本篇博文主要内容为 2026-10-14 从Arxiv.org论文网站获取的最新论文列表，自动更新，按照NLP、CV、ML、AI、IR、MA六个大方向区分。
说明：每日论文数据从Arxiv.org获取，每天早上12:30左右定时自动更新。
提示: 当天未及时更新，有可能是Arxiv当日未有新的论文发布，也有可能是脚本出错。尽可能会在当天修复。
目录
概览 (2026-10-14)
今日共更新2794篇论文,其中:
- 自然语言处理共388篇(Computation and Language (cs.CL))
- 人工智能共852篇(Artificial Intelligence (cs.AI))
- 计算机视觉共482篇(Computer Vision and Pattern Recognition (cs.CV))
- 机器学习共864篇(Machine Learning (cs.LG))
- 多智能体系统共54篇(Multiagent Systems (cs.MA))
- 信息检索共62篇(Information Retrieval (cs.IR))
- 人机交互共55篇(Human-Computer Interaction (cs.HC))
多智能体系统
[MA-0] Optimizing Team Behavior via Extremum-seeking Control in Multi-agent Reinforcement Learning
【速读】：该论文旨在解决合作式多智能体强化学习（Cooperative Multi-Agent Reinforcement Learning, MARL）中行为多样性（behavioral diversity）的动态调控问题。在协作任务中，智能体需通过协调行动实现共同目标，而适当的行为多样性有助于形成角色分工与更丰富的联合策略，但其最优水平高度依赖于具体任务场景，且手动调参成本高昂。为此，论文提出自适应多样性控制（Adaptive Diversity Control, ADiCo），其核心在于构建一个双时标反馈控制系统：在快速时标上，通过多样性控制（Diversity Control, DiCo）将系统多样性实时调节至目标值；在慢速时标上，采用极值搜索控制器（Extremum Seeking Controller, ESC）仅基于平均回合回报信号在线调整该目标值，无需显式模型或多样性-性能曲面的解析梯度。实验表明，ADiCo能在不同任务配置下自动维持适配当前需求的多样性水平——在需要同质策略时保持低多样性，在需异质策略时提升多样性，其平均性能优于经过网格搜索优化的DiCo及无约束多样性演化（Unconstrained DiCo）。结果验证了将经典反馈控制机制引入MARL训练流程，可实现高效、自适应的协调机制，为避免任务特异性的人工多样性参数设定提供了系统性解决方案。
链接: https://arxiv.org/abs/2610.18808
作者: Svar Rajankumar Patel,Kshitij Jerath
机构: University of Massachusetts Lowell (马萨诸塞大学洛厄尔分校)
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI); Robotics (cs.RO); Systems and Control (eess.SY)
备注: Accepted for publication at the 65th IEEE Conference on Decision and Control (CDC 2026), Hawaii, USA
Abstract:In cooperative Multi-Agent Reinforcement Learning (MARL), agents must learn to coordinate their actions to achieve shared objectives. A key indicator of effective coordination is behavioral diversity, i.e., the agents’ potential to enact distinct yet complementary control policies given the same set of observations. Although behavioral diversity can enable role specialization and richer joint behaviors, its appropriate level is highly context- and task-dependent and often costly to tune. We propose Adaptive Diversity Control (ADiCo), a two-timescale framework that adapts behavioral diversity online during training. On a fast timescale, Diversity Control (DiCo) regulates system diversity toward a target value. On a slower timescale, an Extremum Seeking Controller (ESC) adjusts this target using only mean episodic returns, without requiring an explicit model or analytical gradient of the diversity-performance landscape. We evaluate ADiCo on six configurations across two cooperative MARL tasks, where it maintains task-appropriate diversity levels: low diversity when homogeneous policies are beneficial and higher diversity when heterogeneous policies are required. Through this adaptation, ADiCo matches or exceeds the mean performance of DiCo (best grid-searched diversity target) and Unconstrained DiCo (diversity allowed to emerge without target regulation). These results demonstrate that incorporating classical feedback control into MARL training enables effective coordination by providing a systematic alternative to task-specific manual selection of the diversity target.
[MA-1] PlanPath: Exploration-Guided Orchestration for Patient-Level Pathology Reasoning
【速读】：该论文旨在解决病理学诊断中跨多个组织切片（Whole Slide Image, WSI）进行患者层面综合分析的难题。传统病理学智能代理通常仅逐张切片进行形态描述，缺乏对多部位、多切片间互补证据的系统性整合与量化分析能力，导致无法实现真正意义上的患者级结论生成。其核心挑战在于如何在海量异质性切片中精准定位相关证据，并实现从组织架构到单个细胞级别的细粒度测量与跨切片量化整合。为此，本文提出PlanPath——一种面向患者级病理视觉问答（Patient-level Pathology VQA）的智能体框架。该方案的关键在于将问题转化为可执行的探索计划（exploration plan），明确所需证据类型，智能路由至潜在相关切片，并驱动“粗粒度至细粒度”的逐步证据采集过程；同时，利用组织与细胞层级的分析工具获取计数、比例及范围等可量化的特征，由PlanPath在不同切片间聚合形成可追溯的答案。为支持评估，研究构建了两个患者级VQA数据集（肾癌和胃癌），每例平均包含15张WSI，且所有问答对均经病理科医生验证，源自真实诊断报告。实验结果表明，PlanPath在两项数据集上的整体准确率分别优于最强基线9.4%和12.1%，且可显著提升现有滑片级代理的性能。该工作推动了病理学人工智能从局部描述向全局量化推理的范式演进。
链接: https://arxiv.org/abs/2610.18664
作者: Jiusong Ge,Mingzhao He,Qidong Liu,Chunbao Wang,Chunze Yang,Di Zhang,Jiashuai Liu,Mireia Crispin-Ortuzar,Huazhu Fu,Chen Li,Zeyu Gao
机构: Xi’an Jiaotong University (西安交通大学); University of Cambridge (剑桥大学); A*STAR (新加坡科技研究局); Chinese Academy of Medical Sciences (中国医学科学院)
类目: Computer Vision and Pattern Recognition (cs.CV); Multiagent Systems (cs.MA)
备注:
Abstract:A patient’s pathology case spans tens of slides from distinct specimens and anatomical sites, each carrying non-redundant, site-specific evidence. Forming a patient-level conclusion therefore requires locating the slides relevant to a given question and integrating complementary findings across them. However, existing pathology agents analyze one slide at a time and describe morphology rather than measure it, leaving evidence acquisition across a patient’s specimens and patient-level quantification underexplored. To address this gap, we propose PlanPath, an agentic framework for patient-level pathology visual question answering (VQA). PlanPath turns the question into an exploration plan that specifies the evidence required, routes to the slides likely to carry it, and drives coarse-to-fine acquisition from tissue architecture down to individual cells. Tissue- and cell-level tools supply the counts, proportions, and extents, which PlanPath aggregates across slides into a traceable answer. To enable evaluation, we build two patient-level VQA datasets for renal and gastric cancer, averaging 15 WSIs per patient, with pathologist-verified question-answer pairs derived from diagnostic reports. Experimental results show that PlanPath outperforms the strongest baseline by 9.4% and 12.1% in overall accuracy on the two datasets and further improves existing slide-level agents when plugged into our framework. The dataset and code are available at: this https URL.
[MA-2] On Symmetry for Characterizing Agent Heterogeneity in Multi-Agent Systems
【速读】：该论文旨在解决多智能体系统中智能体异质性（agent heterogeneity）的结构性表征问题，即如何在不依赖特定策略或初始状态分布的前提下，对智能体进行基于结构的类型划分。现有方法缺乏统一的结构判别准则，难以有效区分共享策略与个体特异性策略的适用场景。为此，作者在有限时域部分可观测马尔可夫博弈（finite-horizon POSGs）框架下，提出了“策略评估对称性”（policy-evaluation symmetry）的概念：当任意分布式行为策略下的完整策略评估过程与该变换可交换时，该变换即为策略评估对称性。这些对称性构成一个群，其作用下智能体的轨道（agent orbits）定义了结构化的智能体类型。该分类具有内在不变性，不依赖于具体策略选择或初始状态分布。研究进一步证明，所有分量式环境自同构（componentwise environment automorphism）均属于策略评估对称性，但反之不成立。通过将该框架应用于合作型Dec-POMDPs，发现即使智能体共享相同奖励函数，仍可能属于不同结构类型。针对加权奖励聚合情形，作者推导出精确的值函数恒等式，并给出了源POSG的策略评估对称性可被继承至目标Dec-POMDP的充分条件。最后，文章揭示了重复模型因子分解（repeated model factorizations）可诱导策略评估对称性的充分条件，并指出保持最优Boltzmann策略集既非该对称性的必要条件，也非充分条件。因此，该研究的核心解决方案在于建立一套基于策略评估对称性的、不受策略与初始状态影响的智能体结构分类体系，为设计高效且合理的共享/专属策略表示提供理论基础。
链接: https://arxiv.org/abs/2610.18538
作者: Jianhong Wang,Jonathan Lawry
机构: INFORMED-AI Hub, University of Bristol (INFORMED-AI中心，布里斯托大学); School of Engineering Mathematics and Technology, University of Bristol (工程数学与技术学院，布里斯托大学)
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI)
备注: 25 pages
Abstract:Understanding agent heterogeneity can inform the design of shared and agent-specific policy representations, however existing criteria provide no unified structural test. We study finite-horizon POSGs and define a type-compatible transformation as a policy-evaluation symmetry when complete policy evaluation commutes with it for every decentralized behavioral policy. Policy-evaluation symmetries form a group for which agent orbits define structural agent types. This classification does not depend on the particular policy chosen or the distribution of initial states. We prove that every componentwise environment automorphism is a policy-evaluation symmetry, but that the converse does not hold. Applying the framework to cooperative Dec-POMDPs shows that agents can belong to different structural types even when they share a common reward. For prescribed weighted reward aggregation, we derive an exact value identity and sufficient conditions for the resulting Dec-POMDP to inherit the policy-evaluation symmetries of the source POSG. Finally, we give sufficient conditions under which repeated model factorizations induce policy-evaluation symmetry, and show that preserving an optimal Boltzmann policy set neither implies nor follows from this symmetry in general.
[MA-3] CALM: Decentralized Control of Cooperative Aerial Manipulation with Multi-Agent Reinforcement Learning
【速读】：该论文旨在解决多架无人机（UAVs）在风扰、载荷运动及目标广播噪声等复杂环境下，协同空中抓取与操作任务中因共享载荷导致的强耦合动力学问题。其核心挑战在于如何在保持控制架构去中心化的同时，实现高精度的位置与姿态跟踪，尤其是在动态目标和外部干扰下的鲁棒性。解决方案的关键在于提出一种名为CALM（Cooperative Aerial Learned Manipulation）的去中心化控制器，融合了参数共享的多智能体强化学习（MARL）策略（基于多智能体近端策略优化，MAPPO），并引入基于每架无人机的扩展状态观测器（ESO）以估计外部扰动与载荷质量，结合目标运动预测的前馈式编队偏航控制，提前调整编队姿态以减少篮筐偏航滞后，且不增加净控制力。此外，采用条件扰动的集中式训练-去中心化执行（CTDE）框架，使策略在训练时利用全局信息优化价值函数，而执行阶段仍保持完全去中心化。实验表明，CALM在未见轨迹上显著优于多种基线方法（包括机械臂基准、缆绳悬挂实现及两种模型基控制器），在位置与姿态误差方面均表现更优，且在全信息路径退化场景下仍具备统计上相当或更优的性能。
链接: https://arxiv.org/abs/2610.18425
作者: Jiaxi Huo,Petre Ricioppo,Fausto Francesco Lizzio
机构: 未知
类目: Robotics (cs.RO); Multiagent Systems (cs.MA)
备注: 8 pages
Abstract:Cooperative aerial manipulation is difficult under wind, payload motion, and noisy goal broadcasts because multiple Unmanned Aerial Vehicles (UAVs) are coupled through a shared payload. We present CALM (Cooperative Aerial Learned Manipulation), a decentralized controller for aerial manipulators carrying a basket while tracking a moving ground robot. CALM combines a parameter-shared Multi-Agent Reinforcement Learning (MARL) policy, trained using Multi-Agent Proximal Policy Optimization (MAPPO), with goal estimation, per-UAV Extended State Observers (ESOs), basket-mass estimation, and anticipatory formation-yaw feedforward. The anticipatory formation-yaw feedforward uses estimated goal motion to rotate the formation before position error develops, reducing basket-yaw lag without adding net commanded force. A disturbance-conditioned critic supports centralized training while execution remains decentralized. Across held-out trajectories, CALM achieves lower mean position and orientation errors than all evaluated baselines, including the matched arm baseline (7.68 versus 10.95 cm and 3.3 versus 4.7 ^\circ ), the cable-suspended realization, and two model-based controllers. Under unseen full-information-path degradation, CALM remains statistically comparable or superior to the arm baseline. Videos are available at this https URL.
[MA-4] Choosing LLM s for Strategic Agent Interactions FAST NEURIPS
【速读】：该论文旨在解决大语言模型（Large Language Model, LLM）在动态博弈环境中进行战略选择时面临的挑战，即现有评估框架仅针对固定模型在特定游戏类型和交互结构下的表现进行评价，忽视了对手未知、可自适应演化以及反馈信号随时间变化的复杂性。其核心问题是：如何在存在非平稳性和对抗性对手的条件下，实现对LLM战略能力的有效评估与动态选择。解决方案的关键在于提出一种“元博弈”（meta game）范式，将战略选择建模为在基础博弈之上的高层决策过程。在此框架中，一个策略选择器（strategic selector）从模型组合中动态调用不同LLM，以应对在线学习的隐含收益矩阵。研究发现，评估信号具有显著的游戏特异性，需远高于传统基准设置的种子运行次数（N）才能获得可靠信号。实验表明，依赖查询嵌入或模型自我反思的方法无法稳定捕捉深层战略信号，而轻量级多臂老虎机（multi-armed bandit）算法表现出更强的鲁棒性，但仍不及最优固定模型的表现；更重要的是，简单的非平稳及对抗性老虎机变体未能超越静态基线，揭示了当前方法在处理动态对手时存在的关键缺陷。
链接: https://arxiv.org/abs/2610.18354
作者: Arastun Mammadli,Herbert Woisetschläger,Jawad Fayaz,Shiqiang Wang
机构: University of Exeter(埃克塞特大学); IBM Research(IBM 研究院)
类目: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Multiagent Systems (cs.MA)
备注: NeurIPS workshop on Foundations of Agentic Systems Theory (FAST), 2026
Abstract:Game-theoretic benchmarks claim that large language model (LLM) strategic decision-making varies across game types, interaction structures, and capability dimensions. However, they only evaluate strategic performance across individual fixed models. In parallel, contemporary LLM routers pick models under quality-cost trade-offs, but with no notion of a strategic structure. They do not account for an opponent that might be unknown, and that can adapt over time and shift the feedback signal. To address this, we extend the game-theoretic evaluations and define strategic selection as the meta game layered on top of the base game. In each round, a strategic selector fields an LLM from a portfolio of models to navigate a game payoff matrix that is latent and learned online. We find that the evaluation signal is game-specific, and that extracting a reliable signal requires a much higher number of seed runs N compared to traditional benchmarking setups. Evaluating across 5 games and 10 LLMs, we show a model’s strategic competence varies significantly depending on the game setup and the opponent faced. Building on this, we implement a suite of selector methods and find that approaches that rely on query embeddings or LLM self-reflection do not consistently capture the underlying strategic signal. A lightweight multi-armed bandit algorithm performs more reliably across games, though it still falls short of the best fixed model. Finally, we show that simple non-stationary and adversarial bandit variants fail to outperform a stationary bandit baseline, pointing to a gap in how current methods handle a shifting opponent in strategic settings.
[MA-5] Strategic Multi-Agent Resource Allocation via Statistical Contracts
【速读】：该论文旨在解决一类战略性假设检验问题，即当多个拥有与主方目标相关数据或数据生成机制的代理（agent）竞争资源分配时，如何设计一种统计契约（statistical contract），使得主方能够在保证非零假设（non-null）代理参与激励的同时，最小化对零假设（null）代理的资源分配。其核心挑战在于：在代理行为可能具有任意依赖性且存在策略性参与动机的情况下，如何实现对资源配置的严格风险控制和公平性保障。解决方案的关键在于引入e值（e-values）构建“连续闭包”（continuous-closure）——一个能够确保零假设代理在期望上无法获利，并为任意依赖结构下的通用损失函数提供事后风险控制的分配集合。该框架在损失函数为分配给零假设代理的资源比例时，支持多项式时间内的线性优化求解；同时，作者给出了可解释的显式近似形式，并从非零假设代理的视角分析了参与激励条件，明确了其进入契约的盈利边界。此外，研究进一步扩展至在线代理接入场景，提出一种具有时间一致误差控制能力的方法，使主方能在不可逆的实时决策中维持严格统计保障。
链接: https://arxiv.org/abs/2610.18336
作者: Ben Chugg,Michael I. Jordan,Aaditya Ramdas
机构: Carnegie Mellon University (卡内基梅隆大学); UC Berkeley (加州大学伯克利分校); ÉNS (巴黎高等师范学校); Inria (法国国家信息与自动化研究所); Stanford University (斯坦福大学)
类目: Multiagent Systems (cs.MA); Computer Science and Game Theory (cs.GT); Methodology (stat.ME)
备注: 36 pages, four figures
Abstract:We study a class of strategic hypothesis testing problems in which a principal aims to allocate resources among competing agents who possess data or data-generating mechanisms that may be relevant to the principal’s goals. Each agent can pay a participation cost to opt into a statistical contract, in which case they generate data from a private distribution and receive an allocation based on a predetermined (and publicly known) mechanism. Each agent is either “null” or “non-null” and the principal’s goal is to design the contract to limit the allocation to the former while maximizing transactions with the latter. Using e-values, we define a “continuous-closure,” a set of allocations which prevents null agents from profiting in expectation and provides post-hoc risk control for general loss functions under arbitrary dependence. When the loss is the fraction of resources allocated to null agents, the region admits polynomial-time linear optimization. We also give explicit, interpretable closed-form approximations, and we analyze participation from a non-null agent’s perspective, giving conditions under which entry into the contract is profitable. Finally, we extend the framework to support online arrival of the agents, in which case the principal has to make irrevocable on-the-fly decisions, and provide a method which gives time-uniform error-control in this setting.
[MA-6] When Does Opponent Modeling Pay Off? Spatial Inference Stabilises Reinforcement Learning under Partial Observability AAMAS2027
【速读】：该论文旨在解决部分可观测竞争性博弈中深度强化学习智能体在缺乏对手完整信息情况下如何有效推断对手意图并稳定训练的问题。其核心挑战在于，传统基于时序差分（TD(0)）的DQN在部分可观测环境中容易因信用分配困难而陷入训练退化（degenerate seeds），尤其在多步决策配置中表现不稳定。解决方案的关键在于引入一个空间对手模型（spatial opponent model）——基于回溯重标注（hindsight-relabelled）对手目标训练的Transformer架构，能够从对手观测轨迹中生成子目标热力图（subgoal heatmap），并在早期与主网络融合，嵌入到双延迟Q网络（dueling double DQN）中。该模块通过显式建模对手潜在目标，显著提升了训练稳定性：在对称零和觅食博弈中，采用该模块的智能体从未发生训练崩溃，且在成功训练的情况下可恢复接近理论最优的对称博弈价值。此外，该方法具备零样本迁移能力，能有效应对训练阶段未见过的对手，其性能优于基线模型，但在信息论分析预测的边界条件下（如全信息追踪型对手或资源密集地图）会失去优势。消融实验表明，该方法的优势由两机制构成：一是训练期的稳定性增强，主导于信用分配困难且对“盲视”（blinding）具有鲁棒性的高密度地图；二是部署期的推理能力，其作用随环境透明度提升（雾气变薄）而减弱。最终结论为：对手目标推断是一种受可测量边界条件调控的稳定性和迁移性机制，其有效性取决于可见性（雾厚度）、对手意图及资源密度。
链接: https://arxiv.org/abs/2610.18229
作者: Marcel Petran,Stefan Edelkamp
机构: Charles University, Prague, Czech Republic(查理大学，捷克共和国)
类目: Computer Science and Game Theory (cs.GT); Machine Learning (cs.LG); Multiagent Systems (cs.MA)
备注: 9 pages, 7 figures, 2 algorithms, 4 tables, submitted to AAMAS 2027
Abstract:Deep reinforcement learning agents in partially observable competitive games must act on beliefs about an unseen opponent’s intent. We study a spatial opponent model: a transformer trained with hindsight-relabelled opponent goals that outputs a subgoal heatmap from the opponent’s observed trajectory, fused early into a dueling double DQN. In symmetric constant-sum foraging games with partial observability, the module stabilises training. Plain TD(0) DQN suffers degenerate seeds in three of four one-step configurations, whereas the opponent-model agent never collapses and, wherever training succeeds, recovers the majority of the symmetric game value. Transfers zero-shot to a foraging opponent unseen by the learner, with a positive goal advantage over the baseline, and loses this edge exactly where an information argument predicts: against a full-information tracking adversary whose intent is the learner itself, and on resource-dense maps where both agents saturate. A blindfold ablation, zeroing the heatmap at deployment, decomposes the advantage into two mechanisms: training-time stabilisation, which dominates where the credit assignment is hard and survives blinding on the dense map, and deployment-time inference, whose reliance shrinks as fog thins on the sparse map. We conclude that opponent-goal inference is a stability-and-transfer mechanism whose value is governed by measurable boundary conditions: visibility (fog thickness), opponent intent, and resource density.
[MA-7] Fan Out or Relay? Ownership-Aware Routing for LLM Agents Querying Siloed Tables
【速读】：该论文旨在解决组织内部数据所有权分散导致的查询执行效率与准确性难题：当一个查询涉及多个数据拥有者时，传统方法如“广播式路由（fan-out）”或“逐级传递路由（relay）”无法在不预先知晓数据所有权的情况下保证最优性能。其核心问题在于，哪种路由策略更优完全取决于数据所有权分布，而非查询结构本身。解决方案的关键在于提出一种所有权感知路由（Ownership-Aware Routing, OAR）机制：OAR通过仅基于模式（schema-only）生成SQL查询，统计涉及的数据拥有者数量，在读取任何私有数据行之前即完成路由决策。该方法利用公开的数据所有权信息，避免了因错误路由导致的信息泄露或精度损失。实验表明，在10个场景中OAR始终不劣于最优固定协议，且在PeopleJoin上显著优于所有固定协议，最高提升达9.3个百分点，验证了其在真实组织环境中的有效性与鲁棒性。
链接: https://arxiv.org/abs/2610.18157
作者: Zhexi Feng
机构: University of California San Diego(加州大学圣地亚哥分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: 19 pages, 4 figures, 14 tables
Abstract:Inside organizations, a question’s data often sit with several owners, who query their own tables but never hand them over. An LLM coordinator can fan out, sending the question to every owner at once, or relay, querying owners in turn and reusing earlier rows. We show that neither is safe in advance, because data ownership decides which one is right. Across thirteen settings on PeopleJoin and BIRD with three model families, fan-out is more accurate in twelve on questions one owner can answer, yet answers at most 4.3% of those joining two owners’ tables. The flip follows ownership, not query shape: on BIRD two-table joins, fan-out answers 32-44% when one owner holds both and 1-4% otherwise. Relay wins by conditioning on returned rows, and hiding them from an otherwise identical planner cuts two-owner accuracy from 23.2% to 5.1%. Since ownership is public, we propose Ownership-Aware Routing (OAR), which drafts a schema-only SQL query, counts the owners whose tables it joins, and routes the question before any private row is read. In ten settings OAR is never significantly less accurate than the best fixed protocol, and beats every fixed protocol on PeopleJoin, by up to 9.3 points on held-out organizations.
[MA-8] Fresh at Decision Stale at Action: Routing New Information to LLM Decisions Already Made
【速读】：该论文旨在解决大语言模型（Large Language Model, LLM）在做出决策时因信息时效性不足而导致的准确性下降问题。由于决策在生成时即被确定，而执行时环境已发生变化，因此模型所依赖的信息可能已过时，从而影响最终行为的有效性。其核心解决方案是提出“决策有效性的时间戳理论”（timestamp theory of decision validity），并构建一个“决策账本”（decision ledger）以实现该理论。该账本通过六个时间戳精确记录每个决策的生成与更新过程，并引入“新鲜度上限”（freshness ceiling）来界定在执行时刻可保证的信息时效性范围，将后续新增信息带来的不确定性风险显式隔离。模型的准确性由此分解为两大能力：一是“路由”（routing），即确保相关最新信息在决策修订截止前送达；二是“时间判断”（temporal judgment），即准确识别信息的时间效力以决定执行时机。实验验证表明，仅凭新鲜度即可实现理论预测的准确率上限，且模型在关键信息到达后立即响应，准确率随路由效率提升而上升；账本索引机制决定了路由覆盖范围，而决策时设定的“监控条件”（watch conditions）可低暴露地捕捉间接依赖关系，剩余误差主要源于变化生效的时间点判定。因此，决策的正确性不仅取决于初始判断，更依赖于对信息流的精准调度与时间维度的精确理解。
链接: https://arxiv.org/abs/2610.18126
作者: Ayoub Merimi,Souhail Bakkali,Abdeljalil Sakat
机构: 未知
类目: Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT); Information Retrieval (cs.IR); Multiagent Systems (cs.MA)
备注: A timestamp theory of LLM decision validity with a decision ledger
Abstract:Freshness shapes an LLM decision at the moment it is made; the action it commits to is latent, carried out later, in a changing world. We therefore introduce a timestamp theory of decision validity and a decision ledger that realises it. Six timestamps define a decision; a freshness ceiling bounds what up-to-date evidence guarantees at action time, placing the remaining risk in information published afterwards. Accuracy then factorises into two capabilities: routing, which delivers that information to affected decisions before their revision deadlines, and temporal judgment, which reads it for the time of action. The ledger is persistent state holding each decision’s times, evidence and background knowledge, indexed by watch conditions written at decision time. Controlled decision streams with exact labels confirm four predictions across five LLMs, from 8B to frontier scale: freshness alone yields exactly the accuracy the ceiling predicts; models act on decisive news once it arrives, so accuracy follows routing; the ledger’s index sets what routing reaches, and watch conditions reach indirect dependencies at low exposure; and the remaining errors concern when a change takes effect. Freshness makes a decision right when it is made; routing and temporal judgment keep it right when it is carried out.
[MA-9] ECMR: Evidence-Aware Deliberation Control for Multi-Agent LLM Reasoning
【速读】：该论文旨在解决多智能体大语言模型（Multi-agent LLM）系统在推理过程中因采用固定协商协议而导致的计算资源分配不合理问题，特别是由此引发的语义虚假共识（semantic false consensus）——即群体答案虽表现自信，但其支持理由存在狭窄、重复或相关性过高，缺乏真正独立验证的问题。其解决方案的关键在于提出一种基于证据感知的控制器ECMR，该控制器作用于冻结的角色条件化大语言模型（frozen role-conditioned LLMs），在每一步决策中综合评估答案集中度、支持多样性与独立性、矛盾情况、角色行为模式、验证历史及剩余预算，并据此动态选择“提议”“挑战”“验证”“仲裁”“多样化”“修正角色”或“停止”等动作。该决策过程被形式化为有限时域的带预算控制问题，并推导出状态条件化、风险触发验证及角色修正能提升成本调整后效用的充分条件。实验表明，在MedQA、CaseHOLD和FinQA共9,000条公开测试样本上，ECMR-Full将整体准确率从62.1%提升至84.9%，虚假共识率由40.7%降至17.0%，证据支持率从58.0%提高至88.7%；尽管状态随机化导致准确率下降11.7个百分点且仅减少0.7轮次，而更复杂任务则趋向于延迟终止并转向仲裁机制，说明ECMR-Full虽使用最大令牌预算，但实现了最强的绝对可靠性，而非普遍更高的令牌效率。
链接: https://arxiv.org/abs/2610.17869
作者: Yuan Liu,Haoyue Peng,Fanliang Hu,Lei Huang,Wenli Cai
机构: Medical Artificial Intelligence Laboratory, Chongqing University Cancer Hospital; Chongqing University, Chongqing 400030, China
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: 32 pages, including references and technical appendices
Abstract:Multi-agent Large Language Model (LLM) systems typically commit to a deliberation protocol before inference. A fixed protocol cannot allocate computation differently to resolved and unresolved instances, and it may mistake repeated support for independent confirmation. We study this failure as semantic false consensus: a confident group answer supported by narrow, duplicated, or correlated rationales. We introduce ECMR, an evidence-aware controller over frozen role-conditioned LLMs. At each step, ECMR summarizes answer concentration, support diversity and independence, contradictions, role behavior, verification history, and remaining budget, then selects whether to propose, challenge, verify, arbitrate, diversify, correct a role, or stop. We formulate this decision as finite-horizon budgeted control and derive sufficient conditions under which state conditioning, risk-triggered verification, and role correction improve cost-adjusted utility. Across a declared held-out evaluation of 9,000 records from MedQA, CaseHOLD, and FinQA, ECMR-Full increases pooled accuracy from 62.1% to 84.9%, reduces false-consensus rate from 40.7% to 17.0%, and raises evidence-support rate from 58.0% to 88.7%. State shuffling loses 11.7 accuracy points with only 0.7 fewer rounds, while harder traces shift from early stopping toward arbitration. ECMR-Full uses the largest token budget, yielding the strongest absolute reliability rather than uniformly superior token efficiency.
[MA-10] Degrees of Freedom in Multi-Agent Coordination: Between Centralized and Decentralized Regimes
【速读】：该论文旨在解决基于大语言模型（Large Language Model, LLM）的多智能体系统（Multi-Agent System, MAS）中集中式协调与去中心化协调之间的权衡问题。集中式协调虽能提供明确控制，但易形成单一瓶颈；而去中心化协调虽具备局部自主性，却需通过分布式决策与经验共享实现有效协作。为系统性探索二者之间的中间协调空间，本文提出“自由度”（Degrees of Freedom, DoF）这一连续配置框架，用于量化下层智能体被委派的决策空间规模。DoF的关键在于调节任务粒度与任务单元分配至各分支的概率，其两端分别对应极端的集中式与去中心化协调模式。实证结果表明，不同任务域对最优DoF配置存在依赖性：多数场景下，中间配置显著优于两端，且在域内保留经验可缩小两类分支间的性能差异。此外，任务领域特性与保留经验对协调模式的影响强于内部路由机制，揭示出任务依赖性的协调性能景观。研究为设计和优化LLM-MAS中的中间协调机制提供了可量化的实证依据。
链接: https://arxiv.org/abs/2610.17751
作者: Yiqing Peng,Hehai Lin,Zixuan Wang,Dongyi Zhang,Zhengjie Zhu,Yi Han,Yangyang Zheng,Hao Wang,Qian Li,Chengwei Qin
机构: The Hong Kong University of Science and Technology (Guangzhou); Wake Forest University; Shandong University; University of Science and Technology of China
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Large language model (LLM)-based multi-agent systems (MAS) are commonly organized around two coordination regimes. Centralized coordination provides explicit control but can create a bottleneck at a single coordination locus, whereas decentralized coordination offers local autonomy but must achieve effective collaboration through distributed decisions and experience. These trade-offs motivate a systematic study of the space between the two regimes. We introduce Degrees of Freedom (DoF), a continuous configuration framework that quantifies the size of the decision space delegated to downstream agents. DoF controls the granularity of delegated work and the target probability that each task unit is assigned to the branches. Its endpoints recover the two coordination regimes. Across multiple task domains, DoF reveals a hierarchy of organizational effects. Delegated granularity and within-regime routing produce structured but bounded profile reallocations, whereas task domain and retained experience are associated with stronger differences between the centralized- and decentralized-branch profiles. The resulting performance landscape is task-dependent: different domains favor different DoF configurations, while atomic-target conditions are more consistent across runs. In-domain retained experience brings the two branch profiles closer together. The same controlled sweep reveals a task-dependent coordination landscape: some domains exhibit higher-performing configurations, whereas others show flatter responses and no single configuration dominates across domains. In five of the six evaluated state, an intermediate DoF configuration achieves higher mean accuracy than both endpoints, demonstrating the value of exploring the coordination space between them. Together, these findings provide empirical guidance for studying and designing intermediate coordination regimes in LLM-based MAS.
[MA-11] MAS-Gate: Unlocking Multi-Agent System Lifecycle via Node-Centric Quantification
【速读】：该论文旨在解决多智能体系统（Multi-Agent System, MAS）在设计、运行与事后分析三个阶段中因研究碎片化导致的目标不一致与信号难以对齐的问题。现有工作通常局限于特定阶段或单一环节，缺乏跨阶段统一的量化框架，制约了系统整体性能的提升。其解决方案的关键在于提出MAS-Gate——一种基于节点中心的统一量化机制，将整个MAS生命周期建模为有向无环图（DAG），通过结合“节点内效用”（within-node utility，衡量节点对参考答案似然度变化的贡献）与“节点间强度”（inter-node strength，近似语义吸收能力及潜在下游影响），实现跨阶段的结构化表征。该统一量化方法被应用于三大核心任务：为编排器（Orchestrator）训练提供结构感知的密集奖励信号；训练编辑器（Editor）基于已编辑与未编辑轨迹的配对奖励进行运行时前缀干预；以及为大语言模型（LLM）驱动的事后归因提供风险排序节点作为辅助证据。实验结果表明，MAS-Gate在多跳问答、通用推理与错误定位等任务中均显著优于各阶段专用基线，验证了其作为贯穿MAS全生命周期的统一结构接口，在生成优化、动态编辑与错误归因方面的有效性与普适性。
链接: https://arxiv.org/abs/2610.17613
作者: Hehai Lin,Zixuan Wang,Ming Yan,Peidong Wang,Shilei Cao,Minzhi Li,Bosheng Ding,Juepeng Zheng,Chengwei Qin
机构: 未知
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注:
Abstract:Large language model (LLM)-based multi-agent systems (MAS) can solve complex problems beyond the reach of a single agent by decomposing tasks and coordinating complementary capabilities. Despite extensive research on MAS, existing work largely focuses on specific aspects or individual phases of their development and operation, resulting in fragmented perspectives with objectives and signals that are difficult to reconcile across these different settings. To bridge these disconnected perspectives, we conceptualize MAS development and operation as a three-phase lifecycle spanning design-time orchestration before execution, runtime intervention during execution, and post-hoc attribution after execution, and introduce MAS-Gate, a unified node-centric quantification applicable across these phases. Modeling each execution as a directed acyclic graph, MAS-Gate combines within-node utility, which estimates a node’s contribution through changes in reference-answer likelihood, with inter-node strength, which approximates semantic uptake and potential downstream reach. We adapt this shared quantification to three objectives: providing a structure-aware dense reward for Orchestrator training, training an Editor to act on execution prefixes using paired rewards from completed edited and unedited trajectories, and supplying risk-ranked nodes as auxiliary evidence for LLM-based post-hoc attribution. Experiments spanning interaction environments, knowledge-intensive multi-hop question answering, general reasoning, and post-hoc attribution demonstrate that MAS-Gate improves workflow construction, targeted runtime editing, and LLM-assisted error localization over corresponding phase-specific baselines. Together, these results establish MAS-Gate as a unified structural interface for improving MAS generation, dynamic editing, and error attribution throughout the lifecycle.
[MA-12] Mobility-Budget-Constrained Submodular Maximization for Distributed Semantic Multi-Robot Exploration IROS2026
【速读】：该论文旨在解决在通信受限与移动能力受限条件下，多机器人协同探索中如何平衡信息增益与可达性的问题。其核心挑战在于，在分布式环境下，各机器人需在有限的资源预算下，自主决策具有高信息价值且可执行的探索路径，同时避免因过度依赖局部最优而导致全局性能下降。解决方案的关键在于提出一种分布式语义探索框架，融合狄利克雷映射（Dirichlet mapping）、预算约束的轨迹选择机制以及资源感知的分布式贪心（Resource-Aware distributed Greedy, RAG）协调策略。其中，每个机器人通过聚类前沿区域，基于单次预算受限的Dijkstra算法计算地形与航向感知的最小代价路径，并依据边际信息增益（以狄利克雷熵的期望降低量表示）筛选可行路径；该方法保证了每轮个体选择的精确性，团队整体每轮信息增益不低于集中式最优解的一半，且在通信受限时仍能保持良好的渐进性能衰减。实验表明，该方法在网格世界与逼真全系统仿真中均实现了更高的覆盖率、语义mIoU和地图置信度，同时单位距离的遍历成本与熵值更低，且唯一满足每条轨迹的移动预算约束，验证了其在实际应用中的有效性与鲁棒性。
链接: https://arxiv.org/abs/2610.17562
作者: Changxuan Yao,Sandilya Sai Garimella,Lu Gan,Panagiotis Tsiotras
机构: Georgia Institute of Technology(佐治亚理工学院)
类目: Robotics (cs.RO); Multiagent Systems (cs.MA)
备注: 5 pages, 3 figures. Accepted to the IROS 2026 SeMaNa Workshop. The first two authors contributed equally
Abstract:Multi-robot exploration under communication and mobility constraints requires balancing information gain with reachability. We present a distributed semantic exploration framework that combines Dirichlet mapping, budget-constrained trajectory selection, and Resource-Aware distributed Greedy (RAG) coordination. Each robot clusters frontiers, computes terrain- and heading-aware least-cost paths with a single budget-limited Dijkstra search, and selects feasible paths by marginal information gain, represented by the expected reduction in Dirichlet entropy. Each robot’s per-round selection is exact, the team’s joint per-round information gain is at least half the centralized optimum, and the bound degrades gracefully under limited communication. Ten-run gridworld ablations show improved coverage, semantic mIoU, and map certainty with lower entropy and traversal cost per unit distance than shortest-distance frontier selection. Budget-sensitivity experiments show that an intermediate budget best balances path reach and receding-horizon execution. In photorealistic full-system simulation against five coordination rules run in their native published forms, the proposed method attains the highest mean coverage and semantic accuracy while being the only one that respects the per-trajectory mobility budget.
[MA-13] Origin Attribution for Unsafe Behavior in Self-Evolving Multi-Agent Coding Systems
【速读】：该论文旨在解决自演化多智能体编程系统中安全缺陷的溯源难题，即当不安全行为在多个智能体间传播并经过代码迭代后被发现时，如何准确识别其最初引入的智能体及对应程序版本。其核心挑战在于：代码重写会削弱文本相似性、独立发生的安全问题可能干扰时间顺序判断，且中间传递者与源智能体存在相似性，导致混淆。解决方案的关键在于提出一种融合行为指纹匹配（behavioral fingerprint matching）、历史定位（historical localization）与递归追溯（recursive tracing）的方法：通过比较候选实现的多维度行为特征，保留多个历史版本的证据，并沿代码演化路径向上追踪匹配项。在85个受控场景的158个查询中，该方法实现了96.8%的精确溯源准确率，显著优于无递归的全量指纹匹配（77.2%）。组件消融实验和34个额外查询验证表明，多代表样本与递归追溯具有互补作用；进一步扩展至审批边界错误与资源所有者错误的评估，充分证明了行为与历史证据协同在代码修订与跨智能体传播情境下恢复不安全行为起源的有效性。
链接: https://arxiv.org/abs/2610.17488
作者: Hanyi Yu,Jiaming Cui
机构: 未知
类目: Multiagent Systems (cs.MA)
备注:
Abstract:As language-model-based coding agents evolve their programs and reuse code from one another, safety failures can acquire a history that extends beyond the agent in which they are detected. An unsafe change may originate in one agent, propagate through intermediate recipients, and survive subsequent revisions before an audit reveals it. Understanding this history is important for identifying the introducing change and distinguishing inherited behavior from independently occurring failures. We study origin attribution in self-evolving multi-agent coding systems: recovering the agent and program version that introduced an observed unsafe behavior into the target’s code lineage, or determining that it arose locally. This task is challenging because rewriting obscures textual similarity, independent occurrences confound temporal ordering, and intermediate recipients can resemble the source. We propose a method that combines behavioral fingerprint matching with historical localization and recursive tracing. It compares candidate implementations beyond the unsafe outcome, retains evidence from multiple historical versions, and follows matching candidates upstream. On 158 queries across 85 controlled scenarios, the method achieves 96.8% exact attribution accuracy, compared with 77.2% for exhaustive fingerprint matching without recursion. Component ablations and evaluation on 34 additional queries using the same method and hyperparameters demonstrate the complementary roles of multiple representatives and recursive tracing. Further checks extend the evaluation to approval-boundary and resource-owner errors. Together, these results show how behavioral and historical evidence can recover the origins of unsafe behavior through code revision and cross-agent propagation.
[MA-14] Scaling Inference-Time Compute with Teams of Coding Agents
【速读】：该论文旨在解决多智能体系统（multi-agent systems）在推理扩展过程中面临的效率瓶颈问题：尽管引入多个智能体可并行协作以提升性能，但实际应用中常因任务重复、协调失败及计算成本上升而导致资源浪费。其核心挑战在于如何在有限的计算预算下，实现奖励与成本之间的帕累托最优（reward-cost Pareto frontier）。解决方案的关键在于将团队设计建模为一个计算资源分配问题，并通过系统性实验揭示不同阶段的计算投入对整体性能的影响。研究发现，将额外计算资源优先分配给规划阶段的智能体最具价值；使用低成本模型作为规划者能以极低代价获得接近全强模型团队的奖励增益，而廉价执行者或验证者则会降低性能；此外，规划者的边际收益随团队规模增长持续提升，而执行者和验证者的性能在超过四名成员后趋于饱和。这些发现共同构建了一个高效计算资源配置策略，指导用户根据任务特性、成本和延迟约束，精准分配智能体预算以实现最优性价比。
链接: https://arxiv.org/abs/2610.17471
作者: Keenan Samway,Daniel D’souza,Omar Darwiche Domingues,Marzieh Fadaee,Tan Yi-Chern,Nouha Dziri
机构: Cohere Labs(酷睿实验室); Cohere(酷睿)
类目: Multiagent Systems (cs.MA); Computation and Language (cs.CL)
备注:
Abstract:Multi-agent systems promise a new axis for scaling inference: rather than having one agent think longer or sample more attempts, a task is split across many agents that work in parallel and coordinate. However, in practice, more agents can easily produce duplicate work, coordination failures, and higher costs. We therefore formulate team design as a compute allocation problem: starting from a solo agent, where should additional inference compute go to reach the reward-cost Pareto frontier? We build a minimal multi-agent harness from the ground up and trace this frontier by varying, one at a time, the workflow stage at which a lead delegates to its workers (planning, implementation, or verification), the models that form the team, and the number of workers. On ProgramBench, teams outperform solo agents but cost three to seven times as much. With the right compute allocation, however, a heterogeneous 17-agent team comes within 1.4 points of a strong homogeneous 5-agent team at 70% lower cost. This efficiency stems from three findings. First, planning is the most valuable use of workers: in a homogeneous team whose lead decides how many workers to spawn, a lead that delegates only planning to its workers, and then implements and verifies the code itself, matches the reward of a lead that delegates to workers throughout planning, implementation, and verification, at half the cost. Second, cheap models pay off as planners but not as implementers: a strong lead with cheap planners recovers 47.5% of the reward gain of an all-strong team for only 13.9% of its additional cost, whereas cheap implementers lower reward. Third, planners keep improving reward as the team grows, whereas implementers and verifiers plateau beyond four workers. Together, our results offer a guide to spending an agent budget where it counts, given the task, cost, and latency constraints a user has.
[MA-15] Agency as a New Organizing Abstraction for Computer Science
【速读】：该论文旨在解决当前生成式人工智能系统在实际部署中频繁出现的结构性缺陷问题，这些问题表现为资源失控消耗、行为被对抗性内容误导以及运行过程缺乏透明度等。其核心观点是：尽管现有研究多将智能体系统（agentic AI systems）视为基础模型（foundation models）的应用，但这种框架并不完整；真正的突破在于将“代理”（agency）作为一种新的计算机科学组织抽象（organizing abstraction），而不仅仅是人工智能建模的概念。解决方案的关键在于重新构建计算系统的抽象范式——从传统的以程序执行为中心的“过程”（process）抽象，转向以目标追求为核心的“代理”抽象。在此新范式下，系统需在有限资源约束下，基于目标、观测与记忆，自主合成跨工具和环境的行动序列，且无需预先定义特定任务的完整模型。因此，要克服当前系统在执行、接口、正确性、监管、协调与持久性六个维度上的结构性抽象鸿沟，必须推动系统层面与人工智能技术的协同创新，实现对代理行为的可预测、可控与可解释管理。
链接: https://arxiv.org/abs/2610.17410
作者: Ian Foster
机构: Argonne National Laboratory (阿贡国家实验室); The University of Chicago (芝加哥大学)
类目: Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC); Human-Computer Interaction (cs.HC); Multiagent Systems (cs.MA)
备注: An extended version of a paper to appear in Communications of the ACM, November 2026
Abstract:Advances in large-scale machine learning have enabled agentic AI systems: persistent, autonomous computational entities that pursue goals, initiate actions, and interact with open environments. These systems are often framed as applications of foundation models. We argue that this framing is incomplete: agency is also a candidate new organizing abstraction for computer science. What is new is not the concept of an agent, nor runtime policy generation per se, but its general-purpose, open-ended form, in which systems can synthesize courses of action from goals, observations, and memory, over tools and environments for which no complete task-specific model was specified in advance. Thus agency becomes a systems-level abstraction as well as an AI modeling concept. Where the process abstraction organizes computation around executing programs, agency organizes it around pursuing goals under bounded resources: a process executes a program; an agent pursues a goal. From this perspective, the recurring failures of deployed agentic systems (runaway resource consumption, behavior redirected by adversarial content, opaque continuous operation) are symptoms of structural abstraction gaps, not isolated engineering defects. We examine these gaps across six dimensions (execution, interfaces, correctness, oversight, coordination, and persistence) and argue that closing them requires coordinated advances in systems,
[MA-16] DiTSG: Context-Aware Air Traffic Scene Generation with Multi-Agent Diffusion Transformers
【速读】：该论文旨在解决现有空中交通模拟器难以有效捕捉空中交通管制员在实际运行中对多架航空器轨迹进行干预（如速度建议和雷达引导）的规律性问题，进而影响对空域内可安全高效管理的航空器数量及交通构型的准确评估。其核心解决方案是提出一种基于扩散模型的多智能体交通场景生成框架——扩散变换器交通场景生成（DiTSG），其关键在于采用堆叠的多智能体逆向变换器（multi-agent inverted Transformer）作为去噪主干网络，联合生成同一交通场景中多架航空器的轨迹，并以各航空器特定的预计到达时间（Required Time of Arrival, RTA）作为生成条件。该方法能够有效保留真实空管运行中观测到的航空器间交互特性，实验表明所生成的交通场景在统计特性上复现了真实数据特征，同时具备高度多样性与现实合理性。
链接: https://arxiv.org/abs/2610.16959
作者: Seokbin Yoon,Kyungmin Kim,Taejung Kwon,Keumjin Lee,Peng Wei
机构: George Washington University (乔治华盛顿大学); Korea Aerospace University (韩国航空航天大学)
类目: Multiagent Systems (cs.MA); Machine Learning (cs.LG); Robotics (cs.RO)
备注: 8 pages, 6 figures
Abstract:Understanding how many aircraft, and which traffic configurations, can be safely and efficiently managed by air traffic controllers within a given airspace is fundamental to air traffic management. Air traffic controllers continuously modify aircraft trajectories through interventions such as speed advisories and radar vectoring, and the patterns of these interventions are difficult to capture effectively using existing air traffic simulators. To address this limitation, we propose Diffusion Transformers for Traffic Scene Generation (DiTSG), which learns to generate air traffic scenes from recorded multi-aircraft trajectories. DiTSG employs stacked multi-agent inverted Transformer layers as the denoising backbone of a diffusion transformer, jointly generates the trajectories of multiple aircraft within a traffic scene, and conditions the generation on agent-specific required times of arrival. The resulting trajectories generated by DiTSG preserve realistic inter-aircraft interactions observed in actual air traffic operations. Experiments show that the DiTSG generates diverse traffic scenes that reproduce the statistical properties of recorded traffic while preserving realistic inter-aircraft interactions.
[MA-17] Game-Creation-Agents : Autonomous Game Production
【速读】：该论文旨在解决从自然语言指令自动生成可运行的网页游戏（基于Phaser 4框架）这一复杂任务中的系统性挑战，核心问题在于如何将非结构化的文本需求转化为兼具功能完整性、视觉一致性与可执行性的高质量游戏产品。其解决方案的关键在于构建一个分阶段、多智能体协同的生成式流水线：首先通过多模态游戏设计文档（multimodal game design document）将游戏机制与场景契约（scene contracts）与具象化视觉参考相耦合，形成可执行的设计基准；随后将设计文档投影为可机器读取的运行时架构、确定性测试用例与合约规范；再由八个专业化智能体并行执行场景实现、资产生成与构建整合，其中通过一个简化版清单（reduced manifest）实现双分支融合；最后通过构建关联的实时UI验证与代码盲评（code-blind critic）进行质量门控检查，任何失败项将触发归属智能体的有限修复，直至所有质量关卡通过后方可发布。该方法在独立评估中展现出更强的需求满足度、核心机制完备性、内容深度及功能性视觉表现，并在人机对比实验中获得更高评分与偏好。
链接: https://arxiv.org/abs/2610.16776
作者: Tencent Multimodal RL Team:Xiangxin Zhou,Jiaqi Tang,Ke Ma,Wenhao Yang,Jiarui Yao,Boye Niu,Haiyang Guo,Yifei Zhao,Huanjin Yao,Pu Jian,Ruihuang Li,Haitao Wu,Yushi Huang,Xiangyan Liu,Yan Sun,Kaixin Li,Tianyu Pang
机构: 未知
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: this https URL
Abstract:Game-Creation-Agents (GCA) turns a natural-language request into a runnable Phaser 4 browser game. The pipeline first builds a multimodal game design document that couples mechanics and scene contracts to grounded visual references. It then projects that document into a runtime scaffold, machine-readable contracts, and deterministic tests. Eight specialized agents next carry out production: scenes are implemented serially, assets are produced in parallel, and one reduced manifest joins the two branches into a single build. Build-linked real-UI checks and a code-blind critic then verify that build and decide whether the game may ship. Each blocker returns to the agent that owns the affected artifact for a bounded repair, and the game is released once all the gates pass. We evaluate finished games separately, through their rendered interfaces alone, over prompt-matched tasks run under a common protocol. GCA achieves stronger requirement satisfaction, core mechanics, content depth, and functional visuals than a comparable open pipeline. In a side-by-side human study, raters also score GCA’s games higher on every dimension and prefer them overall.
[MA-18] Agent Cue: Statistical Summaries for Guiding Decentralized Multi-Agent Self-Organization
【速读】：该论文旨在解决大规模多智能体系统（Multi-agent Systems, MAS）在去中心化协作中因信息爆炸导致的效率下降问题。随着智能体数量增加，各智能体在共享板上不断记录参与与协作行为，导致信息量呈指数级增长，智能体需耗费大量计算资源读取历史记录，从而削弱了自组织能力。针对这一挑战，论文提出AgentCue方法，其核心在于构建一个基于潜在空间模型（latent space model）的新框架，将智能体与任务类型共同嵌入同一潜在空间中。通过持续更新协作网络与参与网络，并利用潜在空间模型对这些网络进行参数估计，生成紧凑且可解释的统计摘要。智能体仅需参考该统计摘要即可高效选择子任务与合作者，避免重复遍历共享板。研究证明了潜在空间模型的可识别性，并给出了在网络增量下的非渐近误差界，进一步表明每次网络更新后进行固定次数的梯度迭代即可维持稳定的估计精度。在LongBench v2多文档问答任务上的实验验证了AgentCue在有效性、效率和可解释性方面的优越表现。
链接: https://arxiv.org/abs/2610.16621
作者: Guojun Zhu,Xinyou Chen,Kejian Zhang,Sanguo Zhang,Doudou Zhou
机构: University of Chinese Academy of Sciences(中国科学院大学); National University of Singapore(新加坡国立大学); Tsinghua University(清华大学)
类目: Multiagent Systems (cs.MA); Graphics (cs.GR)
备注: 80 pages, 5 figures
Abstract:Multi-agent systems (MAS) combine multiple agents to solve complex tasks, and recent MAS scale to more than a thousand agents. At this scale, decentralized MAS remove the central orchestrator that assigns sub-tasks, and the agents, which we call workers, self-organize through their own choices of sub-tasks and collaborators on a shared board. As workers keep adding participation and collaboration records, however, the information explodes: workers spend more and more effort on reading the records of other workers, and self-organization loses efficiency. In the statistical view, we represent the execution history of an MAS by two continually updated networks, a collaboration network among workers and a participation network between workers and task types. We then propose AgentCue, a method built on a new latent space model that places workers and task types in a shared latent space. Each time the networks are updated, the estimated parameters of the latent space model form a compact and interpretable statistical summary, and workers use the statistical summary to choose sub-tasks and collaborators without repeated searches of the shared board. We establish the identifiability of the latent space model and non-asymptotic error bounds for the estimates under network increments, and we show that a fixed number of gradient updates after each network update suffices for the same error rate. Experiments on LongBench v2 Multi-Doc QA show that AgentCue achieves competitive effectiveness, efficiency, and interpretability.
[MA-19] Learning Graded Sparse Mechanisms: A Representation Theorem and Certified Generalization
【速读】：该论文旨在解决多智能体系统与物理环境中普遍存在的一种分级稀疏机制（graded sparse mechanism）的建模与学习问题，即系统中每个元素的更新依赖于少数关键元素，而对大多数元素的影响较弱，且哪些元素重要及其影响程度随系统状态动态变化。尽管此类机制在群体行为、人群模拟、粒子模拟器及图网络的邻域聚合中广泛存在，但其作为一类具有“分级”特性的函数尚未得到系统的理论研究。本文的核心贡献在于：首先通过一组公理化定义该类函数——包括权重对影响的上限约束、权重依赖的平滑性以及对称性（无特殊元素被强调），并证明了一个表示定理：该类函数恰好构成加权多重集上某种运输度量（transport metric）的Lipschitz球，且该度量由公理唯一确定。这一几何结构为泛化误差提供了双重保障：一方面，泛化界依赖于数据的稀疏性分布特征（sparsity profile）——即有多少元素参与作用及其权重分布，该界以稀疏性分布决定的指数形式呈现，不随总元素数量增长而恶化，并在多个自然情形下给出了匹配的下界；另一方面，基于表示结果，经典的经验偏差界可直接应用于该函数类，其上界为样本两半在该度量下的运输距离（可通过匹配精确计算），并附加一个浓度项。最后，作者在不同交互规则与噪声水平下的模拟维克塞克（Vicsek） flock 模型中验证了该泛化证书的有效性，表明其非平凡（non-vacuous），仅占原始上界的一小部分。因此，解决方案的关键在于：通过公理化构建具有明确几何结构的函数类，并利用运输度量的性质实现对稀疏性依赖的精准刻画与泛化分析。
链接: https://arxiv.org/abs/2610.16537
作者: Mark Kozdoba,Shie Mannor
机构: Technion, IIT(以色列理工学院); NVIDIA(英伟达)
类目: Machine Learning (cs.LG); Multiagent Systems (cs.MA); Machine Learning (stat.ML)
备注:
Abstract:In a variety of multi-agent and physical environments, an element’s update depends strongly on a few other elements and weakly on many, and both the identity and the number of the elements that matter change with the state of the system. Such graded sparse mechanisms are common, for instance, in flocks, crowds, learned particle simulators and the neighborhood aggregation of graph networks, yet they have received little systematic study as a class of functions with the grading given. In this paper we study the supervised learning of the mechanism when the grading is provided by the environment, as a weight on every element. First, we define the class by axioms, where each element’s influence is capped by its weight, the dependence on the weights is regular, and no element is distinguished, and prove a representation theorem: the class is exactly the Lipschitz ball of a transport metric on weighted multisets, and the metric is forced by the axioms. Next, the geometry of the representation gives generalization bounds that depend on the data in two ways. One is via the sparsity profile of the data – the law of how many elements matter and by how much – which bounds the rate over all designs with that profile, with an exponent from the profile and no dependence on the number of elements. In a number of natural cases we also show matching lower bounds, up to constants in the exponent. The other is empirical: due to the representation result, classical empirical discrepancy bounds apply to the class as they stand, and the resulting bound is the transport distance between the two halves of the sample in the class’s own metric, computed exactly by a matching, with an additional concentration term. Finally, we study a simulated Vicsek flock under four interaction rules at two noise levels, and show that the certificate is non-vacuous, a small fraction of the trivial value on every rule.
[MA-20] RelLedger: Relational Information Governance for LLM Agent Organizations ICLR2027
【速读】：该论文旨在解决大语言模型（Large Language Model, LLM）智能体组织中共享记录机制所引发的因果模糊性问题，即在多轮交互历史中，难以追溯特定事实报告决策的驱动因素，也无法判断该记录对下游决策的实际影响。其核心挑战在于：当不同历史路径间存在“交付差距”（delivery gap）时，无法明确是何种关系或线索控制了发送者的报告行为，以及该信息是否真正影响后续任务。为此，论文提出RelLedger这一审计框架，其关键在于通过固定事件、证据、风险与消息内容，交叉对比发送者在不同历史背景（如目标对象与无关角色）下的行为差异，结合角色重分配与历史重写技术以定位控制性线索；并通过消息干预与成本干预识别效应出现的时间节点。此外，归因账本（attribution ledger）将每项声明与其所需响应单元关联，区分缺失干预与采样不确定性；接收方回放机制则追踪已传递信息在下游决策中的演化路径。在模拟单人公司的可执行工作流中，该方法进一步扩展至检索历史、工具调用及共享账本更新的审计。实验表明，不同模型存在特异性的控制路径，且显式损失规则下关系差距更小；例如，Claude在冲突历史中报告同一事实的概率比友好历史高出62.22个百分点，而在可执行工作流中，紧张历史使接收方审查率较温暖历史降低45.83个百分点，揭示了关系动态对决策链的显著影响。
链接: https://arxiv.org/abs/2610.16281
作者: Junxi Huang,Wenshuo Chen,Nuoxuan Li,Kailong Wang,Meitong Shen,Yuehan Kou,Leihao Zhao,Yutao Yue,Xiucheng Wang,Nan Cheng
机构: Xidian University(西安电子科技大学); The Hong Kong University of Science and Technology (Guangzhou)(香港科技大学（广州)
类目: Multiagent Systems (cs.MA); Computation and Language (cs.CL); Computers and Society (cs.CY)
备注: Under review at ICLR 2027
Abstract:In LLM agent organizations, shared records determine which facts are available for later decisions. A sender consults interaction histories before deciding whether to report an event. A delivery gap across histories does not show whose relationship drove the choice or whether the resulting record matters downstream. We introduce RelLedger to audit these links. It holds the event, evidence, risk, and message fixed while crossing the sender’s histories with the event target and an unrelated actor. Role reassignment and history rewrites locate the controlling cue. Message and cost interventions show when the effect appears. An attribution ledger maps each claim to its required response cells and separates missing interventions from sampling uncertainty. Recipient replay follows delivered information into downstream decisions. Executable workflows for a simulated one-person company extend the audit to retrieved histories, tool calls, and shared ledger updates. Across GPT, Gemini, Claude, DeepSeek, and Qwen, the audit reveals model-specific control paths and smaller relation gaps under explicit loss rules. Claude sends the same factual notice 62.22 percentage points more often with conflict histories than with friendly histories. In executable workflows, strained histories reduce recipient review rates by 45.83 points relative to warm histories for Claude.
[MA-21] Pollard: Governed Semantic Execution Trees for Framework-Independent AI Agent Control and Replay ICTAI2026
【速读】：该论文旨在解决代理框架在执行、治理与重放过程中因身份分离而导致的一致性与可验证性问题。现有框架通常将执行、治理和重放逻辑割裂，导致状态不一致或难以追溯。其解决方案的关键在于提出Pollard——一种基于受控语义执行树（governed semantic execution trees）的框架无关运行时。该机制通过严格定义步骤身份（step identity），使其绑定到规范输入、操作类型、尝试次数及父节点身份，确保所有执行结果（完成、拒绝、资源结算、分支历史）均记录于同一不可变树结构中。通过签名确定性冻结（signed deterministic freeze）实现严格重放，无需重新调用被封装操作即可返回经验证的存储结果；同时采用事务性存储协调并发工作负载间的资源预留。实验验证表明，在租约保留、边界约束和封印托管等假设下，系统能有效检测所有语义变更且无误报，多对框架矩阵保持参考行为一致性，且在真实服务路径测试中每个完成的Pollard试验均生成可验证的执行树，充分支持其核心机制的可靠性与可审计性。
链接: https://arxiv.org/abs/2610.16145
作者: Muntaser Syed,Marius C. Silaghi
机构: Florida Institute of Technology(佛罗里达理工学院)
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: Accepted at IEEE ICTAI 2026 (short paper)
Abstract:Agent frameworks can persist checkpoints, trace tool calls, and limit resource use, but these facilities often assign different identities to execution, governance, and replay. We present Pollard, a framework-independent runtime based on governed semantic execution trees. A step identity commits to canonical input, kind, attempt, and parent identity; the same tree records completion, refusal, resource settlement, and branch history. Strict replay returns a verified stored result without invoking the wrapped operation, while a transactional store coordinates reservations across concurrent workers. Under a signed deterministic freeze, two state-model replicas each executed the evaluation transitions without an invariant violation, retained-lease PostgreSQL groups were tested against exact limits, strict replay avoided external calls, and all semantic mutations were detected with no false alarms on the controls. A multi-pair framework matrix preserved reference behavior and verified the trees. A separate cost-capped hosted study exercised live provider paths, and every completed Pollard trial produced a verified tree. Together, the studies support the core mechanisms under explicit lease, boundary, and seal-custody assumptions.
[MA-22] Scaling Multi-Agent Intelligence with Substrate-Centric Coordination
【速读】：该论文旨在解决基于大语言模型（Large Language Model, LLM）的多智能体系统在规模扩展时面临的三大瓶颈问题：随着智能体数量增加导致的阅读负担上升、因未知质量的共享工作而造成有用信息被稀释，以及对失败智能体的脆弱性。现有方法主要采用以智能体为中心的设计模式，其协调机制固化于固定流程、中心化协调器或点对点通信中，尽管近期的黑板式方法将协调机制转移到共享媒介中，仅缓解了最后一个挑战。为克服上述局限，作者提出ScaleMAS系统，其核心创新在于构建一个异步智能体通过三个操作与共享消息数据存储进行交互的架构：发布消息并可选择性引用先前消息以构建依赖关系、基于内容搜索消息、以及通过地址访问消息。其中，搜索操作最多返回k条结果，确保任意群体规模下智能体的读取开销保持恒定；搜索结果按相关性和引用频率排序，后者被证实是信息质量的有效代理指标；同时，消息在生产者离开后仍持久存在，从而实现个体智能体失效后的系统持续运行。实验表明，ScaleMAS在四项代理任务中均以更低的推理计算成本优于以智能体为中心和黑板式基线方法，并可稳定扩展至128个智能体，且具备良好的容错能力。
链接: https://arxiv.org/abs/2610.16120
作者: Zhenting Qi,Zhuokai Zhao,Rulin Shao,Kerem Oktar,Zhiyuan Yao,Himabindu Lakkaraju,Wen-tau Yih,Ansong Ni,Bhargavi Paranjape,Yilun Du
机构: Meta AI(Meta AI); FAIR at Meta(FAIR at Meta); Harvard(哈佛大学); Kempner Institute(肯普纳研究所)
类目: Computation and Language (cs.CL); Multiagent Systems (cs.MA)
备注:
Abstract:Existing large language model (LLM) based multi-agent systems have been shown to scale poorly, and we identify three problems: reading burden that grows with the group, useful information diluted by shared work of unknown quality, and fragility to failed agents. We argue that the problems are largely due to agent-centric designs: the coordination structure is wired into fixed workflows, central coordinators, or peer-to-peer messaging, and recent blackboard-style methods that move the coordination into a shared medium relieve only the last problem. To close the gaps, we build ScaleMAS, where asynchronous agents interact with a shared datastore of linked messages via three operations: publishing messages with optional references to earlier ones that they build on, searching for messages by content, and visiting messages by address. Search returns at most k messages, so an agent’s read stays bounded at any group size; results are ranked by relevance and by how often they are cited, which we find to be an effective proxy for quality; and messages persist after their producers leave, so the group carries on when individual agents fail. Across four agentic tasks, ScaleMAS outperforms agent-centric and blackboard-style baselines at lower inference compute, continues to scale up to 128 agents, and tolerates losing agents midway.
[MA-23] Beyond Perfect Teammates: Multi-Agent Collaboration with Unreliable Colleagues
【速读】：该论文旨在解决大语言模型（Large Language Model, LLM）驱动的多智能体协作系统在现实团队合作场景中面临的核心挑战：现有系统与评估普遍基于“理想同事”假设，即协作智能体始终提供准确、完整且及时的信息，而忽视了实际协作中常见的信息错位、内容不全或响应缺失等行为失效问题。为分离任务求解能力与应对不可靠信息源的鲁棒性，研究提出Unreliable Colleagues Benchmark（UnCoBench），通过Misaligned（信息错位）、Incomplete（信息不全）和Missed（信息缺失）三类设置系统化地模拟协作中的行为失败。其解决方案的关键在于构建一个具备可验证性和可扩展性的自动化生成流水线，结合基于LLM的场景生成与可执行的答案唯一性检验及角色必要性验证，以生成具有真实挑战性的测试用例。实验结果表明，多个模型在可靠设定下表现相近，但在面对不可靠证据时展现出显著差异的鲁棒性；进一步的轨迹分析揭示，智能体普遍存在纠正性跟进的缺失，恢复行为常表现为低效等待与冗余查询。这表明，真正的鲁棒协作不仅依赖于良好的推理能力，更需要主动诊断证据缺陷、针对性恢复相关信息并将其有效整合至决策过程中的行为机制。
链接: https://arxiv.org/abs/2610.16117
作者: Chen Huang,Yiyang Zhang,Yuhao Wu,Yang Deng,Wenxuan Zhang
机构: Singapore University of Technology and Design(新加坡科技设计大学); Singapore Management University(新加坡管理大学)
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: 24 pages, 8 figures, 8 tables
Abstract:LLM-based multi-agent collaboration is increasingly used to solve complex tasks by dividing work among specialized agents and integrating complementary expertise. Yet existing systems and evaluations largely rest on an implicit perfect-colleague assumption: collaborators provide correct and complete information and respond promptly to requests. In realistic teamwork, however, collaborators may return information that does not answer the intended question, provide insufficient detail to support a decision, or fail to respond to a request. To separate task-solving competence from the ability to recover an unreliable evidence-gathering process, we introduce the Unreliable Colleagues Benchmark (UnCoBench), which organizes behavioral failures through Misaligned, Incomplete, and Missed settings. The main agent need to verify contextual correctness, diagnose information gap and track unresolved dependencies in order to obtain the correct evidence for solving the task. We also introduce a verifiable and scalable automated generation pipeline that pairs LLM-based scenario generation with executable checks of answer uniqueness and role necessity. Results across ten models show that similar reliable-setting performance can result in different robustness to unreliable evidence. Trace analysis reveals frequent omissions of corrective follow-up, and the recovery behavior involves inefficient waiting and redundant queries. These findings show that robust collaboration requires not only sound reasoning, but also targeted action to diagnose evidence failures, recover relevant information, and integrate it into decisions. Our data and source code are available at this https URL.
[MA-24] GUIDE: Computationally Augmented Consensus for Trustworthy Clinical Practice Guidelines
【速读】：该论文旨在解决临床实践指南（Clinical Practice Guidelines, CPGs）更新滞后的问题，其核心挑战在于证据爆炸式增长与专家共识形成过程复杂性之间的矛盾。为应对这一问题，研究提出了一种多智能体协作式人工智能助手系统GUIDE，其关键在于通过专业化分工的AI代理承担文献检索、证据评级等重复性高强度任务，同时保留人类专家在关键判断环节（如偏倚风险评估、推荐意见制定）中的最终决策权，实现人机协同的高效指南生成与更新。实验表明，GUIDE在眼科领域12个NICE数据集上实现了88.85%的文献召回率、86.06%的证据确定性评级一致性及91.67%的推荐意见一致率，平均每个数据集仅需162.17分钟且仅需28.68%的人工干预；在心内科、内分泌科、消化科和呼吸科共20个额外数据集上表现稳定，并在去除专家干预后显著降低关键判断步骤性能，验证了专家参与的必要性。此外，在前瞻性真实世界验证中，GUIDE显著减轻专家工作负荷，生成的推荐意见临床可接受性（均值3.73 vs. 1.53）和一致性（66.7% vs. 13.3%）远超十年旧版指南，且与纯人工更新效果相当（P = 0.588），但耗时与投入大幅减少。因此，GUIDE提供了一个兼具时效性、可信度与专家主导性的指南开发范式，验证了计算增强型共识形成的可行性。
链接: https://arxiv.org/abs/2610.16108
作者: Yuanfan Lin,Jie Di,Xujin He,Zhenzhe Lin,Jialing Chen,Ruiyang Li,Changyu Zheng,Yuanrui Shang,Weiqin Peng,Jingxi Qin,Xiong Wang,Dongxin Ye,Yining Huang,Guoqiang Li,Andi Xu,Sijia Yu,Ruixin Wang,Yahan Yang,Xiaohang Wu,Duoru Lin,Jingjing Chen,Zhenzhen Liu,Wenben Chen,Haotian Lin
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC); Multiagent Systems (cs.MA)
备注: 108 pages total, 28 pages main text, 5 figures, 4 supplementary figures, 8 supplementary tables. Research article
Abstract:Maintaining up-to-date clinical practice guidelines (CPGs) is essential for evidence-based medicine, but rapid evidence growth and complex expert consensus formation limit timely updates. We developed GUIDE, a multi-agent copilot system for human-AI collaborative guideline generation and updating. GUIDE delegates labor-intensive tasks to specialized AI agents while preserving expert judgment and final arbitration. In 12 NICE ophthalmology datasets, GUIDE achieved 88.85% literature recall, 86.06% alignment in certainty-of-evidence ratings, and 91.67% recommendation agreement, requiring 162.17 minutes per dataset on average (28.68% human intervention). Performance remained stable across 20 additional datasets from cardiology, endocrinology, gastroenterology, and respiratory medicine. Ablation analysis showed that removing expert intervention reduced performance in judgment-intensive steps such as Risk-of-Bias assessment. In six post-cutoff datasets, GUIDE maintained robust end-to-end performance on unseen evidence. Prospective real-world validation showed that GUIDE reduced expert workload and produced more clinically acceptable recommendations than the original decade-old guideline (mean score, 3.73 vs. 1.53; agreement, 66.7% vs. 13.3%), with outcomes comparable to human-only updating (P = 0.588) but substantially less time and effort. GUIDE offers a practical framework for timely, trustworthy, expert-centric guideline development and validates computationally augmented consensus.
[MA-25] Dream2Discern: Learning from Hard Latent Variants for Runtime Failure Prediction in Multi-Agent Systems
【速读】：该论文旨在解决基于大语言模型（LLM）的多智能体系统在实际部署中缺乏可靠运行时审计器的问题，即如何从已观测到的执行前缀（execution prefix）准确预测最终执行结果。现有方法依赖有限且经过验证的轨迹数据集进行训练，但扩展该数据集需耗费大量执行与验证成本，且仅扩大失败覆盖范围无法提供明确标准以识别当前审计器难以分类的前缀。为此，本文提出Dream2Discern，其核心创新在于通过自监督的隐式增强机制，利用审计器自身预测损失驱动生成更具挑战性的难例。具体而言，在Dream阶段，通过对审计器的表示算子施加扰动，使已有前缀的预测损失最大化，从而生成“困难的潜在变体”；在Discern阶段，审计器基于原始标签学习从这些变体中预测执行结果。随着模型持续学习，变体被动态重生成，将有限的数据集转化为可再生的难例来源，无需额外轨迹采集或结果验证。实验表明，Dream2Discern在AFTraj-2K和StepShield两个基准上均达到超过90%的轨迹级加权准确率，显著优于现有方法，在预测准确性与实时性之间取得良好平衡。
链接: https://arxiv.org/abs/2610.16071
作者: Boxuan Zhang,jianing Zhu,Xiaotian Han,Ruixiang Tang
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multiagent Systems (cs.MA)
备注: 27 pages, 6 figures
Abstract:Reliable deployment of LLM-based multi-agent systems calls for runtime auditors that predict terminal outcomes from observed execution prefixes. These auditors typically learn from a limited corpus of trajectories with verified outcomes. Expanding this corpus requires costly execution and verification, yet broader failure coverage alone provides no explicit criterion for identifying prefixes that the current auditor finds difficult to classify. To this end, we propose Dream2Discern, a runtime auditor trained through implicit augmentation guided by its own prediction loss. During Dream, we select perturbations of the auditor’s representation operators to increase prediction loss on recorded prefixes, yielding hard latent variants. During Discern, the auditor learns to predict execution outcomes from these variants using the original labels. The variants are regenerated as the auditor learns, turning a limited corpus into a renewable source of hard examples and requiring no additional trajectory collection or outcome verification. Extensive evaluations on AFTraj-2K and StepShield show that Dream2Discern outperforms competitive failure auditors and achieves over 90% trajectory-level weighted accuracy on both benchmarks while balancing prediction accuracy and timeliness. Project page: this https URL
[MA-26] PopGen: Generating Semantically Diverse Populations for Zero-Shot Coordination ICLR2027
【速读】：该论文旨在解决多智能体强化学习（Multi-Agent Reinforcement Learning, MARL）中合作智能体在训练阶段因缺乏真正语义多样性策略而难以实现鲁棒零样本协作的问题。现有方法通常依赖轨迹层面的代理指标来衡量多样性，但此类指标易将同策略的不同轨迹误判为多样，导致训练环境中的策略分布失真。针对这一问题，论文提出关键解决方案——PopGen框架，其核心在于利用大语言模型（Large Language Models, LLMs）对智能体行为进行语义层面的描述与生成。具体而言，PopGen首先通过LLMs生成一组具有语义差异的智能体行为文本描述，再将每种行为转化为可执行的代码策略（code policy），从而构建出语义上真正多样化的智能体群体。实验表明，基于PopGen生成的多样化群体训练出的合作智能体，在未见过的伙伴面前展现出更强的零样本协作能力，显著优于传统基于轨迹多样性的人工群体训练方法（population-play baseline）。此外，研究还发现语义多样性随PopGen群体规模增加而提升，且通过合理的设计选择可在保持群体质量的同时最大化多样性，验证了该框架的有效性与可扩展性。
链接: https://arxiv.org/abs/2610.15969
作者: Paul Teiletche,Kanghyun Ryu,Kyungjoon Park,Dhruva Tirumala,Markus Wulfmeier,Negar Mehr
机构: EPFL(洛桑联邦理工学院); UC Berkeley(加州大学伯克利分校); Yonsei University(延世大学); Google DeepMind(谷歌深度思维); Nomagic
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: Submitted to ICLR 2027
Abstract:In multi-agent reinforcement learning, exposing a cooperator to partners with diverse strategies during training is key to better collaboration at test time. However, existing methods typically rely on trajectory-level proxies to measure diversity, which may label two distinct trajectories of the same strategy as diverse. Instead, our key insight is to leverage LLMs’ ability to describe behaviorally distinct agents in environments, similar to how humans can recognize high-level strategies. To this end, we introduce PopGen, a framework that synthesizes semantically diverse agents from scratch using LLMs. PopGen first generates a diverse set of textual descriptions of agent behaviors, then translates each behavior into a code policy that can be executed within the environment. We show that cooperators trained against PopGen populations achieve robust zero-shot cooperation with unseen partners, outperforming population-play baselines across four collaborative environments. We further show that semantic diversity scales with PopGen population size, improving the cooperator’s worst-case performance, and we identify design choices that preserve population quality while maximizing diversity. Our code is available at this https URL.
[MA-27] DiffNash: Finding Nash Equilibrium in High-Dimensional Continuous Action Spaces with Diffusion Policies
【速读】：该论文旨在解决高维连续博弈中均衡计算的问题，核心挑战在于需要学习和表示复杂、多模态的策略分布，而传统神经策略模型在这一任务上表现受限。其解决方案的关键是提出一种基于扩散模型（Diffusion Model）的框架，通过构建重加权得分匹配（score-matching）目标，实现无需直接采样自每次更新后的策略分布即可训练扩散策略。该方法的核心创新在于设计了一种无偏估计器，用于估算当前策略与历史策略在证据下界（Evidence Lower Bound, ELBO）之间的差异，其指数形式可作为重要性权重的可计算近似，从而有效缓解因复用历史动作导致的分布偏移问题。所提出的DiffNash算法融合了在线镜像下降（OMD）的快速收敛特性与虚构对手法（FP）的样本效率优势，通过引入回放缓冲区复用历史动作，并利用上述重要性权重修正机制，在森林保护博弈和连续版“将军博弈”中分别将纳什差距（Nash gap）相较于最强基线降低了35.6%和24.7%，且仅需极少超参数调优。
链接: https://arxiv.org/abs/2610.15932
作者: Haichuan Wang,Anagha Satish,Jiale Zheng,Saisneha Ghatti,Lingkai Kong,Kai Wang,Milind Tambe
机构: Harvard University (哈佛大学); Massachusetts Institute of Technology (麻省理工学院); Georgia Institute of Technology (佐治亚理工学院)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT); Multiagent Systems (cs.MA)
备注: 41 pages, 11 figures
Abstract:Computing equilibria in high-dimensional continuous games requires learning and representing rich, multi-modal strategy distributions, something conventional neural policies struggle to achieve. Diffusion models offer a natural solution, but standard diffusion training requires direct samples from a fixed target distribution, which are generally unavailable under iterative game-solving updates. We propose a diffusion-based framework for equilibrium computation that enables diffusion-based variants of Online Mirror Descent (OMD) and Fictitious Play (FP). Our approach derives a reweighted score-matching objective that trains diffusion policies without requiring direct samples from each updated policy. Our main method, DiffNash, combines the fast convergence of OMD with the sample efficiency of FP by reusing past actions through a replay buffer. To correct the resulting distribution shift from sample reuse, we derive an unbiased estimator of the difference between current and past policies’ Evidence Lower Bounds, whose exponential provides a tractable proxy for the corresponding importance weight without evaluating diffusion likelihoods. We evaluate our approach on forest protection games and continuous Colonel Blotto games, reducing the Nash gap over the strongest baseline by 35.6% and 24.7%, respectively, with minimal hyperparameter tuning.
[MA-28] Scalable Weak-to-Strong Verifier Hardening for Agent Environments
【速读】：该论文旨在解决大语言模型（LLM）在代理环境（agent environment）中因验证器（verifier）可被攻击而引发的奖励劫持（reward hacking）问题。具体而言，现有验证器可能被恶意代理利用，使其在未完成目标任务的情况下仍获得全额奖励，导致评估结果失真及训练过程出现偏差。随着代理环境数量与复杂度的提升，验证器的安全性亟需增强。为此，论文提出一种可扩展、全自动的“弱到强”验证器加固方法，其核心在于构建一个隔离的验证器沙箱以防止代理直接篡改验证逻辑，并引入“黑客-修复者循环”（hacker-fixer loop）机制，通过迭代式漏洞探测与修复实现防御能力的持续增强。该方法在Terminal Bench、KernelBench和SETA等多个基准环境中显著降低了攻击成功率，且随着迭代次数增加，防御强度呈递增趋势，优于静态非迭代基线方案。该方法的早期版本已被官方Terminal Bench仓库采纳，用于审计环境中的漏洞，体现了其实际应用价值。
链接: https://arxiv.org/abs/2610.15924
作者: Chong Zeng,Ziqian Zhong,Kexun Zhang,Ivan Bercovich,Ivgeni Segal,Yuzhen Mao,Shashwat Saxena,Aditi Raghunathan
机构: Stanford University (斯坦福大学); Carnegie Mellon University (卡内基梅隆大学); Fewshot Corp; Independent Researcher
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multiagent Systems (cs.MA)
备注: Project page: this https URL
Abstract:Agent environments are essential for large language model (LLM) development. They use programmatic verifiers to turn complex outcomes into rewards. Verifiers can be attacked into awarding full credit to cheating agents without completing the intended task, causing inaccurate evaluation and misalignment in training. As agent environments scale in numbers and difficulty, so must the repairing of exploitable verifiers. To this end, we ask whether today’s model with scaled compute can automatically harden agent environments to prevent potentially stronger future models from reward hacking. We introduce a scalable, fully automatic approach to weak-to-strong verifier hardening. It first sets up an isolated verifier sandbox that agents cannot directly modify, then uses a hacker-fixer loop to find and repair remaining verifier vulnerabilities. We apply this loop to Terminal Bench, KernelBench, and SETA, substantially reducing attack success across these agent environments. The iterative design scales with the hardening budget, producing progressively stronger defenses as the number of loop iterations increases and outperforming the static, non-iterative baseline. An earlier version of our hacker-fixer loop has been adopted by the official Terminal Bench repository as a way to audit exploits in its environments. Project page: this https URL. Comments: Project page: this https URL Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multiagent Systems (cs.MA) Cite as: arXiv:2610.15924 [cs.CR] (or arXiv:2610.15924v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2610.15924 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[MA-29] Mind Your Own Role: Reducing Role Overstepping for Effective and Efficient Multi-Agent Collaboration via Dynamic Role Activation Steering
【速读】：该论文旨在解决多智能体系统（Multi-Agent Systems, MAS）在执行复杂任务时，智能体因职责越界（role overstepping）而执行其他角色的子任务所导致的工作流紊乱、系统效率下降及任务性能退化的问题。其核心解决方案是提出一种轻量级且高效的推理阶段激活干预方法——动态角色激活引导（Dynamic Role Activation Steering, DRAS）。DRAS的关键在于通过分析模型内部激活空间中的角色相关模式，发现角色指令会引发激活空间中一致的方向性偏移，且遵循角色与越界行为的轨迹在激活模式上具有显著差异。基于此，DRAS为每个角色构建一个角色引导向量和一个轻量级越界检测器；在大语言模型（LLM）生成过程中，检测器根据当前激活状态实时估计越界概率，并动态调整引导强度，从而将智能体的生成轨迹精准引导回其本职职责。实验表明，相较于原始MAS，DRAS可将角色越界轨迹减少高达86.6%，显著提升任务质量评分10.0%–11.1%，实现1.50×–1.61×的任务完成速度提升，并降低26.8%–38.7%的令牌消耗。
链接: https://arxiv.org/abs/2610.15907
作者: Xingke Yang,Sicong Li,Lingfeng Yao,Rui Chen,Gen Li,Hao Wang,Xin Fu,Miao Pan
机构: University of Houston(休斯顿大学); Stevens Institute of Technology(史蒂文斯理工学院)
类目: Multiagent Systems (cs.MA)
备注:
Abstract:Multi-Agent Systems (MAS) have shown strong capability in solving complex tasks by assigning different agents specialized roles and corresponding subtasks. However, agents may overstep their assigned responsibilities and perform subtasks belonging to other roles, which can disrupt workflow, reduce system efficiency, and degrade task performance. To address this issue, we investigate role-related patterns in the model’s internal activation space and uncover two key observations: (i) role instructions induce consistent directional shifts in activation space, and (ii) role-adherent and role-overstepping trajectories exhibit distinct activation patterns. Based on these observations, we propose Dynamic Role Activation Steering (DRAS), a lightweight and effective inference-time activation intervention approach to reduce role overstepping in multi-agent collaboration. Specifically, DRAS constructs a role steering vector and a lightweight role-overstepping detector for each role. During LLM generation, the detector estimates the probability of role overstepping from the current activation state, and DRAS dynamically adjusts the steering strength accordingly to guide the agent toward its assigned responsibilities. Experimental results show that DRAS effectively reduces the number of role-overstepping trajectories by up to 86.6% compared with Vanilla MAS, thereby improving task quality score by 10.0%–11.1%, achieving a 1.50 \times --1.61 \times task-completion speedup, and reducing token consumption by 26.8%–38.7%.
[MA-30] Agent Sociodrama: Social-Dynamics-Guided Persona Recasting for LLM Agent Teams
【速读】：该论文旨在解决大型语言模型（LLM）代理团队在协作过程中表现出的典型社会性失效问题，如信息持有者沉默、主导发言者固执于错误答案以及规范违规行为未被纠正等。其核心解决方案是提出“代理社会剧”（Agent Sociodrama）框架，通过统一的外部LLM裁判对代理的情绪状态及具有方向性和非对称性的社会关系进行每轮更新，从而动态建模和验证代理团队中的社会演化过程。在此基础上，进一步引入轨迹引导型人格重塑（Trajectory-Guided Persona Recasting, TPR），利用交互日志与社会轨迹诊断群体动态失败，并重构代理人格以修复协作缺陷。实验表明，TPR在四个协作基准任务中显著提升了性能，相较于初始人格配置最高实现83.3个百分点的准确率提升；在分布式信息合作与混合动机谈判等复杂场景中，其帕累托效率优于SOTOPIA和AgentVerse，最高提升达37.5个百分点。结果证明，经过验证的社会演化轨迹可作为可操作的信号，用于主动修复和优化LLM代理团队的协作效能。
链接: https://arxiv.org/abs/2610.15788
作者: Yirui Qi,Xiaoming Zhang,Jaehong Yoon
机构: Beihang University (北京航空航天大学); Nanyang Technological University, Singapore (南洋理工大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multiagent Systems (cs.MA)
备注: Project Page: this https URL
Abstract:Large language model (LLM) agents are increasingly deployed as teams, yet they fail in characteristically social ways: information holders stay silent, dominant speakers anchor groups on wrong answers, and norm violations go unsanctioned. We present Agent Sociodrama, a framework for modeling and validating evolving social dynamics in LLM agent teams. Under fixed personas, agents’ emotional states and directed, asymmetric relationships are updated every round by a unified external LLM judge and shape their subsequent interactions. Across six LLM backbones and five social scenarios, we assess social fidelity through classical social-theory patterns and agreement with human judgments. The resulting trajectories exhibit structured social patterns and model-specific differences. We then introduce Trajectory-Guided Persona Recasting (TPR), which diagnoses group-dynamic failures from interaction transcripts and social trajectories and recasts agent personas before rerunning the task. Across four collaborative benchmarks, TPR improves performance from both heterogeneous and randomly initialized agent persona compositions, yielding up to 83.3%p of accuracy improvement over the initial casting. On representative distributed-information cooperation and mixed-motive bargaining tasks, we outperform SOTOPIA and AgentVerse in three of four comparisons, by up to 37.5%p in Pareto rate. These results show that validated social trajectories provide actionable signals for repairing LLM agent teams.
[MA-31] Dynamic Conflict Clustering with Lightweight Learning and Strategic Repair for Distributed Constraint Satisfaction
【速读】：该论文旨在解决分布式约束满足问题（DCSP）中因传统求解器依赖广泛且重复的点对点通信、缺乏战略协调与学习能力而导致的通信开销大、易陷入停滞状态以及难以求解复杂实例的问题。其核心解决方案是提出一种基于修复机制的动态冲突聚类-不完全求解器（DCC-Incomplete），关键在于通过冲突中心化的分解策略，将未求解部分划分为冲突簇，并利用协调层限制通信范围至同一冲突区域内的代理；同时引入记忆引导的重启机制，使代理与冲突簇能够根据邻近冲突状态暂停、修正并自适应调整修复策略。实验结果表明，该方法显著提升了求解成功率（从37.50%提升至61.11%），降低了中位运行时间（从0.565s降至0.367s）和中位通信量（每约束消息数从12.493降至6.503），验证了动态冲突聚类结合轻量化战略与学习引导修复在提升通信效率与求解鲁棒性方面的有效性。
链接: https://arxiv.org/abs/2610.15658
作者: Raziyeh Moghaddas,Fabio Caraffini
机构: 未知
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI); Discrete Mathematics (cs.DM)
备注:
Abstract:Distributed Constraint Satisfaction Problems (DCSPs) model distributed decision-making in multi-agent systems, where agents assign values to local variables while satisfying shared constraints. However, many DCSP solvers still rely on broad, repeated peer-to-peer communication and are largely reactive, with limited support for strategic coordination or learning. This increases communication overhead, can lead to stagnation, and reduces the success of hard instances. Hence, we propose Dynamic Conflict Clustering-Incomplete (DCC-Incomplete), a repair-based solver that combines conflict-centred decomposition, structured coordination, strategic repair, and lightweight learning for stagnation recovery. DCC-Incomplete groups the unsolved part of the problem into conflict clusters and restricts communication via a coordination layer to agents within the same conflict region. It also introduces memory-guided restart, enabling agents and clusters to pause, revise, and adapt repairs based on neighbouring conflict states. A thorough evaluation of DCC-Incomplete on distributed graph-colouring benchmarks with planted satisfying assignments shows that its memory-guided restart improves the success rate from 37.50% to 61.11%, reduces median runtime from 0.565s to 0.367s, and cuts median communication from 12.493 to 6.503 messages per constraint. These results suggest that dynamic conflict clustering, combined with lightweight strategic and learning-guided repair, improves both communication efficiency and robustness in incomplete DCSP solving.
[MA-32] Representation Theorems for Actual Powers over Two-Agent General Concurrent Game Frames
【速读】：该论文旨在解决传统并发博弈框架（concurrent game frames）中权力表征的局限性问题，特别是基于α效用性（alpha effectivity）定义的联盟权力所隐含的单调性缺陷：该定义无法区分联盟通过不同联合行动实现的相同结果集，从而掩盖了实际行动策略的差异。为此，论文采用由van Benthem、Bezhanishvili和Enqvist（2019）提出的“实际权力”（basic power）概念，即以联盟的每种联合行动及其对其他参与者可能行动的全集所对应的可达状态集合来刻画其权力结构。进一步地，针对Li与Ju（2025, 2026）指出的标准并发博弈框架中三个可能过强的假设——序列性（seriality）、代理间的独立性（independence of agents）以及确定性（determinism），论文在仅考虑双代理人情形下，系统分析了由此产生的八类广义并发博弈框架，并证明这些框架在实际权力意义下可被八类对应的邻域框架（neighborhood frames）精确表示。这一结果扩展了van Benthem等人（2019）的代表性定理，为博弈论与逻辑之间的连接提供了更精细、更具包容性的形式化基础。
链接: https://arxiv.org/abs/2610.15476
作者: Zixuan Chen(Institute for Logic, Language and Computation, University of Amsterdam, Amsterdam, The Netherlands),Fengkui Ju(School of Philosophy, Beijing Normal University, Beijing, China),Thomas Ågotnes(Department of Information Science and Media Studies, University of Bergen, Bergen, Norway and School of Philosophy, Shanxi University, Taiyuan, China)
机构: 未知
类目: Logic in Computer Science (cs.LO); Computer Science and Game Theory (cs.GT); Multiagent Systems (cs.MA)
备注: In Proceedings GandALF 2026, arXiv:2610.08898 . Corresponding author: Fengkui Ju
Abstract:One of the most well-known connections between logic and games is Pauly’s representation theorem: the induced powers of individuals and coalitions in concurrent game frames correspond, in a precise sense, to a certain class of neighborhood frames. The precise sense here is what is called alpha effectivity (or alpha power): the power of a coalition is characterized by the sets of states in which it can ensure that the outcome lies by taking some joint action. This definition is inherently monotonic, and, as pointed out by van Benthem, Bezhanishvili, and Enqvist (2019), that fact can obscure relevant information about the power structure in the game: we do not know whether two sets that a coalition has the power to enforce correspond to the same joint action or to different joint actions. An alternative is to characterize the power of a coalition by its actual power (called basic power by van Benthem, Bezhanishvili, and Enqvist (2019)): the collection of sets of states, each of which corresponds to one joint action by the coalition and all possible joint actions by the other agents. It has recently been argued by Li and Ju (2025, 2026) that standard concurrent game frames rely on three assumptions that may, in some cases, be too strong: seriality, independence of agents, and determinism. This yields a total of eight different classes of general concurrent game frames. In this paper, assuming two agents, we prove that, in terms of actual powers, the eight classes of general concurrent game frames are represented by eight corresponding classes of neighborhood frames. This generalizes a result by van Benthem, Bezhanishvili, and Enqvist (2019).
[MA-33] CoLA: Multi-VLA Coordination via Learned Adapters
【速读】：该论文旨在解决多智能体协作中何时需要显式通信的问题，特别是在共享工作空间的视觉-语言-动作（Vision-Language-Action, VLA）模型场景下。现有方法通常依赖于隐式观察或显式消息传递来实现协调，但缺乏对通信必要性的系统性分析。其解决方案的关键在于提出一种名为CoLA（Coordination via Learned Adapters）的新架构：在保持各智能体原有VLA主干网络冻结的前提下，仅通过训练轻量级的消息通道与每个机械臂对应的扩散动作头，实现完全去中心化的执行。实验表明，当信息具有私密性（如仅一个智能体可见遮挡托盘上的颜色标记）时，显式通信显著提升性能（成功率达76.0%），而当信息可通过共享视角（如共用俯视摄像头）获取时，移除消息对性能影响可忽略；此外，该架构具备良好的泛化能力，适用于不同VLA主干网络，并在参数效率和任务表现上优于基于LoRA微调的基线方法。
链接: https://arxiv.org/abs/2610.15337
作者: Ahaskar Kashyap,Jiawen Zong,Boyu Ma,Chuhao Zhou,Jingliang Li,Stefano V. Albrecht,Jianfei Yang
机构: MARS Lab; Nanyang Technological University, Singapore; Autonomous Agents Research Group
类目: Robotics (cs.RO); Machine Learning (cs.LG); Multiagent Systems (cs.MA)
备注: 22 pages, 5 figures. Equal contribution: Ahaskar Kashyap, Jiawen Zong, Boyu Ma. Code: this https URL
Abstract:Vision-Language-Action (VLA) models are typically deployed as isolated single-agent policies. When several such agents share a workspace, they can coordinate implicitly, by observing one another, or explicitly, by exchanging messages, but it remains unclear when explicit communication becomes necessary. We introduce CoLA (Coordination via Learned Adapters), which keeps each agent’s VLA backbone frozen and trains a lightweight message channel and a per-arm diffusion action head on top, with fully decentralised execution. Across three tasks that differ in how much each agent can observe, the channel matters only when information is private. On an occluded tray handover, where the box’s colour marker is visible to one agent only, CoLA reaches 76.0% success versus 1.3% with messages zeroed throughout training. On two-arm and three-arm handover tasks, removing messages has no measurable effect with a shared overhead camera (90.6% vs. 88.3% at two arms; 63.5% vs. 60.0% at three), but lowers success on the two-arm task’s wrist-only variant (74.8% vs. 64.3%). This architecture generalises across different VLA backbones as well. With a frozen \pi_0.5 backbone on the same occluded tray task, CoLA matches the overall success rate of a LoRA-fine-tuned baseline (37.3% vs. 34.7%) using a fraction of the trainable parameters. When CoLA places the box, it selects the correct tray 71.8% of the time, whereas the independently fine-tuned baseline, whose placing arm never sees the marker, selects it at chance level (34.7% against 33.3%).
[MA-34] Do Rollups Self-Organize? Multi-Agent Reinforcement Learning in the EIP-4844 Blob Fee Market
【速读】：该论文旨在解决以太坊分片（EIP-4844）中数据区块（blob）费用机制在多参与方、动态博弈环境下的稳定性与效率问题，尤其关注多个独立、自利的rollup如何在缺乏协调的情况下共同实现目标数据量（每区块3个blob）的供给。传统分析仅能处理单一代理在外部价格假设下的最优行为或静态均衡情形，无法刻画真实市场中异质rollup之间的复杂互动。为此，作者提出一个校准后的多智能体强化学习（multi-agent reinforcement learning, MARL）仿真环境，其中基础费用由EIP-4844规则根据异质rollup的聚合行为内生决定。研究发现，在无收敛保证的独立学习者设定下，系统仍能稳定趋近目标供给水平（3.32±0.01 blob/区块），并在需求范围扩大50倍（10×至500×）时，供给维持在目标附近（10×至200×需求区间），仅在极高需求下超出目标，此时费用调控失效。此外，滚子规模越大，其数据提交频率越高，呈现出与前期理论预测一致的“规模阈值”现象。值得注意的是，引入零成本的Layer-1 calldata作为外部选项后，费用下降约一个数量级，但总供给仍保持在目标水平，表明该费用规则本质上起到了对去中心化供给的负反馈控制作用。
链接: https://arxiv.org/abs/2610.15256
作者: Alessio Brini
机构: Duke University (杜克大学)
类目: Multiagent Systems (cs.MA); Computer Science and Game Theory (cs.GT); Trading and Market Microstructure (q-fin.TR)
备注: 5 pages, 4 figures, 3 tables. Accepted for publication in the Proceedings of the 2026 IEEE 4th International Conference on Artificial Intelligence, Blockchain, and Internet of Things (AIBThings 2026)
Abstract:EIP-4844 gave Ethereum rollups a separate fee dimension for data, governed by a multiplicative update rule that targets three ``blobs’’ per block. Whether this target is reached depends on how many independent, self-interested rollups choose to post data, a question that closed-form models address only for a single agent under exogenous prices or in a static equilibrium. We propose a calibrated multi-agent reinforcement learning environment for the blob fee market in which the blob base fee is endogenous, set by the EIP-4844 rule on the aggregated decisions of heterogeneous rollups fit to post-Dencun Ethereum data. After confirming that the environment reproduces the known single-agent analytical optimum, we show that independent learners, with no convergence guarantee, still settle near the target without coordination ( 3.32\pm0.01 blobs per block over five seeds). Over a fiftyfold demand range ( 10\times to 500\times ), supply stays at the target from 10\times to 200\times and rises above it at higher demand, where the fee can no longer hold it. Posting frequency increases with rollup size, an emergent analog of the size threshold predicted by prior theory. A costless layer-1 calldata outside option pushes the fee down by roughly an order of magnitude while aggregate supply stays at the target, so the fee rule acts as a negative-feedback controller on decentralized supply.
[MA-35] Persistent LLM Orchestration of Heterogeneous Robot Teams via Partial Delegation to Multi-Agent Protocols
【速读】：该论文旨在解决在动态、非终止环境中，基于自然语言指令的多机器人团队任务规划中，大语言模型（LLM）因环境持续变化和任务长期性导致的性能退化问题。现有方法依赖于任务终态与静态环境的假设，但在灾害救援等实际场景中，任务持续进行且环境不断演化（如火势蔓延、幸存者移动），使得传统框架失效。其核心挑战表现为：提示（prompt）长度无限制增长、推理能力随代理与目标数量增加而下降、以及完成一次搜索空间覆盖后停止主动监控。为应对这些问题，论文提出基于“具身生成智能”（grounded generative intelligence）框架的改进方案：首先引入马尔可夫状态摘要（Markovian state summary），通过每次调用时从语义地图重建状态，替代不断膨胀的交互历史；其次采用部分委派（partial delegation）机制，将部分机器人交由自主多智能体协议管理，仅保留对需精细参数与顺序控制的子任务进行高层规划；最后通过显式表达观测过期（observation staleness）并结合任务与事件触发的重新提示机制，确保系统持续保持警觉。在模拟城市搜救任务中，该框架在火势扩散与人员移动的动态环境下，成功救助98%的平民，显著优于最强基线的89%，同时使用令牌数减少6倍。消融实验进一步验证了各机制的有效性。
链接: https://arxiv.org/abs/2610.15224
作者: Maria G. Mendoza,Victoria M. Tuck,Zachary Ravichandran,Shankar Sastry,George J. Pappas
机构: University of California, Berkeley (加州大学伯克利分校); University of Pennsylvania (宾夕法尼亚大学)
类目: Multiagent Systems (cs.MA); Robotics (cs.RO)
备注: 8 pages, 7 figures, 3 tables
Abstract:Natural language specifications let non-expert users task robot teams, and large language models (LLMs) have recently proven effective at grounding such semantic specifications into plans for heterogeneous teams. Existing approaches assume missions that terminate and environments that, while unknown, do not change, but both assumptions fail in many emerging applications. For example, a disaster-response mission does not terminate, and its environment changes continuously as hazards spread and evacuees move. We identify three ways an LLM orchestrator breaks in this setting: its prompt grows without bound, its reasoning degrades as agents and targets accumulate, and it stops actively monitoring after covering the search space once. We build on the successful framework of grounded generative intelligence, and address each issue with a classical idea. A Markovian state summary, rebuilt from the semantic map at every invocation, replaces the growing interaction history. Partial delegation lets the orchestrator assign a subset of robots to an autonomous multi-agent protocol and reserve its reasoning for generating subtasks that require detailed parameters and careful ordering. Lastly, observation staleness, surfaced explicitly in the prompt, together with task-and event-triggered re-prompting, keeps the team vigilant. In a simulated urban search-and-rescue mission with spreading fire and moving evacuees, our framework rescues 98% of civilians versus 89% for the strongest baseline while using six times fewer tokens. Additional ablations isolate the contribution of each mechanism.
[MA-36] Continuous Risk-Based Scheduling for Food Safety Inspection: Integrating Historical Decay with Event-Driven Consumer Review Signals
【速读】：该论文旨在解决传统批量风险导向型检查调度系统在识别关键风险类别时的盲区问题，即难以发现近期通过清洁检查后却迅速出现严重违规的食品设施。传统系统依赖静态历史数据，在固定周期内重新计算风险评分，导致检查周期间存在信息空白。为此，本文提出连续检查风险（Continuous Inspection Risk, CIR）系统，这是一种事件驱动的动态调度框架，能够随着新数据的到达实时重排检查队列。其核心解决方案在于：融合衰减加权的历史风险评分与由语言模型分类器筛选的实时消费者评论流，并引入时间分区规则——严格排除最近一次检查前的评论，以防止过时信号干扰当前风险评估。基于加州圣克拉拉县2024–2026年共9,505次例行检查的数据评估表明，仅使用历史评分可使重大违规的平均检测时间缩短69天（20.9%），但无法预测任何严重虫害相关关闭事件（0/8被检测）；而加入检查后的评论信号后，成功预测了8起虫害关闭中的7起，平均提前117天预警，其中4次在评论发布当日即触发检查。进一步引入评论一致性验证规则，将精确率从0.70提升至0.78，且未牺牲召回率。研究还揭示了两个关键实施挑战：数据清洗陷阱（检查边界处陈旧评论泄露）及评估局限性（未观测到设施自我整改行为）。本研究首次基于开放数据对事件驱动式检查触发机制进行案例级评估，为基于评论增强的公共卫生监管系统提供了可复现的基准。
链接: https://arxiv.org/abs/2610.15191
作者: Akriti Bhat,Wayne Zhu,Ishan Gupta
机构: 未知
类目: Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Multiagent Systems (cs.MA)
备注:
Abstract:Batch risk-based inspection scheduling struggles to detect a critical failure class: food facilities with recent clean inspections that subsequently develop severe violations. Traditional systems recompute risk on fixed planning cycles using static historical data, leaving blind spots between inspection cycles. To address this, we introduce the Continuous Inspection Risk (CIR) system, an event-driven scheduling pipeline that re-ranks the inspection queue dynamically as new data arrives. CIR combines a decay-weighted historical score with a stream of real-time consumer reviews triaged by a language-model classifier. A key structural contribution is our temporal partition rule, which strictly excludes reviews predating the most recent inspection to prevent superseded signals from skewing current risk. Evaluating CIR on 9,505 routine inspections in Santa Clara County, California (2024-2026), we show that history-based scoring alone reduces mean time-to-detection of major violations by 69 days (20.9%) under a 30% inspection budget, but fails to anticipate severe vermin-related closures (0/8 detected). Incorporating post-inspection review signals drastically improves performance, predicting 7 of 8 vermin closures with an average lead time of 117 days-firing on the exact day of review publication in four instances. Introducing a review corroboration rule improves precision from 0.70 to 0.78 without loss of recall. Finally, we analyze critical implementation challenges, documenting a data-hygiene pitfall (stale review leakage across inspection boundaries) and an evaluation limitation involving unobserved facility self-remediation. This work provides the first case-level evaluation of event-driven inspection triggering using open data and provides a reproducible benchmark for review-augmented public health systems
[MA-37] Conflict-Aware Autonomous Resource Orchestration for 5G Edge Slicing Using Multi-Agent Systems and LLM s
【速读】：该论文旨在解决5G边缘网络切片中两个相互耦合的核心问题：一是准确理解高层运营商意图，二是动态环境下对稀缺计算与带宽资源进行高效重分配。现有集中式优化方法在快速变化的边缘资源竞争场景下面临计算开销大、响应延迟高的挑战；基于多智能体系统（MAS）的拍卖机制虽可支持资源仲裁，但仍依赖人工设定的经济参数与优先级策略，缺乏灵活性。而直接将大语言模型（LLM）嵌入周期性控制回路则可能引入不可忽视的推理延迟及非确定性行为。为此，本文提出一种双时间尺度架构：在宏观层面，基于LLM的“战略家”（Strategist）将自然语言形式的网络意图转化为经济策略参数（预算、收入及竞标激进度）；在操作层面，由SPADE驱动的切片代理采用确定性的、类维克里（Vickrey）第二价格机制，实现资源争用的实时仲裁。最终，通过直接修改用户面功能（UPF）Pod上的Kubernetes CPU限制与Calico CNI带宽注解，确保分配结果落地执行。在真实Kubernetes/Open5GS边缘集群上，通过30次重复实验验证了该架构在两类互补场景下的有效性：在智慧城市场景中，自然语言公共安全意图可在数轮拍卖内转化为可量化的切片投标与带宽分配变化；在分层切片场景中，该机制实现了基于优先级的资源与服务质量（QoE）差异化，具备对突发需求的自适应能力，并在维持配置最小保障的同时，几乎完全利用共享带宽池。其关键创新在于通过双时间尺度设计，既利用LLM理解高层语义意图，又通过确定性拍卖机制保障实时性与可预测性，从而在语义理解与系统可控性之间取得平衡。
链接: https://arxiv.org/abs/2610.14903
作者: Nathan R. Prediger,Rodrigo Moreira,Juliano Araujo Wickboldt,Tereza C. Carvalho,Pedro Oliveira,Paulo Novais,Flávio de Oliveira Silva
机构: Federal University of Rio Grande do Sul (UFRGS), Porto Alegre, Brazil; Federal University of Viçosa (UFV), Minas Gerais, Brazil; University of São Paulo (USP), São Paulo, Brazil; University of Minho (UMinho), Braga, Portugal
类目: Networking and Internet Architecture (cs.NI); Distributed, Parallel, and Cluster Computing (cs.DC); Multiagent Systems (cs.MA)
备注: Accepted author manuscript for publication in the Proceedings of the 2027 IEEE Consumer Communications Networking Conference (CCNC 2027), Las Vegas, Nevada, USA, January 9-12, 2027
Abstract:5G edge network slicing must reconcile two coupled requirements: understanding high-level operator intent and reallocating scarce computing and bandwidth resources under dynamic contention. Centralized optimization can face computational and responsiveness challenges under rapidly changing edge contention, and Multi-Agent System (MAS)-based auctions can support resource arbitration but still depend on operator-defined economic parameters and priorities. Embedding a Large Language Model (LLM) directly in a recurring control loop, by contrast, may introduce non-negligible inference latency and non-deterministic behavior. This paper proposes a two-timescale architecture that addresses this gap: an LLM-based Strategist translates natural-language network intents into economic policy parameters (budget, income, and bidding-aggressiveness) at the macro scale, while a deterministic, Vickrey-inspired second-price mechanism run by SPADE-based slice agents resolves recurring resource contention at the operational scale, with the resulting allocations enforced by patching Kubernetes CPU limits and Calico CNI bandwidth annotations directly on User Plane Function pods. We evaluate the architecture end-to-end on a real Kubernetes/Open5GS edge cluster across two complementary scenarios, each repeated 30 times: a Smart-City scenario, in which natural-language public-safety intents propagate into measurable changes in slice bids and bandwidth allocations within a few auction rounds, and a Hierarchical-slicing scenario, in which the mechanism provides priority-dependent resource and QoE differentiation, adapts to abrupt demand changes, and preserves configured minimum guarantees while maintaining near-complete allocation of the shared bandwidth pool.
[MA-38] MARL-BattleGrounds: A Novel JAX-native Benchmark for Heterogeneous and Competitive Multi-agent RL
【速读】：该论文旨在解决多智能体强化学习（MARL）基准测试中“简单高效”与“复杂昂贵”之间的空白问题，即现有基准要么过于简化（如SMAX仅依赖单位微操），要么计算成本过高（如Dota 2），缺乏兼具丰富性、可扩展性与计算效率的中等复杂度竞争性环境。其解决方案的关键在于构建一个基于JAX原生实现的可复用、样本高效且以研究者为中心的基准平台——MARL-BattleGrounds（MARL-BGs）。该平台的核心创新包括：1）引入异构五人小队对战模式，每类智能体具有独特基础能力、被动技能和终极技能，强调团队协作与对抗策略；2）设计52张手绘地图，包含12张课程化训练地图与10张独立验证/测试地图，支持渐进式学习；3）提供两种标准化评估机制——硬性泛化挑战（端局场景下从劣势逆转强反应控制器）与月度循环赛制（基于公平算力重训练并发布Elo排名），确保算法性能可比性；4）集成高粒度数据记录（每局11,192个指标，覆盖28个维度）、脚本对手、分析回放工具及自洽/交叉学习、零样本协同等支持框架；5）开放API支持大语言模型（LLM）代理接入，并以Qwen3.5-9B为例验证其在真实对抗中的表现；6）通过BattleClient实现人类与任意策略混合交互，推动人机协同评估。整体系统在单张RTX 5090显卡上实现每秒5万步5v5游戏仿真，显著提升了MARL研究的实验效率与可重复性。
链接: https://arxiv.org/abs/2610.14883
作者: Ulixes Tariq Hawili,Subramanian Ramamoorthy,Yoann Altmann,Ignacio Carlucho
机构: University of Edinburgh (爱丁堡大学); Heriot-Watt University (赫瑞瓦特大学)
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multiagent Systems (cs.MA)
备注: 34 pages, 14 figures, 18 tables. Project website: this https URL
Abstract:We present MARL-BattleGrounds (MARL-BGs), an easy-to-use JAX-native benchmark for heterogeneous and competitive multi-agent reinforcement learning (MARL), designed to be sample-efficient, computationally efficient, and researcher-centric. MARL benchmarks are typically simple and efficient, like SMAX, where winning reduces to unit micromanagement, or rich and costly, like Dota 2. MARL-BGs fills this missing middle. Its first mode, Team Deathmatch, is a simultaneous turn-based game between two teams of five heterogeneous agents. The five classes have distinct basic, passive, and ultimate abilities designed to synergize with allies and counter enemies. Our 52 hand-designed maps with diverse obstacles include a 12-map curriculum and ten held-out validation and test maps. MARL-BGs provides two canonical evaluation modes. First, eight endgame scenarios form a hard generalization challenge: each asks a team to win an unseen, nearly lost position against a strong reactive controller within 5-10 ticks. Second, a monthly round-robin tournament retrains each submission on equal compute, plays it against every member of the current “Big Nine” field, and publishes the Elo results on the official leaderboard. All entrants are publicly released as downloadable opponents. MARL-BGs records 11,192 metrics per game in 28 topics. We also include three strong scripted opponents, an analytical replay viewer, and off-the-shelf tools for self-play, cross-play, curriculum learning, and zero-shot coordination evaluation. We even provide an API for LLM-based agents and use it to evaluate Qwen3.5-9B against our strongest policy. In our BattleClient, humans can play alongside or against any mix of policies, enabling human-AI evaluation. We benchmark MARL-BGs with six additional deep MARL baselines, which we release. On one RTX 5090 with 1,024 parallel games, the simulator runs 50,000 5v5 game steps per second.
[MA-39] EAIR: Interaction Risks in Enterprise LLM Multi-Agent Systems MICRO
【速读】：该论文旨在解决企业在采用大语言模型（LLM）代理进行多角色协作时，因各代理仅感知局部工作流而引发的跨组织约束违反问题。此类问题表现为：单个代理在本地范围内采取合理行为（如各自发放信用额度），但集体行动却违反全局性企业约束（如总补偿限额）。其解决方案的关键在于提出首个面向企业多代理交互风险的分析框架——EAIR，其核心贡献包括：（1）形式化企业交互风险，基于需保持的组织属性定义风险，并构建可测试的风险分类体系，明确失败标准与证据要求；（2）开发可执行的微世界仿真环境EAIR-Worlds，通过受控、可观测的工作流实例化特定风险，包含显式角色、策略、权限、源事实及执行轨迹；（3）设计一种新型诊断与修复流水线，利用执行轨迹定位关键交互边界，指导针对性工作流修复并重新验证系统；（4）大规模实验验证了多种组织属性下反复出现的交互失效现象。EAIR首次实现了对企业多代理交互故障的系统性诊断与缓解，为复杂企业场景下的安全协同提供可操作的技术支撑。
链接: https://arxiv.org/abs/2610.14800
作者: Yuchen Ma,Swadheen Shukla,Paulo Salem,Yue Huang,Robert Sim,Stefan Feuerriegel,Michel Galley,Jianfeng Gao
机构: LMU Munich(慕尼黑大学); Microsoft(微软); University of Notre Dame(圣母大学)
类目: Multiagent Systems (cs.MA)
备注: Toolkit: this https URL
Abstract:LLM agents increasingly divide enterprise work across roles with different goals, authority, access, and information. Because each agent often sees only a local snapshot of the enterprise workflow, locally reasonable actions can still violate organization-wide constraints (e.g., several service agents can each issue credit that is individually permitted while collectively exceeding a shared compensation cap). We introduce EAIR, the first framework for analyzing enterprise multi-agent interaction risks. We make four contributions. (1) We formalize enterprise interaction risk in terms of the organizational properties that workflow operations must preserve, and operationalize this formulation as a taxonomy of testable risks with explicit failure criteria and evidence requirements. (2) We build EAIR-Worlds, a suite of executable micro-worlds that instantiates selected risks as controlled, observable workflows with explicit roles, policies, permissions, source facts, and execution traces. (3) We develop a novel diagnosis-and-repair pipeline that uses the resulting execution traces to locate contributing interaction boundaries, guide targeted workflow repairs, and retest the repaired system. (4) Large-scale experiments show recurring interaction failures across the organizational properties studied. Together, EAIR is the first operational framework for diagnosing and mitigating enterprise multi-agent interaction failures.
[MA-40] sGLP: Specifying Stochastic Agents that Simulate People
【速读】：该论文旨在解决基层平台（grassroots platforms）在大规模部署前缺乏有效评估与测试手段的问题。由于基层平台依赖参与者自主治理，其真实运行需大量真实用户参与，而大规模真实测试与“由参与者主导”的设计原则相矛盾，难以实现。为此，论文提出一种基于随机逻辑编程的解决方案——随机式草根逻辑程序（stochastic GLP, sGLP），其核心在于通过可扩展的随机代理（stochastic agents）替代真实用户和界面，从而在虚拟环境中对基层平台进行规模化仿真。sGLP是带有概率行为扩展的多智能体并发逻辑编程语言，能够将用户交互行为建模为具有统计特性的行为分布，并以代理形式模拟不同社会特征个体的行为模式。关键创新在于：将原用于表达人类意愿的志愿保护草根逻辑程序（vGLP）中的用户界面与人替换为类型对偶的随机代理，这些代理依据预设的随机行为配置（如亲缘倾向、社交倾向、消费习惯等）生成符合真实社会行为规律的互动模式。研究通过在草根社交图谱和草根货币系统上的实验验证了该方法的有效性，并证明了使用sGLP代理仿真的每一轮执行均为原始vGLP程序的有效运行实例。该方案实现了对复杂社会性系统的可重复、可扩展仿真，为基层平台的设计与验证提供了理论保障与实践工具。
链接: https://arxiv.org/abs/2610.14738
作者: Ehud Shapiro
机构: London School of Economics (伦敦政治经济学院); Weizmann Institute of Science (魏茨曼科学研究所)
类目: Multiagent Systems (cs.MA); Distributed, Parallel, and Cluster Computing (cs.DC); Human-Computer Interaction (cs.HC); Programming Languages (cs.PL)
备注:
Abstract:Grassroots platforms are designed to be owned, operated and governed by their participants, and thus are constituted from person-to-person smartphone apps that run on participants’ smartphones. A grassroots app is specified using a grassroots social contract, which is compiled through an abstraction cascade into a volition-guarded Grassroots Logic Program (vGLP), a multiagent concurrent logic programming language. A vGLP program in turn specifies a smartphone UI via which people express their volitions and thus control the app. Prudence dictates large-scale evaluation and testing of grassroots platforms prior to deployment, but using the many people needed to do that is a contradiction in terms. Hence another solution is needed. Here, we present stochastic GLP (sGLP), designed to allow grassroots platforms to be simulated at scale from the vGLP programs that deploy them, with stochastic agents in place of people. sGLP is GLP with a rate extension for specifying stochastic behaviour. A stochastic agent is an sGLP program replacing both the vGLP UI and the person behind it. It is typed by the dual of the interactive type of the UI and specifies a stochastic profile of a person’s behaviour. We demonstrate sGLP on the grassroots social graph, with a population of men and women drawn from four stochastic profiles — homophile or indifferent, wary or sociable — and on grassroots currencies, with four profiles — spenders or savers, cautious or generous — and report what the mix of profiles does to the social graph that forms and to the currencies that circulate. We prove that every run of a vGLP program in which sGLP agents replace the UI and the people behind it is a vGLP run. We implemented sGLP in GLP by Gillespie’s method, and ran reproducible simulations with it. Subjects: Multiagent Systems (cs.MA); Distributed, Parallel, and Cluster Computing (cs.DC); Human-Computer Interaction (cs.HC); Programming Languages (cs.PL) Cite as: arXiv:2610.14738 [cs.MA] (or arXiv:2610.14738v1 [cs.MA] for this version) https://doi.org/10.48550/arXiv.2610.14738 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Ehud Shapiro [view email] [v1] Sat, 10 Oct 2026 08:58:21 UTC (207 KB)
[MA-41] Does Failure Feedback Direct Agent Self-Modification?
【速读】：该论文旨在解决自改进智能体（self-improving agents）在利用失败反馈进行代码或指令修正时，难以实现目标特定修复（target-specific repair）与泛化能力的问题。现有方法虽能通过选择机制提升性能，但无法确保修复行为具有明确的目标指向性或跨任务迁移能力。其解决方案的关键在于引入目标特定反馈（target feedback），并通过控制实验验证其对修复效果的促进作用：在小规模开源模型上，相关反馈可显著提升目标故障的修复率；在针对Qwen3-4B和Phi-4-mini的对照研究中，目标反馈分别带来20个百分点和7个百分点的修复率提升（仅前者显著）。研究进一步揭示，目标修复并不必然导致未见任务上的性能增益，且实际收益集中于可识别、可修复的故障机制场景，如人为植入错误与自然工具接口失效。此外，编辑定位、回归筛选等策略的效果因情境而异，而八步修订循环与生成式强化学习框架（GEPA）在保持样本外准确率方面表现相当。因此，论文强调评估应区分报告目标修复率、向相关故障的迁移能力以及未见任务的净变化，以更全面衡量智能体的改进质量。
链接: https://arxiv.org/abs/2610.14662
作者: Rong He
机构: 未知
类目: oftware Engineering (cs.SE); Multiagent Systems (cs.MA)
备注: 28 pages, 8 figures, 8 tables. Code, derived data and a script that rederives every number in the paper are included as ancillary files
Abstract:Self-improving agents use failure feedback to revise their code or instructions, but gains after selection do not establish target-specific repair or generalization. We measure directedness by comparing repair of a randomly selected target failure with repair of the parent’s other failures. Across 145 eligible modifications in public Darwin Gödel Machine logs, we find no detectable target-specific advantage. Controlled instruction-revision experiments with small open models show that relevant feedback can improve target repair. In a second study with a pre-specified analysis plan, target feedback raised repair by 20 percentage points for Qwen3-4B relative to matched same-category feedback; the 7-point increase for Phi-4-mini was not significant. Target repair did not reliably imply held-out improvement. Held-out gains were concentrated in settings with identifiable, repairable failure mechanisms, including planted errors and natural tool-interface failures. Benefits of edit localization and regression screening varied across settings, while eight-step revision loops and GEPA ended at similar mean held-out accuracy. Evaluations should report target repair, transfer to related failures, and held-out net change separately.
[MA-42] Beyond Independent Values: Optimal Auctions with Correlated Valuations for LLM -Native Advertising
【速读】：该论文旨在解决大语言模型（Large Language Model, LLM）广告拍卖中因广告主价值存在高度相关性而引发的效率损失问题。在对话式助手场景下，多个广告主竞争同一赞助位，且均基于相同的用户对话内容进行出价，导致其出价意愿呈现显著正相关（affiliation），但现有拍卖机制将这些价值视为独立，从而未能有效捕捉这一经济特性，造成潜在收入损失。论文提出一种名为AffMyerson的新拍卖机制，其核心创新在于通过建模广告主价值为多元正态分布，并将经济上的正依赖关系（affiliation）转化为对精度矩阵（precision matrix）的符号约束条件，进而设计一种满足该约束的神经网络参数化结构。该设计确保了无论先验分布如何选择，拍卖始终是诚实（truthful）且事后个体理性（ex-post individually rational）的，估计误差仅影响收益而不破坏策略一致性。进一步地，该方法可闭式求解条件虚值（conditional virtual value），并通过巧妙的先验设计避免矩阵求逆，实现O(n²)时间复杂度的实时定价。研究还拓展至对数正态族，证明其在正确先验下为收益最优，且当条件方差低于宽松阈值时，虚值函数全局正则，符合实际应用需求。实验表明，在LLM代理投标者场景下，AffMyerson相比调优保留价的第二价格拍卖提升6.2%收益，优于Ronen的前瞻拍卖6.1%，并显著超越假设独立性的Myerson拍卖9.2%；即使在真实关键词搜索查询中，面对难以准确读取的支付意愿，仍能实现4.5%的收益增益。这一性能提升源于对共同需求驱动的相关性建模，尤其体现在基于对话语境联合出价的代理与仅执行计算任务的代理之间的行为差异上。
链接: https://arxiv.org/abs/2610.14646
作者: Shiman Sheng,Mingyu Guo
机构: The University of Adelaide(阿德莱德大学)
类目: Computer Science and Game Theory (cs.GT); Multiagent Systems (cs.MA)
备注: 23 pages, 4 figures
Abstract:Conversational assistants are becoming the place where people ask what to buy, and sponsored content is moving into their answers. Several advertisers then compete for one sponsored slot, and all of them read the same conversation, so their willingness to pay usually moves together. Existing auctions for LLM advertising treat these values as independent. We therefore present AffMyerson, an auction that captures this correlation and turns it into revenue. We model the advertisers’ values given the conversation as a multivariate normal and turn the economic premise of positive dependence (affiliation) into a sign condition on the precision matrix. We design a neural parameterization whose every output satisfies it. We prove that every prior in this family yields a truthful, ex-post individually rational auction, so estimation error can cost revenue but never truthfulness. We derive the conditional virtual values in closed form. By the design of the prior, pricing an auction needs no matrix inversion and takes O(n^2) time, fast enough to run at query time. We also work out the log-normal family used in empirical ad-auction work. There we prove that the auction is truthful for every prior and revenue-optimal when the prior is correct, and that the virtual value is regular everywhere whenever the conditional spread is below a loose threshold, which our fitted priors stay well under. With LLM-agent bidders, AffMyerson earns 6.2% more than a second-price auction with a tuned reserve, 6.1% more than Ronen’s lookahead auction with the same prior, and 9.2% more than Myerson’s auction run as if values were independent. This gain is still 4.5% on real keyword search queries where willingness to pay is hard to read. The correlation comes from common demand. Agents that judge the conversation bid together and agents told to calculate do not. The gain from conditioning moves with the correlation.
[MA-43] Iterate on an LLM Serve on an SLM NEURIPS2026
【速读】：该论文旨在解决在大型语言模型（LLM）推理成本不断上升的背景下，如何高效、经济地实现小语言模型（SLM）在特定任务中具备准确且稳定行为的问题。当前，尽管SLM在成本上具有显著优势（可比前沿模型低5-40倍），但其在实际生产环境中往往难以通过零样本提示（zero-shot prompting）获得可靠表现，而频繁重新训练又无法满足快速迭代的需求。为此，论文提出一种分阶段解决方案：首先在前沿模型上通过提示工程（prompt engineering）在真实流量反馈下快速构建期望行为；待行为收敛后，将真实生产数据中记录的实际决策行为（而非合成模拟）作为监督信号，对SLM进行有监督微调（SFT）。该方法在家庭护理机构调度代理的应用中得到验证，使用GPT-5.4-nano生成单个“是/否”安全判断，再以11.6k条真实生产数据微调LoRA适配的Qwen3-8B（8B参数量），最终达到0.913的一致性评分，显著优于教师模型自身重运行时72.7%的一致性水平，表明学生模型不仅更经济，还具备更高一致性。此外，在分歧案例中，专业调度员盲评更倾向支持学生模型（48 vs 29），且学生模型的中位延迟降低约2.5倍，成本仅为原模型的极小部分。因此，该研究提出的“快速构建—实时反馈—低成本推理”生命周期框架，为构建高效、可扩展的智能体系统提供了可复用的技术蓝图。
链接: https://arxiv.org/abs/2610.14576
作者: Keegan Wang
机构: Carnegie Mellon University (卡内基梅隆大学)
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: Accepted to the NeurIPS 2026 2nd Workshop on Agentic AI Benchmarks and Applications for Enterprise Tasks (AABA4ET)
Abstract:As frontier LLMs gain more capabilities, the cost to run them also goes up. At the same time, small language models (SLMs) also gain more capabilities but have largely been able to keep their cost stable, offering 5-40 times cheaper inference for the narrow, repetitive subtasks that agents decompose work into. However, the standing limiting factor is getting the correct behavior out of a small model in a fast, efficient, and economical way. Unfortunately, in most cases, zero-shot small models fail narrow production tasks. Furthermore, retraining one every time the behavior changes is too slow to iterate on. We propose doing both in sequence. First, develop the desired behavior in prompt space, on a frontier model, with immediate feedback from live traffic. Then, once the behavior converges, distill the recorded production behavior (not synthetic imitations of it) into an SLM through supervised fine-tuning (SFT). We show a proper use case of this method through a scheduling agent for home-care agencies that fans out a single yes/no safety decision per caregiver-shift pair on GPT-5.4-nano. A LoRA-tuned Qwen3-8B trained on 11.6k recorded production decisions reaches 0.913 agreement, while the teacher (GPT-5.4-nano), re-run on its own recorded prompts, agrees with itself only 72.7% of the time, pointing to the fact that the SLM may also be more consistent. Furthermore, when the two disagreed, a blinded professional scheduler sided with the student 48-29. Finally, we find that the SLM has approximately 2.5 times lower median latency and is a fraction of the cost. We believe this lifecycle of quick building, live feedback, and cheap inference is a blueprint for others building similar systems.
[MA-44] FDE-Bench: Evaluating End-to-End Delivery from Underspecified Real-World Business Requests
【速读】：该论文旨在解决现实商业任务中因前端从业者难以准确表述隐性经验规则而导致需求不明确的问题，其核心挑战在于如何在缺乏显式目标的情况下，通过交互式需求澄清与迭代优化实现端到端的解决方案交付。解决方案的关键在于构建首个基于真实客户项目的基准测试集——FDE-Bench，该基准涵盖49个涉及组合优化与机器学习的实际案例，其需求澄清目标源自一线从业者与前线部署工程师（FDEs）的真实对话。评估机制要求智能体通过模拟与客户交互来补全缺失需求，并由针对具体案例的评估器以经客户认可的FDE解决方案作为质量参考，对最终产出进行评分。实验结果显示，在18种智能体配置中，最高平均交付得分仅为0.5510（相对于FDE参考值1.0），最高交付成功率（DSR）仅6.12%，表明当前智能体在处理未充分定义的真实商业任务时，仍远未达到人类专家水平的端到端交付能力。
链接: https://arxiv.org/abs/2610.14539
作者: Huaiming Li,Can Huang,Chufan Wu,Danyu Liu,Xiaomin Yuan,Qinze Li,Yui Lo,Wenhui Bai,Qianlong Wang,Zengle Ge,Xiyu Yan,Yuheng Cheng,Jinhui Ren,Guokai Chen,Jiaqun Liu,Haoran Li,Qianhui Liu,Annan Li,Jianmin Wu,Dawei Yin,Dou Shen
机构: 未知
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注:
Abstract:Real-world business requests often leave key requirements unstated because frontline practitioners struggle to articulate rules they know from experience. Forward-deployed engineers (FDEs) must elicit these requirements through dialogue and iteratively optimize solutions until they meet customer acceptance criteria. Existing optimization benchmarks typically start from explicit objectives, while clarification benchmarks often introduce information gaps into existing tasks. We introduce FDE-Bench, the first benchmark built from real customer projects to evaluate agents on end-to-end delivery tasks that require interactive requirement clarification and iterative optimization. Its 49 cases span combinatorial optimization and machine learning. Clarification targets are distilled from discussions between frontline practitioners and FDEs. Agents elicit missing requirements from a simulated customer, and case-specific evaluators assess final artifacts against acceptance criteria using customer-accepted FDE solutions as quality references. Across 18 agent configurations, the highest mean delivery score is only 0.5510, compared with an FDE reference of 1.0. The highest delivery success rate (DSR), defined as the fraction of planned runs reaching this reference, is only 6.12%. These results show that reliable FDE-level end-to-end delivery remains out of reach for the evaluated agents on underspecified real-world business tasks. The benchmark project page is available at this https URL.
[MA-45] SEPM: Self-Evolving Procedural Memory Grounded in Task Graphs for Multi-Agent Collaboration
【速读】：该论文旨在解决长时程多智能体协作中协同行为与任务依赖关系脱节的问题，即现有记忆系统在支持信息共享或程序复用时，未能有效将共享程序的演化过程与协作智能体的任务依赖关系及执行状态相耦合，从而限制了个体智能体经验向团队场景的直接迁移。其解决方案的关键在于提出自演化程序记忆（Self-Evolving Procedural Memory, SEPM）框架，通过任务依赖图（task dependency graph）作为统一表征，实现在线协作与程序记忆演化的协同。在执行过程中，轻量级共享黑板记录结构化证据与任务依赖关系，支持对目标、计划和世界状态偏差的检测，并基于证据进行实时校准；执行后，经验证的动作与依赖关系被提炼为可复用的程序，重复失败则生成警告与约束。共享程序池通过创建、精炼、专业化和退役等机制动态演化，其过程由成功性与成本效用驱动，并受证据与安全要求约束。该机制有效保留了协作结构与适用条件，实现了跨任务的经验复用。在ALFWorld、WebArena和OfficeBench三个基准测试环境下，结合AutoGen与DyLAN架构的实验表明，SEPM在全部六个设置中均达到最高任务得分，相较于无记忆基线，分别提升29.85、25.86和13.00个百分点。
链接: https://arxiv.org/abs/2610.14515
作者: Yi Liu,Junwei Zhao,Li Sun,Jiansong Zhang,Sen Su
机构: Beijing University of Posts and Telecommunications (北京邮电大学); China Telecom (中国电信)
类目: Multiagent Systems (cs.MA)
备注:
Abstract:Long-horizon multi-agent collaboration requires memory that preserves both successful actions and the coordination structure underlying them. Existing memory systems support information sharing or procedural reuse, but often leave the evolution of shared procedures disconnected from the task dependencies and execution states of collaborating agents. This coupling also limits the direct transfer of procedural memory designed for individual agents to team settings. We propose Self-Evolving Procedural Memory (SEPM), a framework that uses task dependency graphs as a common representation for online coordination and procedural memory evolution. During execution, a lightweight shared blackboard records structured evidence alongside task dependencies, supporting the detection of goal, plan, and world-state divergence and evidence-based realignment. After execution, verified actions and dependencies are distilled into reusable procedures, while recurring failures inform warnings and constraints. The shared procedure pool evolves through creation, refinement, specialization, and retirement, guided by success and cost utility and constrained by evidence and safety requirements. Together, these mechanisms preserve the collaborative structure and applicability conditions needed to reuse experience across tasks. Evaluations on ALFWorld, WebArena, and OfficeBench under AutoGen and DyLAN show that SEPM achieves the highest task scores among the compared methods in all six benchmark settings. Under AutoGen, it improves task scores over the no-added-memory baseline by 29.85, 25.86, and 13.00 percentage points, respectively. The codebase is available at this https URL.
[MA-46] Optimize Once Execute Fast: Latency-Aware Multi-Agent Workflow Learning for Recurrent Queries
【速读】：该论文旨在解决大型语言模型驱动的多智能体系统（LLM-based Multi-Agent Systems, LLM-based MAS）在复杂任务求解中因手动设计工作流（workflow）成本高且泛化性差，而现有自动化方法又仅关注下游任务性能、忽视执行时间（即任务完成时间，makespan）所导致的延迟累积问题。尤其在重复性应用场景（如AI客服支持、AI云故障排查）中，频繁调用同一工作流会使其延迟随时间不断累积，严重影响响应速度。为此，论文提出“延迟感知的工作流生成”（latency-aware workflow generation）这一新问题，其核心目标是在保障任务性能的前提下最小化工作流的总执行时间。解决方案的关键在于提出LAWA（Latency-Aware Workflow Optimization）框架，通过基于理论支撑的图结构编辑策略，精准识别并优化关键路径上的瓶颈环节，从而显著降低工作流的makespan。LAWA通过将图生成器与图编辑过程统一集成，既支持全新工作流的生成，也支持对已有工作流的迭代优化，在七个基准测试中均实现了显著的执行时间压缩，同时保持了与现有方法相当的任务性能，有效支持多智能体工作流在重复查询场景下的高效复用。
链接: https://arxiv.org/abs/2610.14481
作者: Enpei Zhang,Feiyu Qu,Zheng Huang,Dawei Zhou,Elynn Chen,Yujun Yan
机构: Dartmouth College(达特茅斯学院); Virginia Tech(弗吉尼亚理工学院); New York University(纽约大学)
类目: Multiagent Systems (cs.MA)
备注:
Abstract:Large Language Model-based Multi-Agent Systems (LLM-based MAS) have shown remarkable success in solving complex tasks by coordinating specialized agents through multi-step workflows. To overcome the high cost and specificity of manually designed workflows, recent research has shifted toward learning them automatically. However, a key limitation remains: while agent calls introduce significant latency due to model inference time and service congestion, most automated methods optimize solely for downstream performance, ignoring execution time (makespan). This issue is amplified when workflows are reused across recurring queries, making latency accumulate over time. This limits deployment in recurrent applications such as AI customer support and AI cloud troubleshooting, where users expect fast responses. To bridge this gap, we introduce a new problem, termed latency-aware workflow generation, where the objective is to minimize workflow makespan while aiming to retain downstream performance. We propose LAWA (Latency-Aware Workflow optimizAtion), a novel framework that reduces makespan via theoretically grounded graph edits targeting critical-path bottlenecks. By unifying a graph generator with this graph-editing procedure, LAWA supports both de novo workflow generation and refinement of existing ones. Empirical results across seven benchmarks demonstrate that LAWA significantly reduces makespan while achieving competitive task performance, supporting efficient reuse of multi-agent workflows for recurrent queries.
[MA-47] What Counts as the Same Task? Executable Contracts for Robustness Evaluation of Tool-Using Agents
【速读】：该论文旨在解决当前工具使用型智能体（tool-using agents）鲁棒性测试中任务一致性判定标准不严谨的问题。现有评估方法常通过修改环境（如在发票中隐藏恶意文本）来测试智能体的鲁棒性，但其有效性依赖于修改后的测试用例是否仍构成对原任务的有效检验。然而，多数评估采用弱化判据（如仅要求编辑后输入语法正确或嵌入空间距离相近），导致大量实质上已改变任务语义的测试用例被错误接受，从而误导评估结果。为此，论文提出可执行的任务合约（executable task contracts），其核心在于明确定义任务所依赖的关键事实（key facts）、允许的编辑类型以及任务完成的标准，并在智能体执行前对每个修改后的测试用例进行形式化验证。通过在14,336个预存的智能体运行记录上对比不同有效性判据的表现，研究发现基于嵌入空间阈值（如MiniLM余弦相似度）的方法虽能接受75–100%的语义变更（如支付方替换），却误拒24–45%有效的恶意样本；而仅依赖重放历史解法或允许基准字段修改的策略同样存在类似失效问题。研究表明，评估应明确并强制验证何种情形下仍视为同一任务，以确保评估结果的可信性与有效性。
链接: https://arxiv.org/abs/2610.14477
作者: Shaikh Akib Shahriyar,MD. Akil Raihan Iftee,Armon Barton,Matthew Wright
机构: Rochester Institute of Technology (罗切斯特理工学院); Naval Postgraduate School (海军研究生院)
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Multiagent Systems (cs.MA)
备注:
Abstract:Robustness tests for tool-using agents often change the agent’s environment, for example by hiding a hostile note in an invoice, and then judge the agent by what it does. The result tells us about the original task only if the changed test case is still a valid test of that task. Many evaluations decide this with weak rules, such as accepting any well-formed edit or any edit that stays close to the original in embedding space. We propose executable task contracts. A contract states which facts define a task, which edits are allowed, and what finishing the task means, and it checks each changed test case before the agent runs. We compare validity rules by scoring the same 14,336 stored agent runs under each rule. Across four encoders, embedding thresholds allowed 75-100% of edits that changed the task, such as swapping the payee, yet rejected 24-45% of valid hostile notes. With a MiniLM cosine threshold, Mistral appeared vulnerable on all 64 tasks, but only 26 had a successful valid attack. Replaying a saved solution and allowing edits to benchmark fields fail in similar ways. Evaluations should state, and check, what counts as the same task.
[MA-48] Jevolution: A Framework for Studying Communication and Collective Dynamics in AI Agents
【速读】：该论文旨在解决多智能体系统中远程决策代理在实际运行过程中因环境动态变化、仿真器对动作可用性的约束以及个体出生与迁入导致的行动主体不确定性等问题，这些问题使得实验结果难以复现且因果推断困难。其解决方案的关键在于提出Jevolution——一个可仪器化（instrumented）的空间捕食者-猎物框架，通过显式建模世界演化过程、异步请求调度、控制器适配器及实验记录等模块，将原本隐含的系统依赖关系结构化。该框架整合了局部观测、能量与资源约束、预定义信号及继承性标量特征，并基于ODD协议进行机制描述，明确区分可实现的控制变量与需进一步干预的研究问题。通过自由合成的测试用例验证了调度、通信与遗传行为的软件正确性，而非模型智能或生态合理性。一项完整的通信案例研究展示了使用Jev控制双物种的端到端工作流，结果显示在6个不同的栖息地-种子组合下，有信号与无信号条件下兔子数量的标准化差异均值为-0.0095（95%置信区间[-0.0772, 0.0602]），表明信号系统的影响方向尚不显著。因此，本文的核心贡献并非新型学习算法或涌现语言等现象的展示，而是一个可审查的实验系统及其证据边界界定。
链接: https://arxiv.org/abs/2610.14449
作者: Xirui Huang,Nikhil Hooda,Utkarsh Gupta
机构: University of Waterloo (Waterloo大学)
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 29 pages, 5 figures, 12 tables. Project website: this https URL
Abstract:Experiments with remote decision-making agents depend on more than the policy: the world may change during inference, the simulator determines which actions are available, and births and immigration alter who can act. We present Jevolution, an instrumented spatial predator-prey framework that makes these dependencies explicit. Its reference implementation separates the evolving world, asynchronous request scheduling, controller adapters, and experiment records. The model combines local observations, energy and resource constraints, predefined signals, and inherited scalar traits. We describe these mechanisms using the ODD protocol and distinguish implemented controls from research questions requiring additional interventions. Free synthetic fixtures exercise scheduling, signaling, and inheritance against the pinned implementation; they validate specified software behavior rather than model intelligence or ecological realism. A completed communication case study illustrates the end-to-end workflow using Jev to control both species. Across 6 complete habitat-seed pairs, the mean signals-on minus signals-off difference in normalized rabbit abundance was -0.0095 (nominal 95% pair-bootstrap interval [-0.0772, 0.0602]). This small study does not resolve the direction of the signal-system effect. The contribution is an inspectable experimental system and its evidence boundaries, not a new policy-learning algorithm or a demonstration of emergent language, adaptive selection, or biological predictive validity.
[MA-49] ANASSA: From High-Level Framework to Low-Level Architecture for Spatial Agent ic AI
【速读】：该论文旨在解决当前生成式 AI (Generative AI) 系统在地理空间数据处理中面临的架构落地难题，即如何将高层级的治理性责任边界（如任务编排、空间验证、治理控制、可追溯性等）系统性地分解为可执行的低层级架构实现。其解决方案的关键在于采用设计科学研究（Design Science Research, DSR）方法，将 ANASSA 框架中的高层次架构（High-Level Architecture, HLA）责任映射为具体的低层级架构（Low-Level Architecture, LLA），构建出一套具备明确逻辑角色（共47个）、版本化任务图、基于能力的代理选择机制、类型化契约、显式运行时状态、空间验证机制、治理控制、故障处理与恢复路径以及可追溯性记录的可实施架构体系。该设计在保持原有责任边界的前提下，实现了对治理性空间智能工作流的可观测性、可重建性和可测试性，且不绑定特定的 AI 模型、地理空间平台、通信协议或部署环境，为未来系统的实现提供了技术无关的、可扩展的架构桥梁。
链接: https://arxiv.org/abs/2610.14382
作者: Constantinos Papantoniou,Brian Hilton
机构: Anassa.ai(安萨人工智能); Research Scientist(研究科学家)
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA); Software Engineering (cs.SE)
备注: 14 pages, 2 figures, 13 tables. Design Science Research (DSR) study
Abstract:Agentic AI systems operating over geospatial data require more than autonomous reasoning and tool use; they must also support explicit orchestration, spatial validation, governance, provenance, failure handling, and human decision authority. Prior work introduced ANASSA, an agentic AI orchestration framework for spatial intelligence that established high-level components and responsibility boundaries for governed agentic execution (arXiv:2609.14824). However, the framework did not specify how those high-level responsibilities should be systematically decomposed into an implementation-facing low-level architecture. This paper addresses that gap by applying Design Science Research (DSR) to refine the ANASSA framework from high-level architectural (HLA) responsibilities into a low-level architecture (LLA). The resulting design preserves the responsibility boundaries of the foundational framework while translating them into executable architectural structures, including 47 logical roles, versioned task graphs, capability-based agent selection, typed contracts, explicit runtime states, spatial validation mechanisms, governance controls, failure and recovery paths, and provenance records. The architecture is demonstrated through an end-to-end execution trace and analytically evaluated for requirements traceability, internal consistency, responsibility preservation, and technology independence. The resulting high-level-to-low-level design provides an implementation-facing bridge between the ANASSA framework and future system implementations, defining how governed spatial agentic workflows can be made inspectable, reconstructable, and testable without binding the architecture to a specific AI model, geospatial platform, communication protocol, or deployment environment.
[MA-50] Akule: Fast and Scalable Multi-Robot Motion Planning via Sparse Interaction Diffusion
【速读】：该论文旨在解决大规模多机器人运动规划（Multi-Robot Motion Planning, MRMP）中基于扩散模型的轨迹先验在扩展至大规模机器人团队时面临的计算瓶颈问题，即如何在不进行昂贵联合计算的前提下，有效协调个体机器人的运动。其核心解决方案是提出一种稀疏交互扩散框架Akule，通过解耦“哪些交互值得计算”与“如何组合其影响”两个关键步骤，实现高效且可扩展的协同去噪更新。该方法引入三个可学习模块：一个可复用的成对修正模型 $ R_\phi $，一个上下文感知的选择器 $ U_\psi $ 用于决定评估哪些修正项，以及一个基于上下文的加权组合器 $ G_\omega ，以生成依赖上下文的带符号权重来融合修正效应。其中， R_\phi $ 通过匹配的成对轨迹优化进行训练，而 $ G_\omega $ 和 $ U_\psi $ 则从物理上优化的联合轨迹中学习，无需预设真实交互图。最终生成的联合提案由下游基于搜索的规划器验证与修复。实验表明，Akule在多种MRMP基准测试中实现了具有竞争力的规划成功率，最高可达25倍的完整规划加速，支持零样本群体迁移，并成功处理最多256个机器人的场景；消融实验证明，学习型选择与组合机制显著降低了修复前的冲突率及后续修复工作量，优于启发式选择与均匀叠加策略。
链接: https://arxiv.org/abs/2610.14248
作者: Manav Vora,Gokul Puthumanaillam,Aditya Penumarti,Jane Shin,Melkior Ornik
机构: 未知
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: Project website: this https URL
Abstract:Diffusion models provide expressive trajectory priors for multi-robot motion planning (MRMP), but scaling them to large teams requires coordinating individual motions without expensive joint computation. We introduce Akule, a sparse interaction diffusion framework that augments a frozen single-robot prior with three learned modules: a reusable pairwise correction model R_\phi , a contextual selector U_\psi that determines which corrections to evaluate, and a composer G_\omega that combines their effects through context-dependent signed weights. This decomposition separates which interactions merit computation from how their effects should be combined into coordinated denoising updates. We train R_\phi using matched pairwise trajectory optimizations, then learn G_\omega and U_\psi from physically optimized joint trajectories without requiring a prescribed ground-truth interaction graph. The resulting joint proposals are passed to a downstream search-based planner for validation and repair. Across diverse MRMP benchmarks, Akule achieves competitive planning success, complete-planner speedups of up to 25\times , zero-shot population transfer, and successful planning with up to 256 robots. Ablations further demonstrate that learned selection and composition reduce pre-repair conflicts and downstream repair effort relative to heuristic selection and uniform summation.
[MA-51] CoEvoke: Evoking Multi-Agent Collaboration for Trajectory Collection
【速读】：该论文旨在解决多智能体协作训练中因任务求解轨迹缺乏有效反馈与修正机制而导致的“即时通过”或“重复失败但无纠正”的问题。其核心挑战在于如何在轨迹收集阶段主动引入可恢复的困难情境，以激发智能体间的检查、反馈与修复等协作行为，从而生成具有可解释性和指导意义的纠错数据。解决方案的关键在于提出CoEvoke框架，通过施加受控扰动（controlled perturbations）制造可恢复的挑战，并结合受限的基于参考的引导（restricted reference-based guidance）来约束探索方向，从而在不依赖私有参考或额外标注的前提下，构建出包含真实交互证据、决策链与任务原生结果的高质量重建轨迹数据集（共4,567条可重构轨迹，覆盖4,130个软件工程与推理任务）。实验表明，在相同训练任务、决策记录数量和更新次数条件下，引入扰动增强的评判器训练使ACPBench-Hard成功率从35.38%提升至42.12%（+6.73点），而SWE验证器训练相较未训练的接口匹配参考获得7.60–9.40点增益；同时，工作流与后训练评估在未扰动输入下仍能保持任务求解拓扑结构，证明了该方法在不改变求解代理（solving agents）的情况下，通过设计合理的轨迹采集条件即可获取有效的纠正性监督信号，显著提升协作决策质量。
链接: https://arxiv.org/abs/2610.14021
作者: Lin Wu,Shu Yao,Yuheng Wang,Jingru Fan,Huatao Li,Yihan Li,Liu Yixuan,Cheng Liu,Jian Chang,Zeyu Zhao,Aohua Zhang,Huiying Zhao,Shuyuan Zhu,Chen Qian
机构: Shanghai Jiao Tong University (上海交通大学); China Unicom Online Information Technology CO.,Ltd (中国联通在线信息技术有限公司)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA); Software Engineering (cs.SE)
备注: 14 pages, 4 figures, 3 tables
Abstract:Training multi-agent collaboration requires trajectories of checking, feedback, and repair, yet task-solving rollouts may end in immediate approval or repeat unsuccessful attempts without useful correction. We introduce CoEvoke, a cross-domain framework that evokes these interactions during trajectory collection. Controlled perturbations create recoverable difficulties, while restricted reference-based guidance directs investigation. The resulting dataset contains 4,567 reconstructable trajectories over 4,130 software-engineering and reasoning tasks, linking public interactions to evidence, decisions, and task-native outcomes. Separate experiments over 2,287 same-task pairs show feedback increases of 47.72-100 percentage points across seven perturbation groups; guidance has smaller, model-dependent effects. With identical training tasks, decision-record counts, and updates, perturbation-enriched judge training improves ACPBench-Hard success from 35.38% to 42.12% (+6.73 points). SWE verifier training gains 7.60-9.40 points over interface-matched untrained references. Workflow and post-training evaluations preserve task-solving topology with unperturbed inputs and no private-reference guidance. These results support designing collection conditions to obtain corrective supervision that improves collaborative decisions with frozen solving agents.
[MA-52] LLM -Based Multi-Agent Collaboration for Constrained Multi-Objective Container Placement
【速读】：该论文旨在解决数据中心中容器部署（container placement）的复杂优化问题，核心挑战在于同时最小化能耗、最大化应用亲和性（affinity）偏好，并满足多资源容量与反亲和性（anti-affinity）等硬约束。传统方法依赖静态规则，缺乏对动态集群状态的适应能力且难以扩展；而元启发式方法虽具灵活性但计算开销大、收敛慢且易陷入局部最优。为此，本文提出一种基于大语言模型（LLM）驱动的多智能体协作框架，其关键创新在于通过四类专用智能体在闭环ReAct循环中协同决策：能耗代理（Power Consumption Agent）与亲和性代理（Affinity Agent）就冲突目标进行辩论，部署代理（Placement Agent）采用容差门控仲裁（Tolerance-Gated Arbitration, TGA）实现冲突消解，重构代理（Rearrangement Agent）则在部署后阶段利用单调改进规则（Monotone Improvement Rule, MIR）进一步优化决策。所有智能体基于环境提供的确定性可行候选表进行推理，确保动作始终满足硬约束。实验基于Google集群日志数据，在100个应用与25台机器配置下，相比能耗优先基线，所提框架降低8.36%能耗并提升39.30%亲和性；消融实验表明，多轮辩论机制、TGA及MIR均对性能有显著贡献，任一组件移除均导致能耗与亲和性双重下降。
链接: https://arxiv.org/abs/2610.07030
作者: Suvarthi Sarkar,Abhishek Hazra,Akshat Mittal,Aryabartta Sahu
机构: IIT Guwahati(印度理工学院古瓦哈蒂分校); NIT Agartala(印度国家技术学院阿加尔塔拉分校)
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Multiagent Systems (cs.MA)
备注: Accepted in IEEE Cloud 2026
Abstract:Container placement in data centers must simultaneously minimize power consumption and maximize affinity preferences, while satisfying multi-resource capacity and anti-affinity constraints. Traditional approaches typically rely on fixed rules, which lack adaptability to dynamic cluster states and are difficult to extend for adaptive decision-making. On the other hand, meta-heuristic methods, although more flexible, are often computationally expensive, slower and prone to getting trapped in local optima. In this work, we propose an adaptive and efficient approach based on an LLM-driven multi-agent collaboration framework, where four specialized agents operate in a closed-loop ReAct cycle at each placement step. A Power Consumption Agent and an Affinity Agent debate over competing objectives, while a Placement Agent resolves conflicts using Tolerance-Gated Arbitration (TGA). A Rearrangement Agent further refines decisions through the Monotone Improvement Rule (MIR) in a post-placement refinement phase. All the agents reason over deterministic, feasibility-filtered candidate tables provided by the environment, ensuring that every proposed action inherently satisfies hard constraints. Based on experimental evaluation using the Google Cluster Trace with a configuration of 100 applications and 25 machines, the proposed framework consumes 8.36% less power and achieves 39.30% higher affinity than the power-greedy baseline. Ablation studies confirm that each architectural component, the multi-round debate mechanism, the TGA in the Placement Agent, and the MIR in the Rearrangement Agent, contributes meaningfully to the overall performance, with removal of any single component degrading both power consumption and affinity satisfaction.
[MA-53] Collective Benefits and Efficiency in Coupled Echo State Networks
【速读】：该论文旨在解决集体智能中互动机制如何影响群体决策准确性和效率的问题，特别是探究在何种条件下成员间的交互能够超越独立响应的“群体智慧”并提升整体表现。其解决方案的关键在于引入基于回声状态网络（Echo State Network）的储层计算模型，通过耦合反馈机制模拟成员间的动态交互，揭示了交互过程中的协同效应。研究发现，采用池化反馈（pooled-feedback）进行交互的模型对能够重现巴赫拉米等人（Bahrami et al., 2010）报告的双人优势现象，而单纯的独立响应聚合则无法实现这一效果；进一步分析表明，交互有助于维持系统稳定性，避免输出因失控反馈（runaway feedback）而发散，从而提升准确性。此外，在效率方面，由较小规模但相互连接的模型组成的大群体，相较于少数大型独立模型，在同等总计算量下表现出更优性能。因此，该研究提出了一种形式化的理论框架，阐明了交互在特定条件下可增强集体智能的内在机制。
链接: https://arxiv.org/abs/2610.17506
作者: Polyphony Bruna,Rick Dale,Michael J. Spivey,Christopher T. Kello
机构: 未知
类目: Physics and Society (physics.soc-ph); Multiagent Systems (cs.MA); Neural and Evolutionary Computing (cs.NE); Neurons and Cognition (q-bio.NC)
备注: Experiment one presented at the 48th Annual Meeting of the Cognitive Science Society as a full paper talk in “Collective Intelligence.”
Abstract:Simple aggregation of independent responses (“wisdom of crowds”) generally improves collective performance through statistical facilitation, but interaction among members can sometimes hinder group outcomes. However, under certain circumstances, coordination studies show that interacting groups can outperform their individual members and nominal groups, and interaction can also be more efficient than independence. We present a reservoir computing model to investigate how interaction can enhance joint performance in terms of accuracy and efficiency. Echo state networks were trained independently to identify Japanese speakers from vowel recordings and tested jointly with and without coupled feedback. Model pairs reproduced the dyadic advantage reported by Bahrami et al. (2010) with interaction via pooled-feedback, but not when independent responses were aggregated together. Model dynamics revealed that interaction was beneficial whereas aggregated outputs were prone to diverging from runaway feedback over time. In terms of efficiency, larger groups of smaller interacting models outperformed smaller groups of larger models while equating for total amount of group computation. Our results offer a formal framework for explaining when and why interaction enhances collective intelligence.
自然语言处理
[NLP-0] ChunkAct: Action Chunking for Agent ic Robotics
【速读】： 该论文旨在解决代理型机器人（agentic robotics）中因大语言模型（LLM）频繁推理导致的效率瓶颈问题。在现有框架下，每个动作原语（primitive）都需要调用一次完整的LLM推理以回顾全部交互历史，使得任务执行成本随原语数量线性增长，严重制约了实际应用中的实时性与可扩展性。其核心解决方案是提出分块执行（ChunkAct）机制，关键在于将记忆中已知的动作序列预先编译为“动作块”（action chunk），在线执行时仅需一次LLM调用即可触发一整串链式原语的自主执行，直至失败才返回控制权。该方法实现了运行时内存免依赖，并将LLM计算资源集中于真正需要决策的高层选择，而非重复的底层动作规划。因此，LLM的计算复杂度由原语数量降为决策次数，显著提升效率。在LIBERO-Pro基准上，ChunkAct相较当前最优方案提升了5.0个任务成功率，同时减少2.9倍的LLM调用、4.4倍输入token及2.3倍输出token，充分验证了其高效性与实用性。
链接: https://arxiv.org/abs/2610.18874
作者: Bingxin Xu,Wenrui Bao,Xinxin Liu,Jiawei Zhou,Yuzhang Shang,Emilio Ferrara
机构: University of Southern California(南加州大学); University of Central Florida(中佛罗里达大学); Stony Brook University(石溪大学)
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Agentic robotics places a large language model (LLM) agent in the robot control loop, where it composes perception, motion, and vision-language-action (VLA) tools under the guidance of a memory and thereby generalizes to new tasks without retraining. This generality comes at a steep price in efficiency. Every primitive requires a round of LLM reasoning over the full interaction history, so the cost of an episode grows with the number of primitives. We find this reasoning largely redundant: once a sub-goal is chosen, its primitives follow a routine already prescribed by memory. We present ChunkAct, which introduces action chunking to agentic robotics. Online, a single call launches a chunk of chained primitives, which executes autonomously and returns control only upon failure. Offline, ChunkAct compiles the memory into these chunks, so execution is memory-free at test time and the planner is reserved for decisions that require reasoning. LLM computation therefore scales with the number of decisions rather than the number of primitives. On LIBERO-Pro, ChunkAct improves task success by 5.0 points over the state of the art in agentic robotics while requiring 2.9x fewer LLM calls, 4.4x fewer input tokens, and 2.3x fewer output tokens.
[NLP-1] v: Scaling Long-Horizon Agent ic Environments with Dense Progress Rewards
【速读】： 该论文旨在解决开源生成式智能体在长时程任务（long-horizon tasks）上表现不佳的核心问题，其根本挑战在于缺乏高质量的训练数据与可扩展的训练方法：长而复杂的任务难以有效构建、验证，且奖励信号稀疏，导致模型难以学习。为此，论文提出两个关键创新：一是构建首个具备可验证性、密集且渐进式奖励的长时程智能体训练数据集Tev-Gym，涵盖7个领域共1,170个经验证的任务环境，采用“隐世界优先”（latent-world-first）方法，先构建任务领域的完整隐藏规范（包括工作流与每一步后的预期状态），再逐步渲染为智能体可感知的环境，从而实现复杂度可控与自然进度度量；二是提出进度潜力奖励塑造（Progress-Potential Reward Shaping, PPRS）策略，将进度指标转化为每步密集奖励，显著提升训练效率并使原本不可训练的任务变为可训练。基于此方案训练出的Tev-27B模型在Agents’ Last Exam（ALE-CM）计算与数理科学领域达到与GPT-6 Astra和Claude Opus 5相当的性能（Pass@5），且仅使用ALE相关数据即展现出对Terminal-Bench 3.0/4.0的强泛化能力，超越GLM 5.2与Gemini 3.7 Flash，同时在中等时程任务（如TB 2.1）上达到接近Claude Opus 4.8的水平。
链接: https://arxiv.org/abs/2610.18868
作者: Wenbo Hu,Shraman Pramanick,Krishna Kumar Singh,Nanyun Peng,Kai-Wei Chang,Yong Jae Lee
机构: Adobe Research(Adobe 研究院); University of California, Los Angeles(加利福尼亚大学洛杉矶分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: this https URL
Abstract:Long-horizon agents have drawn growing attention for their ability to work autonomously – from hours to days – on economically valuable tasks, a capability essential to automated discovery and recursive self-improvement (RSI). Yet open-source models still struggle on long-horizon tasks, largely for lack of both training data and a working recipe, and scaling both is hard: long, complex tasks are difficult to curate and verify, and their rewards are sparse. We address both. First, we introduce Tev-Gym, the first long-horizon agent training dataset with verifiable, dense, progressive rewards – 1,170 verified task environments across 7 domains. It is built on a latent-world-first approach that first constructs a latent world: a complete and correct hidden specification of the task domain, workflow, and expected state after each step. Each task is then progressively rendered from this specification into an agent-facing environment, enabling controllable complexity and a natural progress metric. Second, we turn this metric into a training recipe: via progress-potential reward shaping (PPRS), we convert progress into a dense per-step reward that improves training efficiency and makes otherwise-intractable tasks trainable. Training with this recipe yields Tev-27B, which achieves substantial gains on the Agents’ Last Exam computing and math sciences domains (ALE-CM), matching GPT-6 Astra and Claude Opus 5 at Pass@5. Despite training only on ALE-related data, it generalizes to the Terminal-Bench (TB) 3.0 and 4.0 subsets, surpassing GLM 5.2 and Gemini 3.7 Flash, and reaches Claude Opus 4.8-level performance on middle-horizon tasks such as TB 2.1.
[NLP-2] DoubtLess: Training-Free Contrastive Decoding for Efficient Reasoning
【速读】： 该论文旨在解决生成式 AI (Generative AI) 在推理过程中因长思维链（Chain-of-Thought, CoT）导致的冗余计算与高显存消耗问题，即模型在求解任务时产生过长且不必要的推理路径，从而增加推理成本。现有方法如基于训练的压缩、添加固定控制向量或采用启发式截断策略虽可缩短思维链，但普遍存在需额外训练、影响准确率或破坏推理逻辑等缺陷。本文提出一种无需训练的解码方法 DoubtLess，其核心创新在于利用一个小型推理模型对大型模型进行引导：该小模型分别以“简洁”和“冗长”两种角色评估大模型当前的思维链，并将两者输出的对数概率向量之差（按强度缩放后）动态注入大模型的每一步解码过程，从而激励其生成更紧凑的推理路径。实验表明，在七项涵盖竞赛数学、研究生级科学及编程竞赛的基准测试中，DoubtLess 使 Qwen3-32B 的思维链长度平均缩短 31% 且保持原有准确率；在自调试编码代理中，每轮思考令牌数减少 20% 而任务性能不变。此外，该方法具有良好的泛化能力，适用于其他模型架构，并可与训练驱动的压缩技术协同进一步提升效率。消融分析显示，其主要节省来自正确答案首次出现后的重复验证行为，有效提升了推理效率。
链接: https://arxiv.org/abs/2610.18862
作者: Minghao Luo,Liang Chen
机构: The Chinese University of Hong Kong(香港中文大学); EPFL(洛桑联邦理工学院)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Reasoning models achieve strong performance through long chains of thought (CoT), but often spend more tokens than needed to solve a problem. Existing ways to shorten the CoT require additional training for each model, add a fixed steering vector to the model’s hidden states, or use decoding heuristics such as stopping early or banning words like “wait”, which can reduce accuracy. We propose DoubtLess, a training-free decoding method in which a much smaller reasoning model steers the large one towards brevity. The small model reads the large model’s CoT so far twice, once prompted to be brief (the brief thinker) and once prompted to be verbose (the verbose thinker), and the difference between its two logit vectors, scaled by a strength, is added to the large model’s logits at every step of the CoT. On seven benchmarks of competition mathematics, graduate-level science and competitive programming, DoubtLess shortens the CoT of Qwen3-32B by 31% with unchanged accuracy, outperforming five inference-time baselines. We also show that DoubtLess reduces thinking tokens per turn by 20% in a self-debugging coding agent with unchanged task performance. It also generalizes to two other model families with unchanged accuracy. When combined with training-based compression, it further shortens the CoT of a model already trained for brevity. Ablations and behavioral analysis show that DoubtLess reduces repeated checking, with most token savings coming after the correct answer first appears in the CoT. Our code is available at this https URL.
[NLP-3] Hybrid Reasoning Agents
【速读】： 该论文旨在解决当前大语言模型（LLM）智能体在推理与外部环境交互中难以兼顾高效内部思维与可解释性动作表达的矛盾问题。现有方法要么依赖文本进行全程推理（缺乏效率），要么采用隐式表示（如潜在空间）进行推理但丧失了对工具调用的原生支持和推理过程的透明性。其核心解决方案是提出混合推理智能体（Hybrid Reasoning Agents, HybridAgents），通过动态切换“潜在推理”与“文本交互”两种模式，在统一生成流程中实现二者无缝衔接。关键设计包括：1）轻量级混合链接（HybridLink），实现文本与潜在表示间的平滑转换；2）持续工作记忆机制，确保多轮交互中混合上下文无损保留。在此架构基础上，进一步引入专用后训练算法：HybridSFT 用于协调主干模型与 HybridLink 的协同行为，提升工具集成能力；HybridRL 则通过强化学习优化决策策略，明确何时使用潜在空间思考、何时以文本形式执行动作。实验表明，HybridAgent 在11个涵盖数学、科学、医学、搜索及自主推理的基准上，相较于前沿文本或潜在空间方法，平均性能提升10.4%，同时实现35.5%–54.7%的令牌消耗减少、1.8x–2.6x加速推理速度，并带来2.7x–3.7x的成本降低。
链接: https://arxiv.org/abs/2610.18861
作者: Jiaru Zou,Yunzhe Qi,Sirui Chen,Federico Bianchi,Shang Zhu,Pan Lu,Jingrui He,James Zou
机构: Stanford University (斯坦福大学); UIUC (伊利诺伊大学厄本那-香槟分校); Together AI; Meta
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 45 pages, 12 figures
Abstract:How can we build an agent that thinks beyond human language, yet still acts through it in the external environment? Modern LLM agents still reason and interact through text, while latent reasoning offers an efficient internal medium but lacks native tool access and reasoning transparency. We bridge these two directions with Hybrid Reasoning Agents (HybridAgents), which dynamically navigate between latent reasoning and text-based interaction throughout a unified generation process. Architecturally, a lightweight HybridLink enables seamless transitions between text and latent modes, while the agent’s continuous working memory preserves lossless hybrid context across multiple turns. These designs preserve tool-integrated agentic capabilities while making latent thoughts directly inspectable. Building on this architecture, we further develop dedicated post-training algorithms for hybrid reasoning: HybridSFT stabilizes tool-integrated hybrid generation by teaching the backbone to coordinate with HybridLink, then HybridRL scales reasoning by learning when to think in latent space and when to act through text. We evaluate HybridAgent at 3 model scales across 11 benchmarks covering mathematics, science, medicine, search, and agentic reasoning. Compared with frontier LLM Agents and agentic frameworks spanning both text-based and latent methods, HybridAgent consistently improves performance by 10.4%, while delivering 35.5%-54.7% token reduction, 1.8x-2.6x faster inference, and 2.7x-3.7x cost savings. Code and data are available at this https URL.
[NLP-4] he Last Remote Sensing Model Built by Humans? Casting Any Earth Observation Task as Recursive Self-Improvement
【速读】： 该论文旨在解决地球观测（Earth Observation, EO）模型依赖人工设计与调优的瓶颈问题，即当前EO任务仍需专家手动构建训练方案并基于基准数据进行反复调整，这一过程耗时且难以泛化。其核心解决方案是将模型构建流程自动化，引入一个通用编程代理（coding agent），通过定义“测试框架”（harness）和“验证器”（verifier）两个关键组件，使代理能够自主完成模型构建与迭代优化。其中，“测试框架”限定代理可用的资源与操作空间，“验证器”则明确成功标准，可通过标注基准、独立传感器数据、物理模型或同一区域的重复观测实现可验证性。在此机制下，代理可在仅使用少量标签（约1%）的情况下，在光学、合成孔径雷达（SAR）及多光谱时序影像等四类典型EO任务上超越现有技术水平，并持续提升性能。该方法将原本割裂的研究方向（如小样本学习、跨域适应等）统一为“测试框架”与“验证器”的配置选择，确立了以人类设计目标、代理自主建模为核心的新型地球观测范式，推动EO模型随代理能力进化而持续演进。
链接: https://arxiv.org/abs/2610.18845
作者: Sijie Zhao,Zeqi Zhou,Heli Qi,Xuanchao Fu,Zhuo Zheng,Junshi Xia,Junjue Wang,Weihao Xuan,Naoto Yokoya
机构: The University of Tokyo (东京大学); RIKEN (理化学研究所); Stanford University (斯坦福大学); Yokoya-Lab-UTokyo/RSI4EO
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
备注: Ongoing project
Abstract:Earth observation (EO) models are still built by hand. For each new task, experts design a training recipe and tune it against a benchmark. We show that this loop can be handed to an agent. Any verifiable task can be cast as a harness and a verifier, where the harness defines what the agent may use and the verifier defines what counts as success. Every EO task can be made verifiable, through labeled benchmarks, independent sensors, physical models or the repeated observation of the same place. Given only these two components, a general-purpose coding agent builds the EO model and improves it recursively on its own outputs. Settings long treated as separate research problems, from learning with a handful of labels to adapting across domains, become choices of harness and verifier. On four EO benchmarks spanning optical, SAR and multispectral time-series imagery, agent-built models surpass the state of the art and keep most of that accuracy from about 1% of the training labels. This establishes a new paradigm for EO, in which people design harnesses and verifiers and agents build the models, so EO models improve as agents do.
[NLP-5] OmniCompass: A Holistic Benchmark for Omni-Modal Video Grounding and Understanding
【速读】： 该论文旨在解决多模态大语言模型（Multimodal Large Language Models, MLLMs）在视频理解能力评估中缺乏全面性与系统性的问题。现有基准测试仅覆盖部分评估维度，难以实现对全模态视频理解能力的深入诊断。为此，论文提出OmniCompass——一个面向全模态视频定位与理解的综合性评估基准。其核心解决方案在于构建双轨评估体系：OmniCompass-VL针对视觉-语言模型，评估需依赖视觉证据的任务；OmniCompass-AV则面向全模态模型，评估需联合音视频证据的任务。每条赛道均包含时序定位（Temporal Grounding, TG）、空间定位（Spatial Grounding, SG）和问答（Question Answering, QA）三类互补任务，分别考察目标出现的时间、位置及事件内容。QA任务进一步细分为感知与推理两类，其中OmniCompass-VL涵盖5种感知和3种推理类别，OmniCompass-AV包含3种感知和2种推理类别。数据集共包含820段视频（短、中、长三种时长）及2,460个实例，其中音频丰富且叙事性强的子集构成OmniCompass-AV，含240段视频与720个实例。所有实例均采用文本仅输入与文本+图像双条件设置进行评估，以检验模型在不同引导方式下的表现一致性。所有标注均经多轮人工质量验证。对50个MLLMs的评估结果表明，开源模型在整体性能上仍显著落后于闭源模型，尤其是在音视频联合理解方面；即便领先模型在时序与空间定位任务中仍存在显著挑战，且在不同提示条件下预测结果存在不一致现象。
链接: https://arxiv.org/abs/2610.18829
作者: Xuyang Liu,Shihao Wang,Xinyu Wei,Minghan Li,Zeqing Wang,Zuhao Yang,Zichen Wen,Lei Zhang
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL); Multimedia (cs.MM)
备注: Homepage: this https URL
Abstract:With the rapid development of multimodal large language models (MLLMs), evaluating their video understanding capabilities has become increasingly important. Existing video understanding benchmarks cover only part of the evaluation dimensions, limiting systematic diagnosis of omni-modal video understanding. We introduce OmniCompass, a holistic benchmark for omni-modal video grounding and understanding. OmniCompass defines two evaluation tracks: OmniCompass-VL assesses vision-language models on tasks requiring visual evidence, while OmniCompass-AV assesses omni-modal models on tasks requiring joint audio-visual evidence. Each track includes three complementary tasks: temporal grounding (TG), spatial grounding (SG), and question answering (QA), assessing when a referenced target appears, where it is, and what happens. QA covers five perception and three reasoning categories in OmniCompass-VL, and three perception and two reasoning categories in OmniCompass-AV. OmniCompass-VL contains 820 videos and 2,460 instances across domains and three duration types: short, medium, and long. A narrative- and audio-rich subset forms OmniCompass-AV with 240 videos and 720 instances. Each instance is evaluated under two paired settings: text-only and text+image, specifying the same target through text and visual-language guidance. All instances are manually annotated with multi-round quality verification. Evaluations of 50 MLLMs show that open-source models still lag behind proprietary counterparts, particularly in audio-visual understanding. Precise temporal and spatial grounding remains challenging even for leading models, and their predictions still exhibit inconsistencies across the paired target specifications.
[NLP-6] Read the Room Before Your Model Generates AACL
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在生成文本时，采样温度（sampling temperature）设置缺乏情境可解释性的问题。现有方法多依赖固定或启发式全局设定，或基于上下文/词元级别的校准，难以体现对话场景的几何可解释结构。为此，本文提出一种简洁且可解释的温度选择框架——Scenario-Actor，其核心在于通过从Transformer隐藏状态中提取与场景相关的结构信息，构建场景嵌入（scenario embedding），即对交际情境的结构化几何表征，并据此动态确定采样温度。该方案的关键在于将温度调节过程与对沟通场景的语义-几何理解相结合，使解码策略能够自适应地反映情境特征。实验结果表明，在Llama-3.1-8B-Instruct模型上，Scenario-Actor在交互式社交场景中实现了优异的目标完成率，在开放式指令遵循任务中取得显著提升，并在多轮对话任务中表现与多个基线相当。
链接: https://arxiv.org/abs/2610.18820
作者: Muyao Kong,Zhihao Lin
机构: New York University (纽约大学); Independent Researcher (独立研究员)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 17 pages, 7 figure, AACL-IJCNLP 2026 Main Conference
Abstract:Large language models (LLMs) generally rely on sampling temperature to regulate the amount of randomness in their outputs. Mainstream methods for selecting temperature typically involve fixed, heuristic global settings and context or token-level calibration, rather than a geometrically interpretable representation of the communicative scenario. In this work, we introduce a simple and interpretable temperature selection framework, the \textbfScenario-Actor, which adapts decoding to the inferred situational characteristics. Scenario-Actor extracts scenario-relevant structures from Transformer hidden states and determines the temperature from scenario embedding, a structured geometric representation of the communicative setting. Empirical results show that on Llama-3.1-8B-Instruct, Scenario-Actor achieves strong goal-completion performance in interactive social scenarios, significant gains on open-ended instruction following, and competitive performance on multi-turn conversation compared to multiple baselines.
[NLP-7] Back-Translation Still Matters: Leverag ing Monolingual Data for Low-Resource Machine Translation with LLM s
【速读】： 该论文旨在解决低资源语言机器翻译中因平行语料稀缺而导致性能受限的问题，尤其关注在仅有少量平行数据（如1K对）但存在中等规模单语数据的情况下，如何更有效地利用这些额外资源以提升翻译质量。其核心挑战在于：传统方法如持续预训练（continued pretraining）对低资源语言的增益有限，而直接使用少量平行数据进行微调又难以充分挖掘模型潜力。解决方案的关键在于引入源自预大语言模型（LLM）时代的回译（back-translation, BT）技术——即利用可获得的单语数据，通过已微调的模型生成目标语言的伪平行句对，再以此反向训练模型，从而实现对源语言到目标语言翻译能力的增强。该方法通过在平行数据上交替更新，确保学习过程稳定；同时结合从真实平行数据中检索的上下文示例进一步优化结果。实验表明，在奥迪亚语和斯瓦希里语上，该方法比持续预训练提升7.7至16.4个XCOMET-XXL分数，在部分情况下甚至超越使用十倍更多平行数据的模型。跨五种低资源语言的评估显示，相较于仅依赖平行数据微调，回译使翻译性能提升16%至173%（以XCOMET-XXL衡量），且人工盲评结果与自动指标一致，验证了其有效性。
链接: https://arxiv.org/abs/2610.18818
作者: Divyanshu Bathla,Parag Singla,Jitin Singla
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Pre-trained Large Language Model (LLMs) are effective on many tasks in zero-shot and few-shot settings. For machine translation of low-resource languages, the obvious options are few-shot prompting and fine-tuning on a small parallel set. We ask: what if moderate amounts of low-resource monolingual data are also available? We argue that, rather than continued pretraining, a more effective use of this data is the pre-LLM era technique of back-translation (BT). The current fine-tuned LLM translates monolingual sentences in each language, and the resulting pairs train it in the reverse direction. Updates on the parallel set are interleaved for stable learning. With 1K parallel pairs on Odia and Swahili, BT beats continued pretraining into the low-resource language by 7.7 to 16.4 XCOMET-XXL points in domain. With beam-decoded intermediate translations, it also beats a model with 10 time more parallel data. In-context examples retrieved from the parallel data further improve BT. We also analyze the source, quality and amount of monolingual text. Across five low-resource languages, with Hindi as a high-resource control, and two 3B backbones, BT improves translation into the low-resource language by 16 to 173% XCOMET-XXL over parallel-only fine-tuning. Blind human evaluation on Odia and Sanskrit agrees with the automatic metrics.
[NLP-8] Self-Evolving Agent Harness
【速读】： 该论文旨在解决生成式 AI（Generative AI）系统在长期运行中因环境变化、任务演化及评估标准迭代而导致的可靠性下降问题，核心挑战在于如何在不更新模型权重的前提下，实现智能体（agent）软件架构（包括提示词、记忆机制、工作流与可执行组件）的持续自进化。其解决方案的关键在于提出一种基于反馈驱动的自进化框架——Harness RSI（Harness Self-Evolution），通过将更新目标与提议、接纳机制分离，明确区分对解释状态的修改与对工具、代码、仓库级控制逻辑及插件组合的重构。该框架采用进化优化策略进行搜索，并借助持续 PAC 学习（continual PAC-learning）分析其可靠性，建立了在特定假设下的期望泛化界、依赖历史回溯预算的保留界，以及基于独立验证数据的高概率接纳规则。此外，研究强调了持久性架构修订对知识共享与组织工作流的影响，并指出可靠部署需伴随任务分布和评估者动态变化的持续验证机制，同时对学习过程、运行时权限和关键操作设置独立保障。
链接: https://arxiv.org/abs/2610.18813
作者: Chuyang Wei,Haoxiang Guan,Minghang Zhu,Kexin Zhou,Jiyan He,Xiren Zhou,Huanhuan Chen,Yifei Shen
机构: Zhongguancun Academy; University of Science and Technology of China; Zhongguancun Institute of Artificial Intelligence
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 18 pages, 3 figures
Abstract:Harness self-evolution revises an agent’s software scaffolding, including prompts, memory, workflows, and executable components, while keeping model weights fixed. We formalize a feedback-driven process in which accepted revisions persist across runs. The framework separates update targets from proposal and admission mechanisms. It distinguishes changes to interpreted state from revisions to tools and code, repository-level control logic, and plugin composition. We characterize search through evolutionary optimization and examine its reliability through continual PAC-learning. Under explicit assumptions, the analysis yields an expected generalization bound, a retention bound conditional on historical regression budgets, and a high-probability admission rule using fresh, independent validation data. We examine how persistent harness revisions affect knowledge sharing and organizational workflows, with implications for scientific reliability, physical safety, and institutional accountability. Cases from prior work illustrate model–harness co-design, distributed execution, and plugin reuse. Reliable deployment requires continual validation as task distributions and evaluators change, with independent safeguards for learning, runtime privileges, and consequential actions. We provide a comprehensive 66-page discussion of related work, theory, algorithms, evaluation, applications, and the broader impact of Harness RSI, together with continuously updated code and supplementary materials, at this https URL.
[NLP-9] SkillScaffold: Scaffolding Agent ic Reinforcement Learning with Policy-Adaptive Skill Credit Assignment and Selective Injection
【速读】： 该论文旨在解决生成式智能体在强化学习（Reinforcement Learning, RL）中因技能注入（skill injection）的盲目性与评估成本高昂所导致的效率低下和性能不稳定问题。现有方法通常假设技能注入必然有益，或依赖高开销的滚动（rollout）评估来间接判断技能效用，而这种评估在采样有限时既不经济也不稳定。针对此问题，论文提出一种自适应策略训练框架SkillScaffold，其核心创新在于通过重用已验证结果的训练轨迹，基于结果感知的似然比较实现低成本、高稳定的技能效用估计，避免了对每个候选技能进行独立环境滚动的开销。该效用估计结果指导是否以及选择何种技能进行注入，在无有效技能时则保持无技能执行，从而实现动态决策。同时，生成器与执行器联合训练机制使生成的技能能随策略演进而自适应优化，促进技能内化以支持无技能推理。实验在ALFWorld、WebShop、ScienceWorld及τ²-Bench等多个基准上验证了该方法在短/长序列任务中的优越性，消融研究进一步证明了基于效用的选择与主动放弃策略的重要性。
链接: https://arxiv.org/abs/2610.18807
作者: Wenlin Zhong,Hao Wang,Kun Yuan,Fan Yang,Jinluan Yang,Dian Jin,Xu Tan,Yiquan Wu,Kun Kuang
机构: Zhejiang University(浙江大学); Kuaishou Technology(快手科技); University of the Chinese Academy of Sciences(中国科学院大学); Zhejiang University of Science and Technology(浙江科技学院)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Agentic reinforcement learning (RL) enables large language models (LLMs) to learn from multi-turn interactions with verifiable outcomes. Textual skills can improve exploration by distilling prior interactions into reusable guidance. However, existing skill-based RL approaches often either assume skill injection is beneficial or rely on indirect feedback or rollout-intensive evaluation to assess its utility. Our pilot studies show that rollout-based utility estimation for skill credit assignment is costly and unstable under limited sampling, while skill benefits vary across candidates and depend on the current policy’s competence on each query. To address these challenges, we introduce SkillScaffold, a policy-adaptive training scaffold that combines low-cost skill-utility estimation with selective skill injection. By reusing outcome-verified training trajectories, SkillScaffold generates candidate skills and estimates their utility through outcome-aware likelihood comparisons, without candidate-specific environment rollouts. The resulting utility guides whether and which skill to inject, retaining skill-free execution when no candidate is estimated to help. Joint training enables the Generator to produce skills tailored to the evolving Executor. Scaffolded RL encourages skill internalization for skill-free inference. Experiments on ALFWorld, WebShop, ScienceWorld, and \tau^2 -Bench demonstrate gains over strong agentic-RL and skill-based RL baselines across both short- and long-horizon tasks. Ablations and further analyses support the importance of utility-based selection and abstention. Our code is available at this https URL.
[NLP-10] SEAR: Self-Evolving Agent ic Robotics
【速读】： 该论文旨在解决当前前沿语言模型驱动的机器人智能体在自主学习与能力迁移方面存在的核心问题：如何区分智能体是否真正实现了持续性学习，而非仅通过大量试错来偶然成功；以及其习得的能力是否具备跨任务、跨仿真环境乃至真实机器人平台的泛化能力。其解决方案的关键在于提出SEAR（Self-Evolving Agentic Robotics）框架与基准测试体系，通过三种运行模式实现对智能体行为的系统性评估——即“代码即策略”（code-as-policy）、“智能体即策略”（agent-as-policy）和“自动研究者”（auto-research）模式，分别对应代码生成、实时控制与视觉-语言-动作模型的自适应微调。同时，基准测试引入多重控制变量，有效分离了经验留存与重复尝试、任务熟悉度、模拟器状态特权、预算规模及运气等干扰因素。实验结果表明：1）智能体的学习效率受总时间预算与任务难度相对于模型能力的相对关系影响显著；2）通用控制代码与共享函数具有良好的跨任务与跨仿真迁移能力，而任务特异性解法则不具备泛化性；3）不同语言模型在不同演化路径上表现各异，部分模型更依赖实时行动，另一些则更适合先生成代码；此外，纯文本模型的表现可媲美多模态模型，凸显了代理式推理与工具使用的重要性，而非单纯依赖感知能力。
链接: https://arxiv.org/abs/2610.18792
作者: Xinyi Wang,Bangzheng Li,Hanyu Wang,Lin Zhang,Andrea Yaoyun Cui,Zhenggang Tang,Shyamal Buch,Dejia Xu
机构: Luma AI
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 16 pages
Abstract:Frontier language-model agents can now improve robots on their own: they write control code, test it, and keep what works as skills and memories for the next task. We introduce SEAR (Self-Evolving Agentic Robotics), a framework and benchmark built around two questions. Is the agent really learning, or just trying more often? And does what it learns still help on new tasks, in a new simulator, and on a real robot? The framework runs an agent in three modes: it writes code that becomes the policy (code-as-policy), acts live in the control loop (agent-as-policy), or works as a researcher that fine-tunes a vision-language-action model (auto-research). The benchmark adds controls that separate retained experience from repeated attempts, familiar tasks, privileged simulator state, larger budgets, and luck. Across seven frontier language models, 330 simulated and five physical tasks, and over four years of agent runtime, we highlight three findings. First, the learning curve of a robotic agent depends both on the overall time budget and the task difficulty relative to model capability. Harder tasks make longer sessions worthwhile; on easier tasks, re-rolling short attempts does just as well. Second, control code and shared routines transfer across tasks and simulators, while task-specific solutions do not. On a real robot, we observe a performance gap between agent-designed practice and novel human-designed scenes. Third, different models evolve more effectively in different ways. Some do best acting live, others writing code first. And a text-only model keeps pace with a comparable multimodal one: perception helps with error assessment, but agentic reasoning and tool use can matter more. We release the benchmark and codebase to support future research.
[NLP-11] Where Do LLM Ensemble Gains Come From? Decomposing Candidate Value Diversity and Aggregation
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）集成方法中评估增益时因基准对照选择不同而导致结果偏差的问题。现有研究通常以单一基线作为参照来衡量集成带来的性能提升，但这种评估方式忽略了不同对照条件对增益归因的影响。论文通过在四种匹配条件下对比分析：直接输入（Direct）、单槽位（Single-slot）、重复候选（Duplicate）与多候选（Multi-candidate），揭示了整体候选增益（Candidate Improvement Value, CIV）可代数分解为额外候选对比（Divergence from Single-Variant, DSV）和候选-上下文对比（Candidate-Context Variance, CCV）两部分；其中，尽管CIV整体值不变，但其分解结构依赖于所选控制条件。在多跳检索增强生成（MultiHopRAG, k=2）任务中，重复间隙（Repetition Gap, Rep）达-8.33个百分点（95% CI: [-13.25, -3.75]），表明以重复候选为基准时测得的增益显著低于以单候选为基准的结果。在更强面板扩展的FRAMES实验中（n=200, k=3, 输出限制4096），配对差值ΔRep为+9.67个百分点（[+6.00, +13.50]），且在面板内重复间隙符号反转（原为-6.33，强面板下变为+3.33），进一步凸显控制条件对增益解释的关键影响。研究发现，在证据聚合场景中，无论采用何种非多样性控制，候选-上下文对比始终呈强烈负向。因此，论文强调：在进行LLM集成评估时，必须明确定义参考控制条件，否则对额外候选贡献的归因将不可靠。
链接: https://arxiv.org/abs/2610.18786
作者: Donghwan Kim
机构: Independent Researcher(独立研究员)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 14 pages, 6 figures
Abstract:LLM ensemble gains are often measured against a single baseline, but the value attributed to additional candidates can change with the comparison used. We study an LLM aggregator under four matched conditions: Direct (no candidates), Single-slot (one candidate in a multi-slot interface), Duplicate (one candidate repeated), and Multi-candidate (separately generated candidates). Under either non-diverse control, overall candidate gain (CIV) splits algebraically into an additional-candidate contrast (DSV) and a candidate-context contrast (CCV); the split, but not CIV, depends on the chosen control. On MultiHopRAG model diversity (k=2), the repetition gap Rep=-8.33 pp [-13.25, -3.75] quantifies the difference between gains measured against Duplicate and Single-slot. On a matched FRAMES stronger-panel extension (n=200, k=3, aggregator output limit 4096), paired Delta Rep=+9.67 pp [+6.00, +13.50]; within panels, Rep changes sign (original -6.33; stronger +3.33). In FRAMES evidence aggregation, candidate-context contrasts remain strongly negative under both non-diverse controls. Our results show why ensemble evaluations should state the reference control before attributing gains to additional candidates.
[NLP-12] Pack Light CarryOn: Building Object-Oriented Context for Long-Horizon Agents
【速读】： 该论文旨在解决长时序智能体（long-horizon agents）在持续任务中因反复处理交互历史而导致的计算冗余问题。现有方法通常依赖压缩后的历史记录或对进展的重构来维持上下文连续性，但这种方式效率低下且难以扩展。其核心挑战在于：如何在保留任务进展可复用性的前提下，避免因历史状态被丢弃而重新构建完整上下文表示。为此，论文提出 CarryOn，一种面向对象的上下文运行时机制，将延续性（continuation）组织于可复用的任务对象（task objects）之上，而非依赖历史轨迹本身。其关键创新在于引入“承载者”（carrier）记录机制，通过维护跨执行轮次的载体状态，在任务状态折叠（fold）后仅对受影响的记录基于绑定证据与环境约束进行增量更新，同时保留未受影响的记录并协同更新执行上下文。该方法显著提升了上下文管理效率，实验表明在 OSWorld 2.0 和 BrowseComp-Plus 上分别实现 5.8% 和 7.0% 的性能提升，同时分别减少 51.9% 和 22.9% 的总令牌消耗。
链接: https://arxiv.org/abs/2610.18784
作者: Yuchen Lu,Julian McAuley,Lei Zhang
机构: University of California San Diego(加州大学圣地亚哥分校)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 24 pages. Project page: this https URL
Abstract:Long-horizon agents accumulate progress in persistent task state, yet repeatedly process interaction histories describing how that progress was achieved. This raises a natural question: can long-horizon context instead be organized around reusable progress rather than the history that produced it? In this paper, we introduce CarryOn, an object-oriented context runtime that organizes continuation around reusable task objects. Rather than treating compressed histories or reconstructed accounts of progress as the primary basis for continuation, CarryOn builds a context view from continuation carriers, complementary execution context, and a live tail of recent interactions. A core challenge is keeping this reusable basis valid after its surrounding history is retired: how can CarryOn update the basis without reconstructing the full representation at every fold? CarryOn maintains carrier records across folds, revising affected records against bound evidence and applicable environment checks while carrying forward unaffected records and jointly updating execution context. Experiments on OSWorld 2.0 and BrowseComp-Plus show improved gains of 5.8% and 7.0% while reducing total token usage by 51.9% and 22.9%, respectively. Project page: this https URL
[NLP-13] BLADE: Auditing Information Use for Failure Attribution in LLM Multi-Agent Systems
【速读】： 该论文旨在解决大型语言模型多智能体系统（Large Language Model Multi-Agent Systems）在协作过程中失败诊断困难的问题。由于各智能体间存在角色、工具与工作流的分工协作，单一智能体对正确结果的误判可能导致错误被后续智能体持续传播，从而难以定位根本原因。现有方法虽能追踪信息依赖关系，但仅知信息来源并不等同于识别其使用过程中的偏差。为此，论文提出无需训练的BLADE（Blame Localization via Audited Dependency Evidence）框架，其核心在于超越单纯的信息溯源，转而对信息使用过程进行审计。该框架基于结构化概率判断（Jev）机制，利用依赖概率筛选关键源-决策交接点，并在这些节点上检验接收方是否忠实且符合任务要求地使用了输入信息，进而区分错误引入、传播与修复三类行为。最终通过全局视角整合所有执行步骤的审计结果，实现精准的责任归属。实验表明，在无参考答案条件下，BLADE在五个WhoWhen及WhoWhen Pro子集共1,228次失败运行中，相较于八种策略均达到最优或并列最优的步骤级准确率，显著优于All-at-once方法2.2–11.1个百分点。显式呈现责任证据为构建更具可解释性与可靠性的智能体系统提供了可行路径。
链接: https://arxiv.org/abs/2610.18780
作者: Xiaoguang Guo,Yaohao Long,Chuxu Zhang
机构: University of Connecticut (康涅狄格大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 16 pages, 6 figures, 3 tables
Abstract:Large language model multi-agent systems extend individual agents’ capabilities through collaboration across specialized roles, tools, and workflows. Yet collaboration makes failures harder to diagnose: one agent may misinterpret a correct result while others carry the mistake forward. Failure attribution therefore seeks to identify the responsible agent and decisive error step. Recent methods trace dependencies, but knowing where information came from is not the same as knowing where it went wrong. We introduce BLADE (Blame Localization via Audited Dependency Evidence), a training-free framework that moves beyond tracing information to auditing its use. Using Jev for structured probabilistic judgments, BLADE uses dependency probabilities to select relevant source-decision handoffs for auditing. At these handoffs, it examines whether receiving decisions use their sources faithfully and within task requirements, distinguishing error introduction, propagation, and repair. These findings inform a global diagnosis that considers all execution steps. Without reference answers, BLADE achieves the best or tied-best step accuracy among eight strategies under a shared Jev decision interface on 1,228 failed runs across five WhoWhen and WhoWhen Pro subsets, outperforming All-at-once by 2.2-11.1 percentage points. Making responsibility evidence explicit offers a path toward more interpretable and reliable agentic systems.
[NLP-14] Rubrics from the Task Verdicts from Evidence: Pointwise Rubric Rewards with Tool-Augmented Judges
【速读】： 该论文旨在解决现有基于评分量表的奖励模型（Rubric-based Reward Models, RMs）中存在的响应诱导偏差（response-induced bias）问题，以及人工评判者在仅依赖阅读时难以应对机械性约束、且常用成对评判仅提供相对偏好而难以直接用于策略优化的局限性。其解决方案的关键在于提出一种点对点式（pointwise）的评分量表奖励模型ReVeR，该模型通过仅基于指令本身进行推理来生成硬性规则与分级原则，并借助工具增强的评判机制独立评估响应。两个核心组件均通过标签无偏蒸馏（label-blind distillation）训练，其中偏好标签仅用于筛选教师模型的输出，从而构建出大规模的“评分量表推理与评判轨迹”数据集ReVeR-Traces。实验表明，ReVeR在平均偏好准确率、测试时缩放能力及下游策略优化方面均优于成对式评分量表奖励模型，尤其在指令遵循任务上表现显著；同时研究揭示了偏好准确率、评分量表保真度与下游效用是相互独立的属性，为未来奖励模型研究提供了重要启示。
链接: https://arxiv.org/abs/2610.18766
作者: Tianci Liu,Haoyu Wang
机构: University of Tennessee(田纳西大学); University at Albany(阿尔巴尼大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Rubric-based reward models (RMs) evaluate responses against explicit criteria. However, existing methods often derive criteria from observed response candidates, introducing response-induced bias that may limit rubric generalizability. Their judges also struggle with mechanical constraints when relying on reading alone, and commonly used pairwise judges provide only relative preferences, complicating their direct use for policy optimization. We propose ReVeR, a pointwise rubric RM that derives hard rules and graded principles by reasoning about the instruction alone and evaluates responses independently through tool-augmented judging. Both components are trained through label-blind distillation, in which preference labels only filter a teacher’s outputs, yielding ReVeR-Traces, a large-scale dataset of rubric-reasoning and judging trajectories. Experiments demonstrate that ReVeR outperforms pairwise rubric RMs in average preference accuracy, most clearly on instruction following, as well as in test-time scaling and downstream policy optimization. We further find that generator RL improves preference accuracy while reducing rubric fidelity, and that higher preference accuracy does not necessarily translate into better downstream policies. These findings highlight that rubric fidelity, preference accuracy, and downstream utility are distinct properties, shedding light on future RM research.
[NLP-15] Looking Complete Is Not Being Grounded: Evaluating LLM Sitreps Against Human-Authored Situation Reports AACL
【速读】： 该论文旨在解决自动危机报告评估中“评估标准应基于实践者真实需求，而非仅便于比较”的核心问题。其解决方案的关键在于构建一个由专业人道主义情境报告通过定性内容分析提炼出的15个信息槽（information schema），并采用大语言模型（LLM）辅助、作者审定的内容分析方法，对121份人工撰写和261份LLM生成的报告在五个灾害事件及三种灾害类型下进行系统评估。研究发现，在气旋伊代（Cyclone Idai）场景中，谷歌Gemini-3.7-flash生成的报告虽名义上略超人类报告表现（0.827 vs. 0.821），但差异在统计上不显著（95%置信区间[-0.049, 0.059]；置换检验p=0.852），且该结果在OpenAI GPT-5.6-Luna模型下无法复现。盲评结果显示作者与Gemini判别器一致性达0.910准确率和0.799的无加权Cohen’s κ，四名裁判均满足原始一致性阈值0.7，唯独Gemini在调整随机一致性的κ值上达标。跨三大主事件分析表明，机器报告仅能再现5.4%-11.0%的当日官方数据，而精确数值匹配中，官方数据与社交流关联率为14.3%-19.6%，机器报告则高达85.4%-90.1%；人工审计确认45/50的缺失声明正确，仅4/25官方匹配与24/25机器匹配具有语义一致性。此外，所有12个事件-裁判组合均存在“发布来源缺口”（issuing-provenance gap），即报告日期、组织、系列与权威信息在专业报告中完全缺失。最终，传统指标如ROUGE-L无效，BERTScore在多重校正后与完整性呈负相关，通用LLM判别器甚至出现方向反转，凸显现有评估范式局限性。
链接: https://arxiv.org/abs/2610.18762
作者: Sachin Kumar
机构: LexisNexis(雷克斯尼斯)
类目: Computation and Language (cs.CL); Computers and Society (cs.CY); Social and Information Networks (cs.SI)
备注: Accepted at LT4CPR Workshop @ AACL-IJCNLP 2026
Abstract:Evaluation for automatic crisis reporting should measure what practitioners need, not what is convenient to compare. Through LLM-assisted, author-approved content analysis, we derive a 15-slot information schema from professional humanitarian situation reports using qualitative content analysis, then evaluate 121 human and 261 LLM-generated reports across five disasters and three hazard types. Under google/gemini-3.7-flash on Cyclone Idai, the best machine arm is nominally above the empirical human ceiling (0.827 vs. 0.821), but the difference is statistically indistinguishable from zero (95% bootstrap CI [-.049,.059] ; permutation p=.852 ), and the ordering does not replicate under openai/gpt-5.6-luna. On a blind 100-item Idai sample, the author agrees with the Gemini judge at 0.910 accuracy and unweighted Cohen’s \kappa=.799 . All four judges clear the plan’s raw-agreement criterion of .7; only Gemini reaches .7 on chance-corrected \kappa . Across the three primary events, machine reports reproduce 5.4-11.0% of same-day official figures. Across those same events, raw exact-number matching links 14.3-19.6% of official figures and 85.4-90.1% of machine figures to the social stream; manual audit confirms 45/50 declared absences, 4/25 official matches, and 24/25 machine matches semantically. The issuing-provenance gap is positive in all 12 observed event-judge cells: report date, organisation, series, and authority were not supplied at professional-report coverage in any cell. On the three primary events, ROUGE-L is uninformative, BERTScore is negatively associated with completeness after multiplicity correction, and a generic LLM judge changes sign.
[NLP-16] PreciseCoder: Faithful and Generalizable Debugging Beyond Unit-Test Rewards
【速读】： 该论文旨在解决基于单元测试（unit test）的强化学习（Reinforcement Learning, RL）在代码生成模型后训练阶段中存在的根本性缺陷：仅依赖通过/失败的二元奖励信号会导致模型倾向于过度修改代码并产生虚假错误（即“幻觉”错误），从而降低代码修复的精确性。其解决方案的关键在于提出一种名为PreciseCoder的新训练算法，该算法通过引入语义验证且局部化的编辑信号（semantically verified, localized edits）来增强原有的单元测试奖励，从而引导模型关注更小、更精准的代码修改。实验表明，在4B、9B和27B规模的模型上，PreciseCoder可将平均编辑行数从7–15行显著降低至3行，同时保持与纯单元测试训练相当的通过率；此外，该方法在单行错误修复训练下具备良好的泛化能力，适用于仓库级程序修复，并能通过推理轨迹（reasoning trace）实现跨模型精度迁移——即使未经过额外训练，前沿大模型如GPT-5.6 Sol和Qwen3.8-Max也能获得4.6–12.4分的编辑精度提升。语言学分析进一步揭示，该奖励机制促使模型聚焦于逐行推理、准确识别问题代码行并形成一致的错误假设，从而显著提升了修复过程的可解释性与准确性。
链接: https://arxiv.org/abs/2610.18755
作者: Yunkai Zhan,Miaosen Chai,Shangshang Wang,Jike Zhong,Yuqing Yang,Deqing Fu,Serina Chang,Wang Bill Zhu
机构: University of Chicago (芝加哥大学); University of Southern California (南加州大学); Google(谷歌); University of California, Berkeley (加州大学伯克利分校)
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 29 pages, 14 figures
Abstract:Reinforcement learning (RL) from unit tests is the common recipe for post-training code models, but a unit test only asks whether the final program passes. We analyze what this unit-test-only reward teaches a model during debugging and find that, while it raises pass rates, it also teaches the model to edit more and to hallucinate bugs in code that is already correct. We introduce PreciseCoder, a training algorithm that augments the unit-test reward with a signal for semantically verified, localized edits, which correlates with smaller edits and higher pass rates. On 4B, 9B, and 27B models, PreciseCoder reduces the average number of edit lines from 7-15to 3 at a pass rate on par with unit-test-only training. Trained only on single-line bugs, PreciseCoder can generalize to repository-level program repair, where it resolves more issues with smaller patches. Its precision also transfers from weaker to stronger models: given the reasoning trace of a 9B PreciseCoder model, frontier models such as GPT-5.6 Sol and Qwen3.8-Max gain 4.6-12.4 points in edit precision with no additional training. Linguistic analysis of the thinking traces shows that our reward encourages the model to reason about individual lines, name the buggy ones, and form consistent bug hypotheses.
[NLP-17] On the Monitorability of Natural Language Autoencoders: Identification vs Discrimination
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在安全监控中缺乏可解释性与可信监督机制的问题，具体聚焦于如何通过自然语言自编码器（Natural Language Autoencoders, NLAs）实现对模型内部激活向量的可理解性监督。其核心挑战在于：人类监督者能否仅依据激活向量的自然语言解释（由激活表述器，Activation Verbaliser, AV生成）准确区分真实描述与精心构造的虚假描述。论文的关键创新在于将NLA的训练与监督部署分别建模为两种不同类型的涌现通信（Emergent Communication）中的参照游戏：训练阶段为生成式参照游戏，强调从向量生成合理解释；而监督阶段则为判别式、多模态、部分可观测的参照游戏，涉及激活向量与对应陈述之间的多对多关系，且要求区分真实与近似虚假的解释。理论分析指出，由于训练过程未强制要求解释排除错误选项，导致AV生成的解释虽能识别其所描述的激活，却难以有效区分与其高度相似的虚假解释。通过引入自编码器自身重构器（AR）作为最友好的代理监督模型进行实验验证，在一个公开检查点上发现，真实解释在全部320次比较中均优于其他激活的解释，但在面对仅在一个主张上存在差异的流畅竞争解释时，仅在60.2%的情况下胜出（显著高于随机水平50%，但远低于理想表现），且经独立模型进行语义保持重写后，虚假解释与真实解释间的距离仅增加2.1%，表明失败并非源于解释中存在私有约定。该结果限定于对话层面的事实正确性，未触及模型内部处理机制的真实性，因此论文进一步提出假设，认为上述现象可被该框架统一解释。
链接: https://arxiv.org/abs/2610.18727
作者: Kevin Yandoka Denamganaï
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: work in progress, open draft
Abstract:Natural Language Autoencoders (NLAs) are proposed as oversight tools for language models: an Activation Verbaliser (AV) writes an English explanation of an activation vector, and an Activation Reconstructor (AR) maps the explanation back to a vector. Their value for oversight depends on a human monitor being able to tell a true account of an activation from a false one, given the Av’s explanation. We formalise NLA training and NLA oversight deployment as two different referential games from Emergent Communication: training is a generative referential game, while oversight is a discriminative, multi-modal, partially-observable referential game played over a many-to-many relation between the monitored language model’s activations and the accounts that are true of them. The gap between the two games predicts that AV’s explanations will identify which activation they describe without discriminating its true account from closely matched false ones, since training never asks explanations to rule anything out. Using the NLA’s own AR as a maximally-sympathetic surrogate monitor model, we experimentally confirm this theoretical insight on one released checkpoint: true accounts beat accounts of other activations in all 320 comparisons, but beat fluent rivals differing in a single claim only 60.2% of the time (this http URL. 6.7%, chance at 50%, n = 35 ). A meaning-preserving rewrite by an uninvolved model costs only 2.1% of the distance to another activation’s account, so the failure is not due to the NLA pair having developed private conventions in the AV’s explanations. Our rival accounts test only transcript-level truths, which the monitored language model’s conversation settles, so these results bound the easiest case and leave claims about the monitored model’s own processing untested. We state conjectures that would make both results consequences of the framework.
[NLP-18] Code-Switching Detection via Language-Selective Dimensions in Speech Foundation Models ICASSP2027
【速读】： 该论文旨在解决多语言语音中语码转换（Code-switching, CS）现象在预训练语料库中代表性不足的问题，其核心挑战在于如何高效识别大规模语音数据中的语码转换样本以增强多语言语料的多样性。解决方案的关键在于提出一种轻量级的直接从语音信号中检测语码转换的方法：利用预训练语音编码器（如Whisper-small）在帧级别提供的语言信息，通过分析单语录音确定具有语言特异性的隐藏状态维度，并将这些维度的帧级激活值聚合为语码转换预测结果。该方法无需依赖高精度的多语言转录文本，仅需少量标注数据即可校准决策规则，实验证明其在CS-FLEURS数据集上对14种语言对的平均F1达到0.798，显著优于基于卷积神经网络（CNN）和HuBERT线性基线的单语语言识别模型。研究结果表明，预训练语音编码器已隐含识别语码转换所需的语言区分特征，从而为构建更具代表性的多语言语音语料库提供了一种高效且可扩展的技术路径。
链接: https://arxiv.org/abs/2610.18726
作者: Jie Chi,Zakaria Aldeneh,Masha Fedzechkina,Maureen de Seyssel,Natalie Schluter
机构: 未知
类目: Computation and Language (cs.CL)
备注: submitted to ICASSP 2027
Abstract:Code-switching (CS), the alternation between languages within an utterance, is common in multilingual speech but underrepresented in pretraining corpora. Increasing CS coverage requires finding relevant utterances in large speech collections so that they can be included in training data. One way to find such utterances is to transcribe the speech and detect language alternation in the text, but this depends on obtaining accurate multilingual transcriptions. We propose a lightweight method that detects CS directly from speech. It uses frame-level language information in pretrained speech encoders to determine whether both languages are present within an utterance. Specifically, we identify language-selective hidden-state dimensions using monolingual recordings and aggregate their frame-level activations into a CS prediction. With a small CS set used to tune a decision rule, we show that the Whisper-small encoder achieves a mean F1 of 0.798 across 14 language pairs in CS-FLEURS, compared with 0.776 for a CNN and 0.751 for a HuBERT linear baseline trained for monolingual language identification. Our findings suggest that pretrained speech encoders already encode the information needed to identify CS utterances, and they provide a lightweight tool for building more representative multilingual corpora.
[NLP-19] From Relation-Specific Heads to Shared Interventions: Contextual Entrainment in Language Models AACL
【速读】： 该论文旨在解决大语言模型在上下文学习中因上下文干扰（contextual entrainment）而导致的鲁棒性问题，即模型在回答后续问题时可能不恰当地复用早期无关上下文中的token。其核心解决方案的关键在于揭示：抑制这种有害复用行为并非依赖于为每种语义关系单独设计特定的注意力头，而是通过使用多样化的攻击性提示（heterogeneous attack prompts）识别出可跨关系共享的通用干预目标。研究发现，尽管部分注意力头表现出关系特异性，但大量关键干预头在不同语义家族间具有可迁移性；当移除这些全局共享头后，原本在同族关系间有效的转移能力显著下降，且子类型特异性的搜索反复重现了这些共享头。此外，干净完成度与上下文使用探测结果表明，此类共享干预可在抑制有害复制的同时保留正常的语义关系补全和基础的上下文学习能力。因此，该研究提出，对上下文干扰的缓解应聚焦于揭示由语义关系结构所塑造的可复用干预机制，而非孤立地寻找关系专属的神经回路。
链接: https://arxiv.org/abs/2610.18674
作者: Jinghan Zheng,Jiajun Shi
机构: McGill University (麦吉尔大学); Dalian University of Technology (大连理工大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted at AACL-IJCNLP 2026
Abstract:Large language models rely on prompt context to perform in-context learning, follow formats, and adapt to local evidence. This ability, however, also raises a robustness concern: models may reuse irrelevant tokens from earlier context when answering a later query. Prior work identifies this behavior as contextual entrainment and finds attention heads whose ablation reduces it, arguing that these heads are largely task- or relation-specific. But are the mechanisms behind contextual entrainment truly isolated by relation, or do relation-conditioned searches recover reusable components shared across relations? Using factual-distractor relation-completion prompts across 19 lexical-semantic subtypes, we identify attention heads whose ablation reduces entrainment and test how the resulting head sets transfer recovery ability across exact relations, semantic families, and mixed relation distributions. Same-family transfer exceeds different-family transfer, family residuals weaken once globally shared heads are removed, and subtype-specific searches repeatedly rediscover many global heads. Clean-completion and context-use probes further suggest that this shared intervention can reduce harmful copying while preserving ordinary relation completion and simple in-context abilities. Our findings provide a more precise perspective on contextual entrainment and its mitigation: harmful context following is not best addressed by searching for isolated relation-specific circuits, but by using heterogeneous attack prompts to reveal reusable intervention targets whose effects are shaped by semantic relation structure.
[NLP-20] SMI: Ontology-Grounded Semantic IDs for Hierarchy-Aware Medical Concept Representation NEURIPS2026
【速读】： 该论文旨在解决生物医学语言模型（Biomedical Language Models, LMs）在处理临床任务时难以捕捉医疗概念间层次结构的问题，尤其针对长尾词汇中罕见且细粒度的临床编码，其传统无层次感知的表示在概念频率降低时信息量显著下降。解决方案的关键在于提出一种无需训练的框架——语义医疗标识符（Semantic Medical Identifier, SMI），通过基于专家定义的医学本体（ontology）为每个诊断、操作或药物概念构建从粗到细的语义路径作为其语义标识（Semantic ID），并利用冻结的生物医学语言模型嵌入，在对应本体层级上进行聚合，从而生成保留粗粒度至细粒度临床语义的层次感知型医疗概念表示。该方法的核心创新在于将本体结构显式融入表示学习过程，通过均匀聚合或在验证数据上优化层级权重的变体实现对稀疏概念的有效表征。理论分析与实验结果均表明，SMI相较于无层次感知的LM嵌入能提供更具信息量的表示，显著提升临床关系预测性能，并在MIMIC-IV和eICU数据集上分别改善了死亡率预测、再入院预测、药物推荐及罕见药物预测等任务的表现，同时在跨医院分布偏移场景下展现出更强的患者表示鲁棒性。
链接: https://arxiv.org/abs/2610.18672
作者: Ziyang Song,Lia Shen,Yixuan Li,Qincheng Lu,Weihao Li,Haoran Xu,He Zhu,Xinyu Wang,Yue Li
机构: Ohio University(俄亥俄大学); University of Southern California(南加州大学); McGill University(麦吉尔大学); Northwestern University(西北大学); University of Washington(华盛顿大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to NeurIPS 2026 (Main Track)
Abstract:Biomedical language models (LMs) provide strong representations for clinical tasks, but they often fail to capture the hierarchical structure among medical concepts. This limitation is especially pronounced for rare and fine-grained clinical codes in long-tailed vocabularies, where hierarchy-agnostic LM representations become less informative as concept frequency decreases. We introduce Semantic Medical Identifier (SMI), a training-free framework for constructing ontology-grounded Semantic IDs and hierarchy-aware medical concept representations. For each diagnosis, procedure, or medication concept, SMI defines its Semantic ID as the coarse-to-fine path through an expert-defined medical ontology. SMI then aggregates frozen biomedical LM embeddings across the corresponding ontology levels to form hierarchy-aware medical concept representations that preserve coarse-to-fine clinical semantics. The base SMI uses uniform aggregation across ontology levels, while a level-weighted variant selects global level weights on validation data. Our theoretical analysis shows that SMI yields more informative medical concept representations than hierarchy-agnostic LM embeddings, particularly for rare and fine-grained clinical concepts. Experiments show that SMI captures meaningful hierarchical structure and improves clinical relation prediction. On MIMIC-IV, SMI improves mortality prediction, readmission prediction, drug recommendation, and rare drug prediction. On eICU, SMI yields more robust patient representations under cross-hospital distribution shift.
[NLP-21] Bayesian Test-Time Inference of Task-Aligned Similarity from Weak Interactive Feedback NEURIPS2026
【速读】： 该论文旨在解决生成式语言模型在测试阶段进行重复决策时，因相似性度量标准不唯一而导致的判断偏差问题。具体而言，模型在检索样例、排序候选答案及修正响应时，依赖于词汇、语义、结构或风格等多重相似性维度，而这些维度在不同任务实例中对“最优”决策的贡献权重可能动态变化，传统静态混合策略难以适应这种准则漂移（criterion shift）。其解决方案的关键在于提出一种无需训练的贝叶斯方法——BASIS（Bayesian Axis Selection via Interactive Signals），通过弱测试时反馈（如标量评分、成对偏好或验证器输出）来动态推断当前决策所遵循的任务特定准则。BASIS维护候选目标与准则之间的联合后验分布，并基于期望信息增益选择探测样本，从而实现对隐含决策准则的自适应识别与加权。实验表明，在概念恢复、推理样例选择与响应优化任务中，BASIS显著优于调优后的静态混合策略，成功将概念恢复成功率@1从56.0%提升至68.1%，推理准确率从72.3%提升至73.1%，优化胜率从61.4%提升至63.8%；且在计算成本相当条件下超越最佳-of-N采样方法。性能提升主要体现在准则错配和候选池较大时，且消融实验表明，自适应准则加权是核心优势，联合目标-准则推断进一步提升了系统表现。该方法仅在预设准则库内选择，不支持发现新准则。
链接: https://arxiv.org/abs/2610.18656
作者: Weida Liang,Kenji Kawaguchi
机构: National University of Singapore(新加坡国立大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: Accepted at NeurIPS 2026. 10 pages main text, 25 pages total including appendix; 4 figures, 15 tables
Abstract:Language model systems repeatedly make test-time decisions by comparing similarity: they retrieve exemplars, rank candidates, and revise responses. Yet similarity is not unique: lexical, semantic, structural, and stylistic resemblance can point to different candidates, and the task-aligned notion can change across instances. We introduce BASIS (Bayesian Axis Selection via Interactive Signals), a training-free Bayesian method that uses weak test-time feedback to infer which criterion in a supplied, task-specific library governs the current decision. It maintains a joint posterior over candidate targets and criteria, updates it from scalar scores, pairwise preferences, or verifier outputs, and selects probes by expected information gain. In a finite noisy model, we characterize when the target-criterion pair is recoverable and show why fixed-criterion and fixed-mixture rules can fail under criterion shift. Across concept recovery, reasoning exemplar selection, and response refinement, BASIS improves six-probe concept-recovery Success@1 from 56.0% to 68.1%, reasoning accuracy from 72.3% to 73.1%, and refinement win rate from 61.4% to 63.8% over tuned static mixtures. It also exceeds compute-matched best-of-N selection on reasoning (72.1%) and refinement (61.2%). Gains are largest under criterion mismatch and with larger candidate pools, and matched controls indicate that adaptive criterion weighting explains much of the benefit, with joint target-criterion inference adding further value. BASIS selects only among supplied criteria and does not discover new ones.
[NLP-22] Does Pretraining Teach Chemical Language Models Chemistry?
【速读】： 该论文旨在解决生成式化学语言模型（Chemical Language Models, CLMs）在分子属性预测任务中预训练效果是否可扩展及实际性能提升的问题。研究发现，尽管CLMs在大规模分子语料上进行预训练，但其下游任务性能并未随预训练规模持续提升，且在多数情况下表现饱和，甚至低于基于梯度提升的基线模型。其解决方案的关键在于系统评估多种预训练策略对模型性能的影响，结果表明：只有基于传统物理化学描述符（physico-chemical descriptors）的监督预训练能一致提升下游表现；而其他形式的预训练反而导致模型性能劣于随机初始化的编码器。研究进一步指出，尽管CLMs能够捕捉分子结构信息，但其能力与随机初始化模型相当，说明其在预训练过程中主要学习的是类似描述符的特征以及低层级的标记统计规律，而非超越经典分子特征的深层化学知识，揭示了当前CLMs在化学表征学习方面的局限性。
链接: https://arxiv.org/abs/2610.18651
作者: Leif Sieben,Yoel Zimmermann
机构: Independent Researcher; Harvard University (哈佛大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Chemical Physics (physics.chem-ph)
备注:
Abstract:Chemical language models (CLMs) are pretrained on large corpora of molecules with the aim of learning representations that transfer to diverse downstream tasks. Here, we systematically study the effects of pretraining using a series of such CLMs across 64 molecular property prediction datasets. We find that the downstream gains of pretraining in CLMs do not scale and performance saturates below gradient-boosting baselines. Across the regimes evaluated, only supervised pretraining on traditional physico-chemical descriptors consistently improves downstream performance for the CLMs tested whereas some forms of pretraining result in models worse than a randomly initialized encoder. While models can resolve molecular structure well, they do so no better than a randomly initialized model, suggesting that, at best, CLMs learn descriptor-like properties in addition to low-level token statistics during pretraining, but surprisingly little chemistry beyond what classical molecular features already capture.
[NLP-23] o Align or Not To Align: Check Your COMPASS Before You Train NEURIPS2026
【速读】： 该论文旨在解决预训练模型组合（如视觉编码器与语言模型对齐或微调检查点合并）过程中模型兼容性难以预测的问题。由于模型兼容性通常仅在实际对齐或合并后才能验证，传统方法依赖耗时的试错搜索，效率低下。为此，论文提出了一种基于理论动机的诊断工具COMPASS（COMPatibility ASsessment Score），其核心在于通过两个低成本计算的维度评估模型对的潜在兼容性：一是“特定对易度”（pair-specific ease），即两编码器在k-近邻（k-NN）空间中的重叠程度；二是“范式上限”（paradigm-level ceiling），表征其训练范式所能支持的最大重叠。这两个维度共同构成候选模型对的排序依据，并指导后续操作策略：优先选择已有高一致性且范式兼容的对，对范式兼容但初始一致性低的对投入额外算力，当现有预训练范式无法支持足够一致性时则切换范式，而对双维度得分均差的对则予以降权。在72个跨模态对齐实验和392个任务加法模型合并实验中，COMPASS的兼容性估计与下游性能高度相关，跨模态对齐任务平均皮尔逊相关系数达0.91，模型合并任务为0.70，显著提升了模型组合的效率与可预测性。
链接: https://arxiv.org/abs/2610.18638
作者: Fabian Gröger,Petar Damjanović,Maria Brbić
机构: EPFL(洛桑联邦理工学院); University of Basel(巴塞尔大学); HSLU(瑞士中部应用科技大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: NeurIPS 2026 camera-ready
Abstract:Combining pretrained models is a common strategy for building AI systems, whether by aligning a vision encoder with a language model or merging fine-tuned checkpoints. However, finding compatible pairs typically requires costly empirical search, as compatibility is known only after alignment or merging. We propose COMPASS (COMPatibility ASsessment Score), a theoretically motivated diagnostic that estimates whether a model pair will be compatible prior to alignment or merging. Building on the observation that representational convergence is most robust at the level of local neighborhoods, COMPASS decomposes model compatibility into two cheap-to-compute axes: a pair-specific ease, defined as the k -NN overlap between the two encoders, and a paradigm-level ceiling, capturing the maximum overlap their training paradigms allow. Together, these two axes provide a ranking of candidate model pairs and prescribe how to act on each pair: (i) prioritize pairs that already agree and whose paradigms are compatible, (ii) invest additional compute on pairs with compatible paradigms but low initial agreement, (iii) switch the pretraining paradigm when it cannot support sufficient agreement, or (iv) deprioritize pairs that score poorly on both axes. We apply COMPASS to 72 cross-modal alignment experiments and 392 task-arithmetic model merges and show that compatibility estimates by COMPASS strongly correlate with downstream success, achieving average Pearson correlation of 0.91 on cross-modal alignment tasks and 0.70 on model merging.
[NLP-24] Breakdown of Conversational Involvement: A Joint Multi-Modal Analysis of Conversation Floors and F-formation Detection with Multiple Features
【速读】： 该论文旨在解决在自由对话中如何准确识别和量化对话参与度（conversational involvement）的问题，特别是针对对话形成（F-formations）与对话主导权（conversation floors）的现有假设局限性。当前研究普遍认为F-formations能够有效捕捉单一对话单元，但本文指出这一假设存在不足，可能导致对真实互动结构的误判。其解决方案的关键在于提出并验证一种融合多模态特征与对话状态分析的综合框架：首先利用说话状态（speaking status）推断对话主导权，并将其与F-formation检测结果结合；其次通过对比不同特征集（如头部、上半身、下半身）在空间分布上的表现，发现基于头部特征的分组易被其他特征所包含；最后通过联合分析空间特征与对话主导权模式，揭示出某些算法管道可同时保留由上半身特征生成的单一主导权群体与由下半身特征保持的原始F-formations结构。因此，该研究的核心贡献在于强调需建立更精细的理论与整合框架，以全面刻画交互中的对话参与机制。
链接: https://arxiv.org/abs/2610.18627
作者: Zonghuan Li,Chirag Raman,Hayley Hung
机构: Delft University of Technology EEMCS (代尔夫特理工大学电子与计算机科学学院), The Netherlands (荷兰)
类目: Computation and Language (cs.CL); Social and Information Networks (cs.SI)
备注:
Abstract:Interactions happen universally and spontaneously in daily lives. Studying involvement in interactions helps us analyze who can influence who and has a wide potential in applications. Specifically, we consider conversational involvement in free-standing conversations. We study two operationalizations of an FCG: F-formations and conversation floors, and propose that current assumption that F-formations capture single conversations is insufficient. To illustrate how different implementations could impact the detected F-formations, we employ three analyses. First, we propose that speaking status can be used for estimating conversation floors, and leveraged it to analyze floors within F-formations. Second, we analyze spatial characteristics of the detection results by applying different feature sets on chosen baselines, and find that groups resulting from head features are more likely to be subsumed by groups detected using other features. Finally, we do a joint analysis which combines the spatial features and floor patterns in the conversation. We observe that certain pipelines can detect groups with single floors from upper body features, while retaining F-formations with lower body features. We conclude that a more detailed theory and framework is needed to holistically demonstrate conversational involvement within interactions.
[NLP-25] Does self-distillation improve reasoning ?
【速读】： 该论文旨在解决生成式模型在利用特权信息（privileged information）进行自我蒸馏（self-distillation）时，其性能提升是否真正源于对特权信息的有效利用，以及这种提升能否在分布外（out-of-distribution）任务中持续泛化的问题。现有方法如自蒸馏策略优化（SDPO）和在线自蒸馏（OPSD）虽在特定任务上表现出更高的样本效率与泛化能力，但本研究发现，这些优势在面对未见过的推理任务时并不稳定，且多数性能增益可归因于过拟合或已有推理行为的迁移，而非真正有效利用了特权信息。此外，实验表明自蒸馏梯度与直接优化期望奖励的策略梯度之间存在较弱的对齐性，暗示其可能并未提供有效的策略改进方向。因此，论文的核心挑战在于：如何在生成式模型中有效建模并利用特权信息以提升复杂推理能力，这一问题目前仍未得到解决。
链接: https://arxiv.org/abs/2610.18613
作者: Jason Ken Adhinarta,Raina Wu,Ashutosh Sharma,Avinash Balakrishnan,Rameswar Panda,Polina Golland,Yoon Kim,Zhang-Wei Hong
机构: Massachusetts Institute of Technology (麻省理工学院); MIT-IBM Computing Research Lab
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 33 pages, 11 figures, 24 tables
Abstract:Self-distillation is an emerging approach that trains a model to imitate its own predictions when conditioned on privileged information, such as correct answers or environmental feedback. Recent methods have reported better sample efficiency and generalization than reinforcement learning (RL), but it remains unclear whether these gains extend beyond in-distribution evaluations and whether they stem from privileged distillation. In this paper, we revisit two representative methods, Self-Distillation Policy Optimization (SDPO) and On-Policy Self-Distillation (OPSD), across code, science, tool use, and math reasoning tasks. We find that their reported gains do not consistently generalize to unseen problems and are largely explained by overfitting or transferring existing reasoning behavior, rather than effectively using privileged information. Moreover, self-distillation gradients show weak alignment with policy gradients that directly optimize expected reward, questioning whether they provide an effective direction for policy improvement. Our experiments across Qwen3 models (1.7B-14B) and OLMo-3-7B suggest that effectively leveraging privileged information to improve reasoning remains an open challenge. Our code is available at this https URL
[NLP-26] Legal-R1: Agent ic Retrieval for Evidence-Based Legal Reasoning
【速读】： 该论文旨在解决法律问答中答案缺乏可验证性与法律依据的问题，尤其针对传统检索增强生成（Retrieval-Augmented Generation, RAG）方法在处理法律证据时依赖单次语义匹配、难以兼顾权威性、适用性及异构来源多样性等缺陷。其核心解决方案是提出一种基于多智能体的法律推理框架Legal-R1，通过分解法律推理为由律师（Lawyer）、法律助理（Legal Assistant）和法官（Judge）三个智能体协同完成的流程：其中律师智能体负责决策何时检索、从何处查询、需获取何种证据以及判断证据是否充分；法律助理智能体则基于权威感知与多样性感知策略，从规范性法律文本和判例中精准检索证据；法官智能体最终基于完整的证据轨迹生成合规且可引用的法律答案。该框架仅对律师智能体进行掩码监督微调与轨迹感知强化学习的后训练，显著提升了答案的法律严谨性与可追溯性。实验结果表明，Legal-R1在多个中文法律基准测试中均优于通用、专用及检索增强型基线模型。
链接: https://arxiv.org/abs/2610.18611
作者: Huaiwen Pang,Jiale Han,Junying Chen,Wenjun Xiao,Benyou Wang
机构: The Chinese University of Hong Kong, Shenzhen(深圳大学); Shenzhen Loop Area Institute(深圳环区研究院); Shenzhen Shenten Digital Technology Co., Ltd.(深圳深腾数字科技有限公司)
类目: Computation and Language (cs.CL)
备注:
Abstract:Legal question answering requires answers that are not only plausible, but also grounded in legally valid and citable evidence. Standard retrieval-augmented generation often relies on one-shot semantic matching, which is insufficient when legal evidence must be selected according to authority, applicability, and diversity across heterogeneous sources. We introduce Legal-R1, an agentic retrieval framework for evidence-based legal reasoning. Legal-R1 decomposes legal reasoning into a multi-agent workflow, where a Lawyer agent learns when to retrieve, which legal source to query, what evidence to seek, and when the collected evidence is sufficient. A Legal Assistant agent retrieves from normative legal sources and precedent cases with authority-aware and diversity-aware selection, while a Judge agent generates the final answer from the complete evidence trajectory. We post-train only the Lawyer agent with masked supervised fine-tuning and trajectory-aware reinforcement learning. Experiments on multiple Chinese legal benchmarks show that Legal-R1 improves over general-purpose, legal-specific, and retrieval-augmented baselines.
[NLP-27] Jev-ToM: Evaluating Theory of Mind in Typed Decision Models
【速读】： 该论文旨在解决生成式大语言模型（Generative Large Language Models, LLMs）在实现有界心智理论（Theory of Mind, ToM）推理时存在的高成本、长延迟及决策不可靠等问题，尤其针对对话与协作类助手在多轮交互中频繁判断对方信念、意图或情感状态的挑战。其解决方案的关键在于提出一种基于类型化决策模型（Typed Decision Models）的新框架——Jev-ToM，通过系统比较五种类型化模型与七种生成式基线模型在11,117个英文文本问题上的表现，验证了类型安全的Jev模型在源平衡准确率上达到73.2%，并在六项能力维度中均位列第一或第二，尤其在信念、目标/欲望/意图和情绪等关键领域表现突出。研究进一步表明，该方法可显著降低重复ToM选择的成本与等待时间；同时揭示出置信度选择策略具有任务依赖性，且附加解释的上下文位置对纠错或误导正确答案具有重要影响。此外，候选集扩展暴露出选择偏差与输出有效性问题，而相关问题提示（related-question cues）是提升下游性能的主要因素，预测状态的影响则具有任务特异性。因此，该研究建议将Jev-ToM作为未来智能助手中的核心ToM选择组件，并优先关注任务特定的置信度校准、候选设计优化与输出验证机制的集成。
链接: https://arxiv.org/abs/2610.18610
作者: Kawai Chung,Chunkit Chan,Yuxuan Liu,Yauwai Yim,Xi Yang,Wenbin Hu,Huihao Jing,Chiu Yeung Tang,Changxuan Fan,Ki Sen Hung,Tsun On Kwok,Haoran Li
机构: The Hong Kong University of Science and Technology(香港科技大学); Department of Computer Science and Engineering(计算机科学与工程系)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Code and evaluation materials: this https URL
Abstract:Dialogue and collaborative assistants repeatedly judge what a partner knows, believes, or intends. Typed decision models offer a potential alternative to generative large language models for bounded Theory of Mind (ToM) selection. We present Jev-ToM, comparing five typed decision models and seven generative baselines on 11,117 English text questions from nine ToM benchmarks across six capability groups. TypeSafe Jev reaches 73.2% source-balanced accuracy including native controls and ranks first or second across all six capabilities, leading beliefs, goals/desires/intentions, and emotions. Other typed models show distinct strengths across tasks. Separate resource comparisons suggest that Jev could reduce the cost and waiting time of repeated ToM selection. Confidence-based selection remains benchmark-dependent, and supplied reasons can repair errors or overturn correct answers depending on their attachment. Candidate expansion reveals distinct selection and output-validity problems. Related-question cues account for most of Jev’s downstream gains, with task-dependent effects from predicted states. The findings motivate testing Jev as a ToM selection component in assistants, with task-specific confidence calibration, candidate design, and output validation as integration priorities.
[NLP-28] Beyond Answer Voting: Fine-Grained Correction of LLM Reasoning through Graph Alignment across Sampled Traces
【速读】： 该论文旨在解决大语言模型在多步推理任务中因中间步骤错误导致最终答案偏差的问题，尤其针对现有自一致性（Self-consistency, SC）方法仅对最终答案进行投票而无法有效处理中间推理环节分歧的局限性。其解决方案的关键在于提出一种名为DDRG（Disagreement-Driven Verification Guided by Reasoning Graphs）的框架，通过构建基于源文本的命题图（claim graphs），利用图对齐技术识别出不同推理路径间的细粒度命题级分歧，并将这些局部冲突传递给一个不依赖答案信息的验证器（answer-blind verifier）。该验证器基于推理路径支持证据进行反证，若能引导至多数答案所对应的支撑路径，则触发一次精准修复；否则返回原始自一致性投票结果。该方法无需扩展初始推理轨迹库，实现了基于局部验证的可选性修正，显著提升了零样本场景下的推理准确性。实验表明，相较于直接回答、思维链（CoT）、GoT、GoV及传统自一致性方法，DDRG在五个基准测试上均取得更高宏平均准确率，尤其在Qwen2.5-7B和Llama3.1-8B模型上分别提升8.65和15.83个百分点。
链接: https://arxiv.org/abs/2610.18606
作者: Haotian Liu,Yingjian Chen,Xiaohang Xu,Lang Huang,Yusuke Iwasawa,Yutaka Matsuo,Irene Li
机构: Xiamen University (厦门大学); The University of Tokyo (东京大学); National Institute of Informatics (NII)(日本信息研究所)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 19 pages, 5 figures, 4 tables. Under review
Abstract:Large language models increasingly tackle tasks that require multi-step reasoning, where errors in intermediate claims can affect the final answer. Self-consistency (SC) improves reasoning by voting over sampled final answers, but its vote leaves intermediate disagreements unresolved. To address this problem, we introduce DDRG (Disagreement-Driven Verification Guided by Reasoning Graphs), a framework for selective correction from a frozen bank of reasoning traces. Our insight is that graph alignment can expose claim-level disagreements and provide correction evidence without expanding the sampled trace bank. DDRG builds source-grounded claim graphs and sends localized conflicts to an answer-blind verifier. A verifier-supported refutation that reaches the majority answer’s supporting path triggers one focused repair; otherwise, DDRG returns the exact SC vote. Experimental results across five benchmarks show that native zero-shot DDRG improves macro-average accuracy over direct answering by 8.65 and 15.83 percentage points for Qwen2.5-7B and Llama3.1-8B. It also exceeds CoT, the evaluated GoT and GoV question-answering adaptations, and SC on this macro metric. DDRG links local verification to selective correction without enlarging the initial candidate bank.
[NLP-29] From Hindsight to Foresight: Dissecting Clinical Evidence Acquisition in Large Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在动态临床诊断中表现不佳的问题，即在面对随时间演进的、非静态的临床情境时，难以有效获取并利用关键诊断证据进行前瞻性判断。现有研究多聚焦于提升模型的知识储备或优化后续决策策略，而本文提出一种全新的“事后反思”（hindsight）分析框架，系统性地剖析LLMs在动态诊断过程中失败的根本原因。其解决方案的关键在于从多维度（包括诊断准确性、证据覆盖度、诊断轨迹动态性、诱发概率与信息熵等）对模型在证据识别、证据利用及诊断更新三个阶段的表现进行深入分解，揭示其在事件发生前（预事件状态）、事件发生时（证据到达前）以及事件发生后（证据到达后的状态转换）的能力瓶颈。通过从能力层面的分解到跨阶段的动态演化分析，研究不仅验证了这些缺陷的普遍性，更将核心挑战从“模型是否具备临床知识”转变为“模型能否在真实诊疗场景中准确识别、获取并有效运用关键证据”，从而推动对诊断机制的精细化机制分析与可解释性改进。
链接: https://arxiv.org/abs/2610.18605
作者: Ziyang Ling
机构: University of Science and Technology of China (中国科学技术大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models (LLMs) often diagnose full sight, static clinical cases well yet struggle in foresight, dynamic diagnosis. Prior work has improved this limitation by quantifying the information models acquire and by developing stronger follow-up strategies to improve diagnostic accuracy. In contrast, we introduce a distinctive hindsight to systematically dissect why LLMs fail to acquire diagnostically critical evidence, tracing the failure from evidence identification, utilization and downstream diagnostic updating through not only diagnostic accuracy, but also evidence coverage, trajectory dynamics, and elicited probability and information entropy. Starting experiments from the breakdown at the level of capabilities, to the diagnosis of the pre-event state before the evidence arrives, and to the post-event state transition after the evidence arrives, our research has revealed the profound limitations and internal causes existing in the dynamic diagnostic process. At the same time, we have also conducted sufficient verification and generalization experiments to prove that these limitations are widespread. Together, these findings shift the central challenge from whether LLMs possess relevant clinical knowledge to whether they can identify, acquire, and effectively use the right evidence under clinical diagnosis and conduct a more detailed mechanism analysis to identify the exact location of the capability gap.
[NLP-30] Rethinking Projector Training in Multimodal LLM s NEURIPS2026
【速读】： 该论文旨在解决多模态大语言模型（Multimodal LLMs）在训练过程中依赖大量图像-文本配对数据进行投影器（projector）训练所带来的效率与可扩展性问题。传统方法通过在大规模图像-文本数据集上进行投影器训练以实现视觉编码器与语言模型之间的对齐，但其关键阶段的作用机制尚不明确。研究发现，仅需10%至20%的训练数据即可获得绝大部分下游任务性能提升，且不同数据集训练出的投影器可通过线性插值保持良好性能，表明投影器训练本质上完成的是粗粒度对齐，多种投影器均可实现类似效果。基于此，作者提出PORTAL——一种无需训练、不依赖配对数据或梯度优化的投影器初始化方法，仅利用各模态的统计摘要信息，通过最优传输（optimal transport）构建视觉与语言特征分布在共享主子空间上的映射关系，从而完全跳过投影器训练阶段。实验表明，PORTAL在六个语言模型骨干网络、两种视觉编码器和16个基准测试中，无需任何配对数据即达到标准训练流程（使用55.8万张图像-文本对）的性能水平，在7组设置中有3组表现更优，并在其余4组平均达到其98%的性能，验证了该方法的有效性与普适性。
链接: https://arxiv.org/abs/2610.18598
作者: Fabian Gröger,Naël Ouerghemi,Shuo Wen,Maria Brbić
机构: EPFL(洛桑联邦理工学院); University of Basel(巴塞尔大学); HSLU(瑞士高等教育学院)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: NeurIPS 2026 camera-ready
Abstract:Multimodal LLMs are typically trained by first aligning a vision encoder with a language model via projector training on large image-caption datasets, followed by joint training of the full multimodal model. Despite its widespread use, the role of the projector training stage remains poorly understood. We find that (i) using only 10 to 20% of the data for training the projector already recovers most downstream performance gains, and (ii) projectors trained on different datasets can be linearly interpolated while largely preserving performance. These results suggest that projector training mainly performs a coarse alignment, which many different projectors achieve equally well. Motivated by these findings, we propose PORTAL, a training-free projector initialization that requires neither paired image-caption data nor gradient-based optimization, but only per-modality summary statistics. PORTAL computes an optimal-transport-based map between Gaussian approximations of the vision and language feature distributions on a shared principal subspace, so that the projector training stage can be skipped entirely. Without any paired data, PORTAL matches the standard pipeline trained on 558k image-caption pairs across six LLM backbones, two vision encoders, and 16 benchmarks. It outperforms that pipeline in 3 of 7 (vision encoder, LLM) settings and reaches an average 98% of its score across the other four.
[NLP-31] Decomposing and Predicting Annotation Disagreement in Harmful Content Across Languages
【速读】： 该论文旨在解决主观任务中注释者分歧（annotator disagreement）的预测问题，其核心挑战在于如何有效识别那些容易引发争议的内容，从而优化标注资源分配并引导复杂案例进入人工审核流程。传统方法常将分歧视为需消除的噪声，而本文提出将其视为具有信息量的属性加以利用。解决方案的关键在于通过大语言模型（LLM）对一系列结构化二元问题进行回答，这些问题涵盖有害内容特征、与分歧相关的线索（如边界性严重程度、上下文缺失）以及语言敏感性特征（如局部侮辱性用语），进而基于这些可解释的语义属性构建分类器来预测分歧。实验表明，该方法在英、德、土三语共四个数据集上，相比直接提示（direct prompting）提升了10.5个ROC-AUC点，优于最强的文本基线模型5.0个点；仅使用其中8个关键问题即可达到接近完整20题集的效果（差距<1.5 ROC-AUC点）。跨数据集评估显示模型具备良好的泛化能力（保留90%-95%的域内性能）。分析揭示，关于贬损性表达、上下文缺失及判断边界的问题是最重要的预测因子，表明有害内容为何具有争议性的关键线索部分可从文本本身可观测。
链接: https://arxiv.org/abs/2610.18588
作者: Irem Ecem Yelkanat,Faeze Ghorbanpour,Alexander Fraser
机构: TU Munich(慕尼黑工业大学); Munich Center for Machine Learning (MCML)(慕尼黑机器学习中心)
类目: Computation and Language (cs.CL); Computers and Society (cs.CY); Multimedia (cs.MM)
备注:
Abstract:Annotator disagreement is increasingly recognized as a meaningful property of subjective tasks, rather than merely noise to be removed. Predicting which items are likely to be contested could help allocate annotation effort and route difficult cases to human review. We propose predicting disagreement from interpretable semantic properties of the text. An LLM answers a set of binary questions covering harmful-content characteristics, disagreement-related cues such as borderline severity and missing context, and language-sensitive cues such as local slurs. A classifier predicts disagreement from these answers, and we analyze which questions contribute most to the predictions. Across four datasets in English, German, and Turkish and three open-weight LLMs, our structured question features outperform direct prompting by 10.5 ROC-AUC points and the strongest text-based baselines by 5.0 points. A compact subset of eight questions performs within 1.5 ROC-AUC points of the full twenty-question set. In leave-one-dataset-out evaluation, models retain roughly 90-95% of their in-domain ROC-AUC. Across models and datasets, questions about derogation, missing context, and decision boundaries serve as the strongest predictors, suggesting that what makes harmful content contested is partly observable in the text.
[NLP-32] he Many Morals of AI: The cautionary bias of North American news narratives in global perspective
【速读】： 该论文旨在解决人工智能（Artificial Intelligence, AI）新闻叙事在不同区域媒体系统中存在差异的深层原因及其对公众认知与政策讨论的影响问题。其核心挑战在于揭示新闻报道中隐含的价值导向性叙事（即“故事寓意”），而不仅仅是表面的主题或情感倾向。解决方案的关键在于利用大语言模型（Large Language Models, LLMs）从12,924篇2022年以来的英文新闻文章中自动提取每篇文章所传递的隐含叙事教训，从而捕捉超越主题和情绪的、具有价值取向的深层解读。研究发现，北美地区的AI新闻叙事以警示性为主导，而对创新与合作类叙事的呈现明显不足，但这一现象主要归因于北美新闻整体存在的负面情绪偏倚，而非对AI技术本身特有的担忧。该研究揭示了AI新闻叙事的文化情境性，强调了媒体系统结构与区域文化心理在塑造公众科技认知中的关键作用。
链接: https://arxiv.org/abs/2610.18573
作者: Andrew Piper,Edie Pearman
机构: McGill University (麦吉尔大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 4 pages, 5 figures
Abstract:The stories told about artificial intelligence (AI) play a central role in shaping public understanding, policy debates, and expectations about technological change. News media are especially influential in this process. Using a cross-regional corpus of 12,924 English-language news articles about AI published since 2022, we examine how storytelling about AI varies across media systems in four regions: North America, China, India, and the Middle East. Leveraging large language models, we extract the implicit narrative lessons conveyed by each article—commonly known as story morals—capturing value-laden interpretations that extend beyond thematic content or sentiment. We find that North American AI coverage is uniquely characterized by narratives of caution, while under-emphasizing narratives of innovation and collaboration that are more prevalent elsewhere. However, this pattern is largely explained by a broader regional negativity bias in North American news sentiment, rather than by heightened concern about AI itself. Our findings bring into view the culturally conditioned nature of AI news narratives.
[NLP-33] CREATE: Creativity Evaluation of VLMs using the Visual Remote Associates Test
【速读】： 该论文旨在解决生成式视觉语言模型（Vision Language Models, VLMs）在创造性任务中缺乏系统性评估方法的问题，尤其关注其类比推理（lateral reasoning）能力与人类水平之间的差距。现有研究虽广泛使用VLMs于视觉叙事、设计创意生成等场景，但尚无可靠基准来衡量其跨模态联想能力。为此，作者提出了CREATE——一个半自动构建的基准数据集，包含155个基于视觉远端关联测试（Remote Associates Test, RAT）的图像问题，要求模型从三幅看似无关的图像中识别出潜在的语义关联。通过开展25名人类参与的对照实验，验证了该基准的有效性与合理性，并对11种主流VLM进行了评估。研究结果表明，多数VLM在该任务上表现显著低于人类水平；进一步分析揭示，模型的联想能力高度依赖文本提示中的概念线索，不具备普遍适用于各类视觉谜题的跨模态泛化能力。因此，该研究的关键贡献在于提出了一种可信赖的评估框架，并揭示了当前VLM在深层创造性推理方面的局限性。
链接: https://arxiv.org/abs/2610.18569
作者: Ziqing Yang,Anubhav Jangra,Smaranda Muresan
机构: Barnard College, Columbia University (巴纳德学院，哥伦比亚大学); Columbia University (哥伦比亚大学)
类目: Computation and Language (cs.CL)
备注: 18 pages, 9 figures
Abstract:Vision language models (VLMs) are used in creative tasks such as visual storytelling, design ideation, art interpretation, and advertising. Yet we lack principled ways to measure the lateral reasoning of these models compared to human potential. We introduce CREATE, a semi-automatically constructed benchmark containing 155 visual Remote Associates Test problems. These problems require a solver to find an underlying connection among three seemingly dissimilar images. We conduct a human study (N=25) to validate and ground the quality of our benchmark, and evaluate 11 VLMs. Our findings show that most VLMs underperform humans on this task. Our experiments indicate that the associative ability of VLMs depends on the cue concepts provided in the text and does not universally extend to visual puzzles.
[NLP-34] Empathy Is Not the Scapegoat for Declines in Factuality and Reliability AACL2026
【速读】： 该论文旨在解决生成式 AI 在压力或不确定情境下如何在保持可靠性的同时提升情感温暖与共情能力的问题。现有研究普遍认为，增强模型的温暖语气会降低其可靠性，但该研究指出这一结论存在局限性——以往工作多基于共情对话数据进行微调（SFT），难以区分语气变化与因SFT引发的分布偏移之间的效应。为此，论文从模型训练与评估双层面重新审视该问题，系统考察了提示工程（prompting）、SFT及强化学习（RL）在引导温暖行为中的作用，并采用包含11个数据集的基准测试套件和基于情感上下文的EMOFAC T评估框架，全面衡量模型能力的保留情况。研究发现，温暖性影响具有任务与模型依赖性，而与任务对齐的自适应语气可同时提升共情表现与目标可靠性，尽管仍存在部分能力权衡。此外，论文引入经人工验证的共情评估器，专门用于评估模型在情感场景下的响应质量。结果表明，通过合理设计，模型可在保持可靠性的前提下显著增强社会互动中的支持性。
链接: https://arxiv.org/abs/2610.18541
作者: Letian Shi,Kanwal Mehreen,Deheng Zhang,Yuxia Wang
机构: 未知
类目: Computation and Language (cs.CL)
备注: Accepted by AACL 2026
Abstract:Language models are increasingly expected to support users in stressful or uncertain situations, where warmth and empathy are desirable. While recent work associates warmer models with reduced reliability, we show that this conclusion is incomplete. Prior studies often train primarily on empathetic dialogue, making it difficult to separate tone effects from SFT-induced distribution shift. We revisit this issue at both the model and evaluation levels. We study prompting, SFT, and RL for inducing warm behavior, and comprehensively evaluate capability preservation using an 11-dataset benchmark suite and EMOFACT under emotional contexts. Across multiple model families, we find that warmth effects are task- and model-dependent, while task-aligned adaptive tone can improve empathy and targeted reliability, although some capability trade-offs remain. We further introduce human-validated empathy evaluators specifically developed to assess empathetic model responses. Our results suggest that models can be more supportive in social interactions while remaining reliable.
[NLP-35] How Reliable Are System-One Decisions in Medicine? A Comparative Study of Jev Jev-like Models and LLM s
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在医疗决策中快速、通用的判断能力与其医学可靠性之间存在的显著差距问题。其核心挑战在于：尽管生成式AI具备高效的推理能力，但在面对数值驱动型任务时表现欠佳，且普遍缺乏对不可靠决策的有效识别与主动回避能力（即“可靠弃权意识”）。研究的关键解决方案在于通过统一协议在15个医疗数据集（共8,742个实例）上系统评估多种“系统一”（System-1）决策模型的性能，不仅关注准确率，更引入决策稳定性与不可靠判断识别能力作为关键评价维度。研究发现，生成式AI（Jev）在以文本为主的任务中表现优异，且在不同问题表述下展现出最强鲁棒性，决策反转率最低；然而，其在缺失正确答案场景下的识别能力及支持可选弃权机制方面仍存在明显缺陷。进一步分析表明，通过将选项级别的二元评分进行归一化处理而非采用传统选择评分策略，可显著提升其基于概率的答案选择准确性。因此，本研究为提升生成式AI在医疗场景中的可信应用提供了早期实证指导。
链接: https://arxiv.org/abs/2610.18529
作者: Yanrui Du,Danyang Zhao,Sendong Zhao,Wenhao Li,Yibo Gao,Haochun Wang,Zhuoer Xu,Shiwen Cui,Bing Qin
机构: Harbin Institute of Technology(哈尔滨工业大学); Ant Group(蚂蚁集团)
类目: Computation and Language (cs.CL)
备注:
Abstract:Jev has attracted considerable attention for its rapid, generalizable decision-making, yet its medical reliability remains under explored. To address this gap, our study compares various types of system-one decision models across 15 medical datasets comprising 8,742 instances under a unified protocol. Beyond accuracy, we assess decision stability and the ability to recognize unreliable decisions. Our results show that Jev rivals leading models on primarily text-driven tasks but is less effective on numerically driven tasks, revealing a distinct capability boundary. Under question-format variations, Jev demonstrates strong robustness, achieving the lowest decision reversal rate and relatively small accuracy fluctuations. Among evaluated open-source alternatives, only StartLux exhibits comparable robustness. But these strengths do not extend to reliable abstention awareness: most evaluated models show limitations in both recognizing missing correct answers and using optional abstention. Moreover, we find that Jev’s probability-based answer selection can be improved by changing how decisions are scored: normalizing option-wise binary scores yields higher retained accuracy than choice scoring. Taken together, our findings provide early guidance for improving and applying decision models in medicine.
[NLP-36] he Geometry of Language Model Logits: Generic Structure and Learned Departures
【速读】： 该论文旨在解决生成式语言模型在自回归解码过程中，其词汇头（vocabulary-head）输出的对数几率（logit）分布结构背后的几何规律及其与模型性能之间关系的问题。具体而言，研究发现：无论在何种预测上下文、模型架构或规模下，词汇索引排序后的对数几率呈现一种高度一致的对数线性关系 $ z_r \approx \mu - \alpha \log r $，且该结构在预训练过程中逐步显现。这一现象首次系统揭示了以往解码方法所依赖的分布假设背后的对数几率几何特性。尽管受控的极值统计基线（如有界、轻尾和重尾分布）可解释该对数线性形式，但无法解释斜率 α 的系统性偏移——训练后模型的 α≈1.4–1.7 显著高于初始化状态下的 α≈0.2，且 α 与模型预测性能显著相关，表明其捕捉到了超越一般极值效应的深层学习结构。因此，解决方案的关键在于识别并量化这种由训练驱动的对数几率几何规律，揭示其作为模型内在表示能力的度量潜力。
链接: https://arxiv.org/abs/2610.18520
作者: Swathi Shree Narashiman,Aravind Ramana Venkatramanan,Nambi Seshadri
机构: Stanford University (斯坦福大学); Indian Institute of Technology Madras (印度理工学院马德拉斯分校); University of California San Diego (UCSD) (加州大学圣地亚哥分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Autoregressive language models produce, at every prediction step, a vocabulary-sized logit vector whose structure determines the conditional next-token distribution. We empirically identify that the ordered vocabulary-head logits follow a remarkably consistent log-linear relation [ z_® \approx \mu - \alpha \log r. ] across prediction contexts, architectures, and scales, emerging progressively during pretraining. To our knowledge, this is the first systematic analysis of the logit geometry underlying the distributional assumptions used in the prior decoding work. While controlled order-statistic baselines spanning bounded, light-tailed, and heavier-tailed distributions can largely explain the observed log-linear geometry, they do not explain the systematic shift in the slope \alpha : trained language models exhibit \alpha \approx 1.4 – 1.7 , sitting well above the initialization baseline ( \alpha \approx 0.2 ) . Moreover, \alpha correlates with predictive performance, suggesting that it captures learned structure beyond generic extreme-value effects. Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) Cite as: arXiv:2610.18520 [cs.CL] (or arXiv:2610.18520v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2610.18520 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-37] Imprinting: Knowledge Distillation Can Shift Student Behavior
【速读】： 该论文旨在解决知识蒸馏（Knowledge Distillation）过程中，教师模型（teacher model）除任务能力外，其隐含的行为倾向是否会被无意中传递至学生模型（student model）的问题。研究发现，尽管蒸馏过程仅针对常规推理与知识类任务进行监督，但学生模型在获得任务能力的同时，其下游决策行为会显著向教师模型的安全性取向（如对伤害的接受或规避倾向）靠拢，这一现象被称为“印记效应”（imprinting）。解决方案的关键在于：通过设计具有不同安全倾向的同源教师模型（如危害接受型与危害规避型），可有意识地将特定行为倾向注入学生模型，且这种影响不依赖于显式的安全标注或强化学习。实验表明，去除少量差异性提示词（divergence tokens）可减弱印记效应，而响应重写虽保留源依赖性影响，还可能引入新的重写者偏差；后续偏好优化虽能提升危害规避程度，但无法完全消除教师源依赖性差异。这表明，教师模型的行为倾向可通过知识蒸馏被系统性地“印刻”到学生模型中，揭示了模型训练中潜在的隐性偏见传播风险。
链接: https://arxiv.org/abs/2610.18517
作者: Yu-Chien Tang,An-Zi Yen
机构: National Yang Ming Chiao Tung University (国立阳明交通大学), Taiwan
类目: Computation and Language (cs.CL)
备注:
Abstract:Knowledge distillation trains student models on teacher-generated responses to transfer task capability, but those responses may also carry behavioral tendencies of their source. We investigate whether the source teacher can shape student behavior beyond the capability that distillation is intended to transfer. We construct harm-accepting and harm-avoiding variants of the same base teacher, use them to generate responses to ordinary reasoning and knowledge questions unrelated to safety, and distill those responses into students with different base models. Students generally acquire task capability while their downstream decisions shift toward the source teacher’s safety-pragmatism disposition, a phenomenon we call imprinting. Distillation from the harm-accepting teacher reduces harm avoidance, whereas distillation from the harm-avoiding teacher increases it. Imprinting varies across teacher-student pairs and does not extend uniformly across different forms of safety behavior. Excluding a sparse set of divergence tokens reduces imprinting, whereas response rewriting preserves the source-dependent effect and can introduce additional rewriter-dependent shifts. Post-distillation preference optimization raises harm avoidance but does not eliminate the source-dependent difference. These results show that behavioral tendencies can be intentionally imprinted through the source teacher, even with ordinary task supervision.
[NLP-38] he Last Mile of Coding Agents for Long-Document Tasks: An Empirical Study on Clinical Trial Protocol Digitization
【速读】： 该论文旨在解决临床试验方案（Clinical Trial Protocol）数字化过程中，如何将非结构化的PDF文档高效、准确地转化为机器可读的结构化数据这一关键问题。当前，临床试验方案作为研究设计的“真相来源”，普遍以非结构化PDF形式存在，难以被临床软件直接分析与执行。尽管生成式AI（Generative AI）中的编码代理（Coding Agent）在处理长文档任务方面展现出潜力，但其在完整文档到精确结构化输出转化过程中的可靠性仍不明确。为此，作者构建了TrialPD-Bench——一个由专家标注的基准数据集，包含150份公开可用的协议PDF，覆盖四大任务类别。评估不仅关注最终答案的准确性，还系统考察了代理行为轨迹中的证据召回率、推理模式及计算成本。实验结果表明，即使是最先进的九个专有与开源模型，在五种编码代理框架下均未能完全解决该任务。深入分析发现，代理通常能准确定位所需信息，但在最后一步的转录与结构化环节仍存在显著缺陷。针对此瓶颈，研究提出一种领域专用的代理框架TrialScribe，通过引入任务特定技能和验证闭环机制，显著提升了代理性能：相较于本地部署的开源模型，其在八项指标上提升6.2至59.4分，达到甚至超越如GPT-5.6-Sol等专有模型的水平。因此，解决方案的关键在于构建具备领域知识嵌入与自我验证能力的专用代理框架，以弥合从信息定位到结构化输出之间的“最后一英里”差距。
链接: https://arxiv.org/abs/2610.18507
作者: Jiacheng Lin,Geoffrey Low,Nishit Agarwal,Rahul Jain,Afrah Shafquat,Mandis Beigi,Jimeng Sun
机构: University of Illinois Urbana-Champaign(伊利诺伊大学厄本那-香槟分校); Medidata Solutions(美迪赛尔解决方案)
类目: Computation and Language (cs.CL)
备注:
Abstract:Clinical trial protocols serve as the source of truth for study design, but are mostly written and used as unstructured PDF documents. Digitizing protocols turns static content into structured, machine-readable data that clinical software can analyze, validate, and execute. Coding agents, which select tools, execute code, and revise their approach as they work, are increasingly applied to such long-document tasks, yet it remains unclear how reliably they turn complete documents into accurate structured outputs. We study this question using protocol digitization as a testbed. We introduce TrialPD-Bench, an expert-annotated benchmark of 150 publicly available protocol PDFs spanning four task categories. Beyond final-answer performance, our evaluation measures evidence recall from agent traces, agent behavior, and cost. Across nine proprietary and open-weight models in five coding-agent harnesses, even the strongest agents do not fully solve protocol digitization. Our analysis further shows that agents usually locate the required evidence but fail in the last mile of transcribing and structuring it. We further develop TrialScribe, a domain-specific harness that equips coding agents with task skills and a verification loop. TrialScribe improves a locally served open-weight model by 6.2-59.4 points across eight metrics, enabling it to match or surpass proprietary models such as GPT-5.6-Sol.
[NLP-39] Aligning Faithfulness for Theory-Scale Autoformalization
【速读】： 该论文旨在解决生成式 AI 在自动形式化（Autoformalization, AF）过程中存在的“忠实性”（faithfulness）评估难题。现有方法多在语句层面进行评价，仅给出整体判断，无法揭示非形式化与形式化表达之间的细粒度对应关系，也难以识别支持判断的上下文解释和隐含假设，导致评估结果缺乏可解释性。由于自动形式化本质上具有理论尺度（theory-scale）特征——目标陈述的意义依赖于周围数学内容及未明言的约定，因此必须在上下文感知的前提下进行精确匹配与判断。为此，论文提出将忠实性评估建模为“对齐-标注”（align-then-label）问题：在显式指定上下文解释与假设的基础上，识别非形式化与形式化语句间的短语级对应关系，并为每一对应分配忠实性标签。为此，研究构建了首个理论尺度的忠实性评估基准 FaithAlign，包含460个样本及4,982条人工标注的短语级链接；同时引入 FaithAlign-Gap 作为补充数据集，其中隐藏非形式化上下文与假设，以测试模型在信息缺失下的推理能力。实验表明，尽管 GPT-6 Astra 在8个模型中表现最优（缺陷检测F1达47.8%），但在存在多个缺陷的样本中仅能识别全部缺陷的4.5%，且在 FaithAlign-Gap 上仅能识别28.2%的不可判定项，凸显了恢复上下文依赖对应关系的挑战性。该研究确立了 FaithAlign 作为细粒度、上下文感知的忠实性评估基准的重要地位。
链接: https://arxiv.org/abs/2610.18498
作者: Yuming Feng,Tadd Mao,Dhruva Arekar,Tianjun Zhong,Jiani Huang,Xujie Si,Ziyang Li
机构: Johns Hopkins University(约翰霍普金斯大学); University of Toronto(多伦多大学); University of Pennsylvania(宾夕法尼亚大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Autoformalization (AF) requires more than compilation: formal statements must preserve the meaning of their informal sources, a property known as faithfulness. An emerging frontier is the automated evaluation of this property, yet most existing methods operate at the statement level and produce whole-item verdicts, failing to reveal how individual expressions correspond or which contextual interpretations support the judgment. This omission is consequential because AF is inherently theory-scale: target statements derive their meaning from surrounding mathematical material and unstated conventions. We formulate faithfulness evaluation as a context-aware align-then-label problem: identifying phrase-level correspondences between informal and formal statements and assigning a faithfulness label to each under explicitly specified contextual interpretations and assumptions. We also introduce FaithAlign, the first theory-scale faithfulness benchmark comprising 460 items and 4,982 annotated phrase-level links, alongside FaithAlign-Gap, a companion set of 146 items with informal context and assumptions withheld. Each FaithAlign item includes the target statement pair, relevant context, formal dependencies, and, where needed, assumptions specifying one given interpretation. Across 8 models, GPT-6 Astra achieves the highest defect F1 at 47.8% and identifies all defects in only 4.5% of items with two or more defects. On FaithAlign-Gap, Astra identifies only 28.2% of undecidable items as such. These results show the difficulty of recovering context-dependent correspondences, establishing FaithAlign as a testbed for fine-grained, context-aware faithfulness evaluation.
[NLP-40] Judges Follow the Reference: The Derivation Gap in Verifying Tool-Call Arguments
【速读】： 该论文旨在解决生成式 AI（Generative AI）在评估大语言模型（LLM）代理所提出论点值时存在的“推导差距”（derivation gap）问题，即当评判模型的输入中缺乏明确参考文本（reference）时，即使论点值正确，其被接受的概率也会显著下降。这一现象出现在所有七种评测模型中，包括系统一模型（System One models）如CLM和Jev，且并非仅由输入截断引起——即使在无截断的基准测试中，正确默认值仍表现出最大差距。关键解决方案是引入DeRef（Derived Reference）机制：一个独立模型在不观察候选值的前提下，基于完整历史记录生成可被评判模型访问的参考文本，从而为评判提供明确依据。实验表明，在三个基准测试中，DeRef显著提升了所有评测模型的整体论点评估性能（AUROC），并在受控测试中将AUROC从随机水平或更低提升至98.40至99.59之间，证明了参考文本可见性对评判准确性具有决定性影响。
链接: https://arxiv.org/abs/2610.18497
作者: Linhai Ma,Yan Wang,Lingfei Qian,Yupeng Cao,Victor Gutierrez Basulto
机构: The Fin AI(金融人工智能公司); Yale University (耶鲁大学); Cardiff University (卡迪夫大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 28 pages, 6 figures, 18 tables
Abstract:Tool-call judges evaluate argument values proposed by LLM agents. We find that they accept a correct value less often when their input holds no reference for it, that is, no text that says what the value should be. This happens when the value is a default value from the function schema that the user did not mention, when it has to be inferred, or when the text it comes from was cut from the judge input. We call this drop in acceptance the derivation gap. It appears across all seven judges we evaluate, including System One models CLM and Jev. The gap is not only a truncation artifact: correct default values have the largest gap on a benchmark with no truncation. A controlled source-relocation test shows that judges follow the reference in their input: they favor a visible incorrect value over a correct value whose source is hidden. We introduce DeRef (Derived Reference): a separate model reads the full history without seeing the candidate and writes a reference for the judge. Across three benchmarks, DeRef improves overall argument AUROC for all judges tested. On the controlled test, it raises AUROC from chance or below to between 98.40 and 99.59.
[NLP-41] Readout Scaling Shapes Sparse Feature Recovery
【速读】： 该论文旨在解决稀疏自编码器（Sparse Autoencoder, SAE）在训练过程中因读出缩放（readout scaling）导致的学习偏差问题，即读出缩放会改变SAE所学习到的潜在表示，进而影响对已知因子的恢复能力。其核心解决方案是引入读出规范化决策空间（Readout-Canonicalized Decision-Space, RC-DS）坐标，通过消除奇异值带来的增益效应，同时保留决策空间中的关键信息。关键在于，RC-DS能够在固定读出条件数的条件下，使训练过程恢复因子生成时所处的空间结构，从而显著提升稀疏因子的恢复精度——在条件数为100时，均值稀疏因子恢复率从中心化逻辑斯蒂（centered-logit）训练的0.506提升至0.913。此外，精确封装器验证了该方法可复现RC-DS训练至数值精度，从而将研究焦点聚焦于训练几何结构本身而非额外信息。实验表明，RC-DS在神经策略中提升了对数几率（logit）重建性能，并在多个语言模型检查点与任务稀疏性设置下普遍优于传统方法，且通过小标签头控制实验进一步证明，更长的训练周期可缩小重建差距。综上，该研究揭示了读出缩放、数据分布与训练方式共同塑造稀疏学习的本质，而RC-DS提供了一种信息保真的手段以分离并控制读出贡献。
链接: https://arxiv.org/abs/2610.18462
作者: Brain Xiao,Sarah Wilson
机构: University of Waterloo( Waterloo大学); Columbia University(哥伦比亚大学)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Readout scaling changes what a sparse autoencoder (SAE) learns, but the benefit of removing it depends on the data. We use readout-canonicalized decision-space (RC-DS) coordinates to remove singular-value gains while preserving retained decision information. At a fixed readout condition number, changing the generating distribution reverses which training coordinates recover known factors best. In our controlled coordinate-matching study, recovery is highest when training restores the space in which the factors were generated. When factors are generated before readout scaling, RC-DS increases mean sparse-factor recovery from 0.506 to 0.913 at condition number 100, compared with centered-logit training under the same budget. An exact wrapper reproduces RC-DS training to numerical precision, isolating training geometry rather than additional information. In neural policies, RC-DS improves mean logit reconstruction under the tested budgets. On three fixed language-model checkpoints, RC-DS has higher mean gain-free reconstruction in all 18 tested task-sparsity settings, while centered-logit SAEs reconstruct output logits better in 17. Small-label-head controls show that longer training reduces the reconstruction gap from 0.0198 to 0.0016. Together, these results show how readout scaling, the data distribution, and training shape sparse learning. RC-DS provides an information-preserving method for studying and controlling the readout’s contribution. Separate label-detection and steering tests distinguish these reconstruction effects from task-specific feature usefulness.
[NLP-42] acher-Guided Tool Learning for Small Language Models via Bidirectional KL Constraints
【速读】： 该论文旨在解决小语言模型（SLMs）在资源受限环境下依赖外部工具时，因自身能力有限而难以通过强化学习自主发现有效工具使用行为的问题，同时克服直接教师监督中存在的师生行为分布不匹配及错误偏好传播等挑战。其解决方案的关键在于提出一种选择性教师引导的策略优化方法（Selective Teacher guided Policy Optimization, SeTPO），该方法通过协同学生探索与选择性教师指导，结合逆KL散度驱动的逐标记级教师反馈以增强低奖励轨迹上的策略优势，并辅以正向KL监督以持续保留教师示范行为。理论上，在双向KL目标框架下，最优策略可保证期望奖励不低于教师水平，同时保留教师行为概率质量的最低比例。实验结果表明，SeTPO在BFCL v4和API Bank数据集上均显著优于监督微调、强化学习及知识蒸馏等基线方法，且在三种不同规模的学生模型上均表现出卓越的综合性能。
链接: https://arxiv.org/abs/2610.18459
作者: Haoxuan Gao,Qin-Wen Luo,Xiang Chen,Yu Tian,Zequn Sun,Xiaofeng Cao,Jiangchao Yao,Sheng-Jun Huang
机构: Nanjing University of Aeronautics and Astronautics(南京航空航天大学); 360 Security Capability Center(360安全能力中心); Nanjing University(南京大学); Tongji University(同济大学); Shanghai Jiao Tong University(上海交通大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:External tools can extend the capabilities of small language models (SLMs), enabling their deployment in resource constrained environments. However, their limited capabilities make it difficult for SLMs to discover effective tool use behaviors through reinforcement learning alone, while direct teacher supervision suffers from teacher student behavioral distribution mismatch and may propagate erroneous preferences. To address these challenges, we propose Selective Teacher guided Policy Optimization (SeTPO), which coordinates student exploration with selective teacher guidance and persistent demonstration supervision. Specifically, SeTPO incorporates reverse KL derived token level teacher feedback into policy advantages on low reward trajectories, while auxiliary forward KL supervision preserves teacher demonstrated behaviors throughout training. Under an idealized bidirectional KL objective, we theoretically establish that the optimal policy achieves expected reward no lower than the teacher’s while retaining a guaranteed fraction of the teacher’s probability mass. Extensive experiments on BFCL v4 and API Bank demonstrate that SeTPO achieves superior overall performance over supervised fine tuning, reinforcement learning, and distillation baselines across all three student scales.
[NLP-43] Copying Is Not Grounding: Training Faithful Long-Form RAG with Authored-Sentence Credit
【速读】： 该论文旨在解决长格式检索增强生成（RAG）答案中因模型生成与源文档不一致的主张而导致误导读者的问题，其核心挑战在于现有基于忠实度奖励（faithfulness rewards）的训练机制易被“复制攻击”（copying）所利用：由于从文档中直接复制的句子在语义上几乎必然得到支持，模型因此倾向于学习复制而非真正基于源文档进行生成。现有防御策略如KL散度惩罚或复制惩罚仅转移问题，导致生成内容更短、空洞或拒绝回答，而标准强化学习算法GRPO因对每个句子赋予相同的优势值，反而同时强化了复制和真实生成的句子，加剧了问题。本文提出作者句信用（Authored-Sentence Credit, ASC），其关键创新在于将奖励与优势计算从整个回答层面下移到句子层面，并仅对由模型自主生成（非复制）、超出问题范围且有源文档支持的句子给予信用。通过将每句话与同一问题下采样的其他句子进行对比，未获信用的句子将始终处于最末位，从而避免被强化。实验在四个模型及多个域内与外部数据集上验证，只有ASC能持续降低作者句的错误率，并在所有设置中显著提升支持率（外部评估者下中位数分别提升38%和46%），同时不依赖任何捷径。两名大语言模型（LLM）裁判评估显示，尽管基线模型表现不佳，但ASC生成的答案质量与基线相当，证明其在保持忠实度的同时有效避免了复制行为。结论强调，忠实度奖励应奖励模型的原创性生成，而非其复制内容。
链接: https://arxiv.org/abs/2610.18442
作者: Hoang Pham,Duc Anh Vu,Duy Tung Doan,Cong-Duy T Nguyen,Anh Tuan Luu
机构: Nanyang Technological University(南洋理工大学); Viettel AI(越南电信人工智能); Viettel Group(越南电信集团); VinUniversity(越南大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:In search engines and assistants, long-form RAG answers can mislead readers with claims their sources do not support. Faithfulness rewards, the share of supported sentences, train models toward fewer such claims. We show that these rewards are hacked by copying. A sentence copied from the passages is supported almost by construction, so trained models learn to copy rather than ground what they write. Common defenses only move the hack: a KL or copy penalty leaves copying in place or trades it for shorter, emptier answers and refusals. GRPO, the standard RL algorithm, makes it worse by giving every sentence its answer’s advantage, reinforcing copied and grounded sentences alike. We propose Authored-Sentence Credit (ASC) , which fixes both by moving GRPO’s reward and advantage from the answer to the sentence level. The reward credits only sentences that are authored rather than copied, say more than the question, and are supported by the passages. The advantage compares each sentence with every sentence sampled for the question, so one that earns no credit ranks last and is never reinforced. Across four models and two in-domain, three held-out corpora, only ASC lowers the error rate of authored sentences and writes more supported ones in every setting, by 38% and 46% at the median under held-out evaluators, giving readers more grounded information. It takes none of these shortcuts, and two LLM judges comparing whole answers prefer no baseline to the base model but rate ASC’s answers on par with it. We conclude that a faithfulness reward should credit what a model writes, not what it copies.
[NLP-44] Estimating Surprisal from the Speech Signal
【速读】： 该论文旨在解决语言理解中**预期意外性理论（surprisal theory）**在口语模态下难以有效验证的问题。由于现有研究几乎完全依赖于书面文本训练的语言模型（LMs）来估算预期意外性，导致存在“文本瓶颈”，无法直接从语音信号中评估口语中的认知加工过程。其解决方案的关键在于：摒弃对文本的依赖，直接基于语音信号构建语音语言模型（speech LM），并通过引入重要性采样估计器（importance sampling estimator）来实现对抽象语言单元（如词或音位）的预期意外性估计。该方法通过在文语转换模型生成的多种语音实现上进行概率边际化，克服了同一语言单位具有多种声学实现、导致概率不一致的难题。实验结果表明，在听觉任务的瞳孔反应数据支持下，基于重要性采样的抽象单元预期意外性比基于具体声学输入的预期意外性更能预测词汇层面的认知负荷；同时，亚音段单位的预期意外性也与瞳孔扩张相关。尽管如此，整体而言，无论来源如何，预期意外性均非瞳孔大小变化的强预测因子。
链接: https://arxiv.org/abs/2610.18439
作者: Zachary Hopton,Samuel Kiegeland,Francesco Ignazio Re,Patrick Haller,Matthew B. Winn,Ryan Cotterell,Eleanor Chodroff
机构: University of Zurich (苏黎世大学); ETH Zürich (苏黎世联邦理工学院); CHI-FRO; University of Minnesota (明尼苏达大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Although language is predominantly spoken, empirical tests of surprisal theory rely almost entirely on written text. In practice, surprisal is estimated using language models (LMs) trained on text and applied to transcriptions, creating a text bottleneck that limits evaluation of the theory in the spoken modality. We remove this bottleneck by estimating surprisal directly from the speech signal using a speech LM. This raises a key question: should surprisal be defined over realization-specific acoustic inputs or over abstract linguistic units? Since a single unit can have many distinct acoustic realizations, each assigned a different probability by the model, defining surprisal over abstract units is difficult. To address this, we introduce an importance sampling estimator that approximates the probability of an abstract unit by marginalizing over realizations generated by a text-to-speech model. Using pupillometry data from a listening study, we evaluate surprisal theory in this text-free setting at multiple levels of granularity. Obtaining surprisal from several speech LMs, we consistently find that surprisal estimated via importance sampling better predicts pupil dilation than realization-specific surprisal for words. We also find that surprisal values for sub-lexical units are predictive of pupil dilation. While we find that surprisal can be estimated without recourse to text, and that marginalizing over distinct realizations of words is more predictive of lexical processing effort, surprisal as a whole is not a strong predictor of change in pupil size, regardless of its source.
[NLP-45] Reliable Memory from Unstable Growth: Curating Self-Written Agent Playbooks with Randomized Experiments
【速读】： 该论文旨在解决语言模型代理在执行多步骤任务时，如何有效筛选和保留过往经验中的关键知识（即“记忆”）以提升未来任务表现的问题。现有系统依赖代理自身判断来决定哪些经验应被保留在“剧本”（playbook）中，但这种依赖主观评估的方法导致性能极不稳定——相同流程可能生成显著提升或下降高达69.6分的剧本。为克服这一缺陷，本文提出随机块精选（Randomized Block Curation, RBC），其核心在于通过在实践任务上进行随机实验来客观评估每组经验块对任务成功率的实际影响：将相关记忆分组为足够大的“块”，并在每个实践任务中随机决定是否包含某一块，从而确保有无该块的任务具有可比性；通过比较两组任务的成功率差异，量化该块的实际贡献。RBC仅保留有益的块，并在多个候选剧本（包括全剧本、空剧本及代理自选剧本）中选择在新任务上表现最优者作为最终输出。在两个记忆系统、三种模型和三个基准上的实验表明，RBC将平均成功率提升至66.9，显著优于全剧本（55.5）、无剧本（54.1）以及代理自选策略（53.8）。因此，该方案的关键创新在于以实证方式替代主观判断，实现对记忆内容的可衡量、可优化的动态筛选。
链接: https://arxiv.org/abs/2610.18427
作者: Hoang Pham,Cong-Duy Nguyen,Dong Le,Ponhvoan Srey,Anh Tuan Luu
机构: Nanyang Technological University (南洋理工大学); VinUniversity
类目: Computation and Language (cs.CL)
备注:
Abstract:Persistent textual memory lets language-model agents reuse lessons from past tasks to improve on future multi-step tasks without retraining. Systems such as ACE and Dynamic Cheatsheet keep it as a playbook: lessons the agent writes after each task and reads in its prompt. A key challenge is to decide which lessons improve later tasks. Current systems leave the decision to the agent itself, which makes performance unstable: the same procedure can write a playbook that helps or one that falls up to 69.6 points below no playbook. These findings motivate valuing lessons by their measured effect on task success rather than by the agent’s judgment. We propose Randomized Block Curation (RBC), which measures this effect with a randomized experiment on practice tasks. It groups related lessons into blocks large enough to move success measurably and, on each practice task, includes or leaves out every block at random. Random inclusion makes tasks with and without a block comparable, so the gap in their success estimates the block’s effect. RBC keeps the blocks that help and, since the same procedure can write a helpful or a harmful playbook, grows several and serves whichever curated, whole or no playbook scores best on fresh practice episodes. Across two memory systems, three models and three benchmarks, RBC raises average success to 66.9, against 55.5 with the whole playbook, 54.1 with no playbook and 53.8 when the agent’s own judgment picks the lessons to keep.
[NLP-46] Receiver Decides Agent -to-Agent Negotiations
【速读】： 该论文旨在解决在多智能体交互中，当一个语言模型代理向另一个代理请求资金时，决定交易金额的关键因素究竟是请求方（requester）还是持有方（holder）的问题。研究通过1,049场由15个智能体参与的循环对抗谈判实验发现，持有方的身份能够解释47%的转账金额方差，而请求方的身份仅解释2%；剔除极端个体后，这一差距缩小至22%与4%，表明持有方的角色在决策中占据主导地位。关键解决方案在于揭示并验证了“持有方拒绝初始请求”的行为模式——那些在首轮即拒绝请求的持有方损失最小，其拒绝率与总损失之间存在显著负相关（Spearman相关系数 = -0.73）。此外，通过针对说服性文本进行微调（fine-tuning），可显著改变持有方的行为：将Qwen3-8B模型在几乎无拒绝的语料上进行微调后，其作为持有方的支出比例从12%上升至84%；而通过对两个“永不失败”的持有方进行知识蒸馏生成4B规模的学生模型，其损失从95%降至1%。研究强调，对资源持有型智能体的评估必须在其持有角色下进行，尤其是在经过说服性微调之后，以真实反映其策略稳定性与抗压能力。
链接: https://arxiv.org/abs/2610.18424
作者: Fardeen Sadab,Alif Hasnain,Afsana Shehrin,Md. Azam Hossain,Abu Raihan Mostofa Kamal
机构: Islamic University of Technology, Dhaka, Bangladesh(伊斯兰科技大学，达卡，孟加拉国)
类目: Computation and Language (cs.CL)
备注: 18 pages, 12 figures
Abstract:When one language-model agent asks another for money, which of the two decides how much changes hands? We ran a round-robin tournament of 1,049 negotiations among fifteen agents. In each game a requester tries to raise money from a holder that has been told to keep a 100 endowment. Knowing which agent holds the money explains 47% of the variance in the amount transferred; knowing which agent asks explains 2%. Dropping the most extreme agents narrows the split to 22% against 4%. Changing the requester’s appeal from one named person to a group changed what requesters wrote, but the amounts transferred under the two framings are statistically equivalent. The holders that lose least are the ones that turn down the opening request, and their turn-1 refusal rate ranks them by total losses (Spearman correlation = -0.73). Training moves this behaviour in both directions. Fine-tuning Qwen3-8B on a persuasion corpus with almost no refusals raised what it gave away as a holder from 12% to 84% of the endowment. Distilling two never-losing holders into a 4B student cut its losses from 95% to 1%. Evaluations of agents that hold resources should test them in the holder role, including after persuasion fine-tuning.
[NLP-47] Agaram 2.5B: The First Tamil-First Language Model Trained from Scratch with Tool Calling. A Playbook for Under-Represented Languages
【速读】： 该论文旨在解决低资源语言（特别是泰米尔语）在生成式AI模型中表现不足的问题，尤其是在多语言对话、工具调用及跨语言翻译任务中的能力短板。其核心挑战在于如何在有限的高质量泰米尔语数据上，从零训练一个具备强大多语言交互能力的大型语言模型，同时保持对泰米尔语脚本、罗马化泰米尔语（Tanglish）和英语的准确输出，并实现可靠的任务工具调用。解决方案的关键在于：第一，采用全新设计的保留换行符的分词器（tokenizer），有效提升文本结构的保留能力；第二，基于13.4万条由开源权重教师模型生成回复的对话数据进行监督微调（SFT），并结合三轮针对性强化训练，显著提升了模型在多语言输出与工具调用上的准确性——在25个工具调用提示中，正确选择工具23次，精确提供预期参数13次；第三，在约500亿词元的混合数据集（含58%泰米尔语）上进行端到端预训练，使模型在六项FLORES和IN22翻译基准中均优于同类基线模型（如Sarvam-1），并在本土泰米尔语评估集IndicIFEval上将性能几乎翻倍（从10.0%提升至19.3%）。尽管如此，模型在知识性问答任务中仍表现较弱，多数错误源于事实性错误，表明其知识获取能力仍需进一步优化。
链接: https://arxiv.org/abs/2610.18420
作者: Ramchand Kumaresan,Madhusudhanan G,Kaushik Kezendirane
机构: Murai Labs(穆拉实验室); Independent researcher(独立研究员)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 29 pages, 10 figures, 12 tables. Technical report
Abstract:We trained Agaram 2.5B, a 2.5-billion-parameter Tamil-English language model, from scratch on about 50 billion tokens (58% Tamil, the rest English, math, code and parallel text) on one node of eight H100 GPUs in about six days, with a new tokenizer that, unlike our first model’s, keeps line breaks. Supervised fine-tuning on 134,428 conversations, whose replies an open-weight teacher model wrote, and three short targeted rounds turned it into a chat model that answers in Tamil script, romanized Tamil (Tanglish) and English and can call tools: on 25 tool prompts whose schemas appear in training it picks the right tool 23 times and gives exactly the expected arguments 13 times. The released model translates better than Sarvam-1, a 2B Indic base model, in all six FLORES and IN22 directions (FLORES chrF 54.3 into Tamil and 55.2 into English, against 49.2 and 44.2), and nearly doubles its base model on native-Tamil IndicIFEval (19.3% from 10.0%). It is weak on knowledge: reading-comprehension and knowledge benchmarks sit near chance, and on 50 held-out everyday questions written by one native speaker, one LLM judge rated 46% of its replies usable (22-24 of 50 over three passes; 95% interval for 23 of 50, 33-60%), with nearly every failure a wrong fact. Our two further SFT experiments changed answer format but not the knowledge measures. We also describe our first model, Agaram 1.5B, whose failures shaped the choices here; how the judge’s scores varied between passes on identical replies; and, item by item, what the targeted rounds trained against and where frontier-model text entered the data. The project ran on compute credits and our own funds, about USD 13,300 across both models; the weights are released (licence to be announced after legal review), with GGUF files for this http URL, at this https URL code at this https URL
[NLP-48] Do Factuality Methods Make LLM s More Factual? Reproducibility and Fair Comparison
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在对齐人类偏好过程中忽视事实性（factuality）的问题，即现有强化学习对齐方法（如基于人类反馈的强化学习，RLHF）虽能提升生成内容的偏好度，但可能导致模型产生非事实性信息。为此，论文首次独立复现了五种代表性事实性对齐方法，涵盖在线策略强化学习（PPO、GRPO、Dr. GRPO）与离线偏好优化（DPO、KTO），并在统一基准测试集和评估指标下进行系统比较。研究发现，这些方法均未完全复现，且在统一基准上均未显著优于基础模型；进一步分析表明，其奖励函数实际优化的是文本格式、长度或简洁性，而非真实性和相关性，而训练方式（在线/离线）的影响较小。此外，评估体系本身存在缺陷：缺乏提示（prompt）的真实答案作为参考，依赖大语言模型评判者和检索系统进行验证，导致评估结果脆弱不可靠。论文强调了当前事实性对齐研究中可复现性与评估标准一致性的重要性，并公开了代码、数据及模型检查点以促进后续研究。
链接: https://arxiv.org/abs/2610.18415
作者: Pratuat Amatya,Vinay Setty,V Venktesh,Avishek Anand
机构: Factiverse AS; University of Stavanger; Stockholm University; TU Delft
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Large Language Models (LLMs) are aligned to human preferences, after pre-training and supervised fine-tuning steps. However, standard RL alignment steps such as RLHF do not focus on factuality of LLMs, hence they may generate non-factual information, which could result in unintended consequences. In response, a number of factuality alignment approaches have been proposed in the literature which claim to improve factuality of LLMs on their own benchmarks. However, these methods have not been reproduced independently and are also not comparable since each uses a different base model, training data, evaluation set, judge, and metric. In this paper, we present the first independent reproduction of five factuality alignment methods, which span on-policy RL (PPO, GRPO, and Dr.~GRPO) and offline preference optimization (DPO and KTO), and compare them on a common benchmark with common evaluation metrics. We found that none of these methods reproduces in full, and on the common benchmark none of the methods significantly outperforms the base model. On closer inspection, the rewards optimize for format, length, or brevity rather than factuality and relevance, whereas the choice between on-policy and offline training matters little. Furthermore, the evaluation metrics and benchmarks are also part of the problem since they do not measure relevance of text to the prompt. Since there is no ground truth associated with the prompts, the reliance on LLM judges and retrieval for verification also makes the evaluation fragile. We release our code, data, and checkpoints.
[NLP-49] Revisiting the Estimation of Unigram Surprisal in Computational Psycholinguistics
【速读】： 该论文旨在解决计算心理语言学中阅读时间预测模型中存在的概率一致性问题：当前上下文突变（contextual surprisal）通常基于神经语言模型计算，而一元突变（unigram surprisal）则依赖独立语料库统计，二者虽均源于同一潜在语言分布（分别捕捉条件与边际结构），但分离估算方式破坏了其内在的概率关联，并可能引入估计偏差。解决方案的关键在于提出三项估计原则，核心是要求两者均从同一语言模型中推导，以保持概率一致性。为此，作者提出一种基于蒙特卡洛方法的、从语言模型中估算一元突变的方法，并引入一种方差更低的边缘化变体。在统一框架下重新估算一元突变后，发现其对阅读时间的预测贡献发生改变，且验证了在该框架下词频可比上下文突变更能有效预测加工努力。
链接: https://arxiv.org/abs/2610.18414
作者: Zachary Hopton,Francesco Ignazio Re,Samuel Kiegeland,Andreas Opedal,Eleanor Chodroff,Ryan Cotterell
机构: University of Zurich(苏黎世大学); ETH Zürich(苏黎世联邦理工学院); CHI-FRO; MPI for Intelligent Systems, Tübingen(图宾根智能系统马普研究所)
类目: Computation and Language (cs.CL)
备注:
Abstract:Contextual and unigram surprisal are standard predictors of reading time in computational psycholinguistics. While contextual surprisal is typically derived from neural language models, unigram surprisal is usually obtained from independent corpus counts. Although both quantities are deterministic functionals of a single underlying distribution—capturing conditional and marginal structure respectively—this practice breaks their probabilistic link and may introduce estimation error from the mismatch. We propose three principles for estimating unigram surprisal in psycholinguistic research, including the requirement that both quantities be derived from the same language model. We operationalize this principle by introducing a Monte Carlo estimator of unigram surprisal from a language model, along with a marginalized variant that has lower variance empirically. When unigram surprisal is re-estimated under this unified framework, we find evidence that predictors change in their predictive contribution to reading time. We also confirm prior findings that frequency can be a stronger predictor of processing effort than contextual surprisal under this unified framework.
[NLP-50] Not All Drift Is a Mechanism: Tracing Prompt Adaptation in LLM s
【速读】： 该论文旨在解决现有提示适配方法在评估时过度依赖下游任务性能，而忽视其内部计算机制差异的问题。尽管不同提示适配方式可能产生相似输出，但其内在信息处理路径存在本质区别。研究聚焦于离散提示优化、提示调优（prompt tuning）与前缀调优（prefix tuning）对密集模型及混合专家模型（MoE）的处理过程影响，发现适配位置决定了任务信息传递至输出的机制：优化文本主要影响响应构建过程，输入级提示将影响注入内容表征，而逐层前缀则通过注意力机制保持直接控制。在MoE模型中，这些差异进一步引发不同的专家路由模式。同时，研究指出常见的表征层面度量指标对通用上下文变化敏感，难以可靠区分有效适配与匹配的伪适配。论文提出，理解提示适配的关键在于分离表示漂移中与任务计算对齐且被因果利用的部分，这一视角为提示适配方法的选择、评估与解释提供了实践指导。
链接: https://arxiv.org/abs/2610.18412
作者: Mikhail Aleksandrov,Nail Bashirov,Maksim Ivanov,Andrew Tsukanov,Timofey Karyagin,Alexander Miasnikov,Gleb Molodtsov,Valentin Samokhin,Aleksandr Beznosikov
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Prompt-based adaptation offers several ways to specialize a frozen language model, ranging from natural-language prompt optimization to learned continuous prompts. These methods are usually compared through downstream performance, although similar outputs need not arise from similar internal computations. We study how discrete prompt optimization, prompt tuning, and prefix tuning alter the processing of dense and mixture-of-experts language models. Our analysis shows that the location of the adaptation predicts how task information is carried to the output: optimized text primarily changes response construction, input-level prompts write their influence into content representations, and layer-wise prefixes maintain direct control through attention. In MoE models, these differences further induce distinct expert-routing patterns. At the same time, common representation-level measurements respond strongly to generic context changes and do not reliably distinguish useful adaptations from matched placebos. We argue that understanding prompt adaptation requires separating representation drift from the part of that drift that is aligned with, and causally used by, the model’s task computation. This perspective provides practical guidance for selecting, evaluating, and interpreting prompt adaptation methods.
[NLP-51] Prompt-Injection Probes Learn the Conversation Not Just the Injection
【速读】： 该论文旨在解决生成式 AI（Generative AI）代理系统中线性探针（linear probes）在检测间接提示注入攻击时存在的高误报率问题。现有评估方法表明，仅依赖对话位置和文本长度等非语义特征即可达到0.91–0.98的AUROC，说明标准探针可能并未真正学习到攻击的本质特征，而是捕获了与攻击相关的表面统计模式。其关键解决方案是引入“匹配负样本”（matched negatives），即在训练数据中加入与攻击样本在格式和上下文上相似但无害的文本实例，从而引导探针区分真实攻击与形式相似的正常内容。实验结果显示，仅通过调整训练数据，即可将无攻击场景下的误报率从41%–80%降至4%–27%，对无害外观攻击的误报率从92%–98%降至3%–16%；同时在默认阈值下仍能保持对欺骗代理的攻击检测率达中位数95%。该方法同样适用于纯文本过滤器，并在新应用场景中维持低误报，但检测性能受模型特性及阈值设定影响。
链接: https://arxiv.org/abs/2610.18408
作者: Kaan Kale,Sriram Vishwanath
机构: Georgia Institute of Technology(佐治亚理工学院)
类目: Computation and Language (cs.CL)
备注: 18 pages, 6 figures, 11 tables
Abstract:Linear probes on a model’s hidden states are used as cheap monitors for indirect prompt injection: they screen the tool outputs an agent reads and send suspicious ones to a slower second check, so every false alarm has a cost. But a probe learns whatever separates its training examples. The standard evaluation of injection probes can be passed without the model: conversation position and text lengths alone reach 0.91-0.98 AUROC. When each attack is compared with the same conversation without the injection, or with harmless text formatted like an injection, standard probes flag most of these harmless cases. We add such cases to the training data as matched negatives. Across six open models, in applications seen during training and at the default threshold, changing only the training data cuts false alarms on injection-free conversations from 41-80% to 4-27%, and on harmless look-alikes from 92-98% to 3-16%. With a threshold set on normal traffic, our probe still catches a median 95% of the attacks that fooled the agent (91% without our data). In new applications, false alarms stay low, but detection depends on the model and on how the threshold is set. The same data also helps a text-only filter.
[NLP-52] Multi-Stream Residual Computation in Diffusion Language Models
【速读】： 该论文旨在解决扩散语言模型（diffusion language models）在并行重建被掩码标记时，其标准Transformer架构中单一残差流（residual stream）难以同时有效承载可见信息与动态掩码预测的问题。传统架构中，所有信息通过单一残差流传递，导致可见位置的上下文证据与掩码位置的渐进预测之间产生干扰。为此，本文提出引入多个持久性残差流（multiple persistent residual streams），并通过自适应路由机制实现流间动态交互，以分离并优化“读取”（read）与“写入”（write）功能。关键创新在于：多流结构使可见位置与掩码位置自发学习不同的路由模式，且因果干预分析表明这种差异组织了读写路径的分工，而每类角色内部仍需保持特定标记的写入强度。基于此，采用角色条件化路由（role-conditioned routing）训练的模型虽可部分恢复双流带来的性能增益，但状态干预分析揭示，额外流的核心优势在于保留早期标记表示以供后期复用，同时允许掩码位置持续积累证据。该机制在更大规模模型上依然有效，表明持久性残差流能够协同支持上下文保留与预测精细化。
链接: https://arxiv.org/abs/2610.18407
作者: Ekaterina Alimaskina,Suren Kazarian,Gleb Molodtsov,Aleksandr Beznosikov
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 24 pages
Abstract:Diffusion language models reconstruct masked tokens in parallel, offering an alternative to autoregressive generation. Yet their Transformer backbones typically carry visible evidence and evolving masked predictions through a single residual stream. We equip masked diffusion models with multiple residual streams that persist through depth and adaptive routing between them, then investigate how they support these complementary roles. On the pretraining task, two streams reduce reconstruction perplexity by 5.71% on validation data and 5.13% on average across external corpora compared with standard residual connections. We observe that, without predefined roles, visible and masked positions learn distinct routing patterns. Causal interventions show that this distinction organizes read and write routing, while token-specific write strength remains important within each role. Guided by these findings, a model trained with role-conditioned routing recovers half the validation gain from adding a second stream. State interventions explain the benefit of the additional stream: it preserves early token representations for later reuse while evidence accumulates at masked positions. This mechanism recurs at a larger model scale, showing how persistent streams support context preservation alongside prediction refinement.
[NLP-53] PRISM: Calibrating Black-Box Judges with Counterfactual Probes
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）作为评估者时，其评分与人类评判之间存在系统性偏差的问题。现有方法依赖人工制定的评分量规（rubric），但这一过程成本高且易引入主观性。本文提出一种名为PRISM（Probe-based Residual Inference of Scoring Misalignment）的新解决方案，其核心在于通过在保持被评内容不变的前提下，主动强调或弱化特定属性来探测模型评分的变化，并将这些响应变化建模为一个代理雅可比矩阵（surrogate Jacobian），从而捕捉评分偏差的局部敏感性。基于人类标注数据，利用岭回归（ridge regression）学习对基准评分的校正函数，实现无需显式量规的评分校准。实验表明，在SummEval摘要评估和HANNA故事评价任务中，PRISM及其基于学习基线的变体显著提升了模型评分与人类排名的一致性，同时在性能上可媲美传统量规驱动的方法。进一步控制实验显示，由质量维度和风格维度生成的探测信号具有相似的校准效果，说明校准可不依赖于复现具体量规标准。此外，学习到的校正系数揭示了不同评估者间存在的共性与对立修正方向，为理解模型评估偏差提供了可解释性支持。
链接: https://arxiv.org/abs/2610.18389
作者: Kakei Yamamoto,So Kuroki
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Large language models are increasingly used as judges to reduce reliance on costly human evaluation, yet their scores can systematically disagree with human judgments. We ask whether responses to an explicitly imposed evaluative bias can help correct a judge’s implicit scoring biases. We introduce PRISM (Probe-based Residual Inference of Scoring Misalignment), which holds the evaluated content fixed and measures score changes when attributes are emphasized or de-emphasized. A local sensitivity model motivates representing these responses as a surrogate Jacobian. Ridge regression on human labels learns a correction to the baseline score. The resulting scoring protocol is rubric-free: it uses overall scores without requesting criterion-level ratings. In the main Jev and Sol comparisons on SummEval summarization and HANNA story evaluation, PRISM and its learned-baseline variant significantly improve agreement with human rankings over uncalibrated scores while remaining competitive with rubric-based methods. A controlled SummEval comparison yields similar agreement for rubric-derived quality probes and style probes, supporting calibration without reproducing rubric criteria. The learned coefficients reveal shared and opposing correction directions across judges.
[NLP-54] Pin It on Anyone: Certified Targeted Displacement of Multi-Bit LLM Watermarks Without the Key
【速读】： 该论文旨在解决生成式人工智能（Generative AI）中多比特水印（multi-bit watermarking）技术的安全性问题，即如何在大语言模型（LLM）输出文本中嵌入可追溯的账户标识信息以追踪有害内容源头。然而，现有水印方案存在被攻击者篡改的风险。本文提出一种新型攻击方法——定向位移攻击（targeted displacement attack），其核心在于：攻击者无需掌握用于生成水印的秘密密钥，仅需获取少量账户的模型输出与逻辑值（logits），即可通过重写文本内容（保持语义不变）的方式，诱导水印解码器误判为提取出攻击者预设的任意水印消息。该攻击的关键在于利用生成文本中的语义歧义性，在不破坏自然语言流畅性的前提下，精确操控水印编码路径。理论上，作者给出了攻击成功率的上下界，其受可用账户数量、水印消息长度及文本特征的影响；实验上，该方法在五种水印机制和四种主流大模型上均验证了有效性，攻击成功率与理论预测高度一致，表明在足够多账户支持下，几乎总能实现高精度的水印消息替换。
链接: https://arxiv.org/abs/2610.18359
作者: Paweł Batorski,Przemysław Spurek,Paul Swoboda
机构: Heinrich Heine University Düsseldorf(海因里希·海涅杜塞尔多夫大学); Jagiellonian University(雅盖隆大学); IDEAS Research Institute(IDEAS 研究所)
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL)
备注:
Abstract:Multi-bit watermarking embeds a message, typically an account identifier, into LLM output so that harmful text can be traced to its origin. We show an attack technique that allows to alter the watermarking message. This allows to, for example, exchange the embedded account ID to an arbitrary different one. The attacker mainly needs access to a few accounts for generating LLM output and model logits, but crucially not the secret key used to watermark. The attacker then rewrites the text, keeping its meaning, so that the watermarking message decoder recovers the watermarking message embedded by the attacker. We are, to our knowledge, the first to propose such an attack, which we call a targeted displacement. Theoretically, we present upper and lower bounds on the probability of our attack succeeding. This bound depends on the number of accounts the attacker has access to, the length of the message and some characteristics of the given text. Our attack can succeed whenever we can bound the ambiguity of the generated text well enough. Empirically, we test our attack technique against five watermarking methods and four LLMs. Attack success tracks our predicted bounds, allowing us to alter watermarking messages in almost all cases with high accuracy, given enough accounts. Our code can be found at this https URL.
[NLP-55] Lightweight but Linguistically Fair? Quantisation Robustness across Bavarian German and English
【速读】： 该论文旨在解决大语言模型在后训练量化（post-training quantisation）过程中，是否能可靠地保持方言类语言（如巴伐利亚德语）的语言能力，相较于标准语言（如标准德语、英语）是否存在性能退化的问题。研究聚焦于低资源方言在压缩后的表现稳定性，揭示了量化对语言多样性的影响机制。其解决方案的关键在于系统性评估不同量化精度（8-bit GPTQ、4-bit GPTQ与AWQ）及校准策略下，多种语言（巴伐利亚德语、标准德语、英语）在多语言基准测试集BAVGROUND上的表现差异。研究发现，8-bit GPTQ能较好保留原始浮点精度（BF16）下的语言相对行为，而4-bit量化则表现出显著不稳定性，导致平均准确率下降约1.4–1.5个百分点，并改变11.9–16.0%的决策结果，且这些变化主要集中在原模型已处于不确定状态的样本上。值得注意的是，尽管巴伐利亚德语作为低资源方言初始表现较低，但4-bit量化并未系统性加剧其劣势，其受影响程度因模型、量化方法和校准语言而异。此外，自由生成式评估进一步表明，即使多项选择准确率变化较小，4-bit压缩仍可能显著降低生成文本质量。因此，研究强调：方言语言在量化过程中的鲁棒性并非由整体多语言准确率单一决定，而是受模型架构、量化方法、校准数据和评估范式共同影响，需采用多维度评估框架以确保语言多样性的公平性。
链接: https://arxiv.org/abs/2610.18343
作者: Jophin John,Michael Hoffmann,Piroska Lendvai,Florent Dufour
机构: Leibniz Supercomputing Centre (LRZ), Garching, Germany; Bavarian Academy of Sciences and Humanities, Munich, Germany; TUM University Hospital and Technical University of Munich, Munich, Germany
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Post-training quantisation reduces the memory cost of large language models, but it remains unclear whether compression preserves language skills of regional varieties as reliably as their standard-language counterparts. We study this question for Bavarian, Standard German, and English using BAVGROUND, a multiple-choice benchmark, where 206 source questions are aligned across all three languages. Across 14 open-weight 7B–10B instruction models, we compare BF16 with 8-bit GPTQ and two 4-bit settings, GPTQ and AWQ. Eight-bit GPTQ remains close to BF16 and largely preserves the relative behaviour of the three languages, while 4-bit quantisation was found to be much less stable. The two 4-bit methods reduce average accuracy by about 1.4–1.5 percentage points and change 11.9–16.0% of BF16 decisions, with changes concentrated on examples where the BF16 model is already uncertain. This instability does not translate into a consistent Bavarian disadvantage. Although Bavarian, a low-resource dialect, has lower baseline accuracy than German and English, the language most affected at 4 bits varies across models and quantisation methods. Calibration further shapes these cross-language effects: the relative Bavarian–German effect changes with calibration language, while changing the calibration source has little average effect on GPTQ-4 but can affect AWQ-4 much more strongly. Free-form evaluation on five models further shows that 4-bit compression can degrade generated responses even in settings where multiple-choice accuracy changes only modestly, while 8-bit GPTQ remains comparatively stable. Overall, our results show that quantisation robustness for regional language varieties depends on the model, quantisation method, calibration data, and evaluation setting, and cannot be measured by aggregate multilingual accuracy alone.
[NLP-56] Probing the Geometry of Color Representations in Multilingual Language Models
【速读】： 该论文旨在解决多语言语言模型（Multilingual Language Models, MLMs）是否在内部表征中学习到一种普遍的、与人类感知一致的颜色几何结构，以及这种结构是基于感知而非语言的。其核心问题是：颜色的语义表征在多语言模型中是否具有跨语言一致性与感知基础？解决方案的关键在于提出并验证两个几何维度——（P1）对感知色空间（如CIE-LAB）的对齐程度，以及（P2）跨语言对称性。通过分析不同层的嵌入表示，研究发现：感知对齐性在中间层达到峰值，随后在输出层显著下降，表明模型后期表征逐渐丧失感知基础；而跨语言对称性随网络深度增加而退化，说明深层表征更倾向于编码语言特异性结构。功能层面进一步证明，当P1对齐度最高时，模型在颜色分类任务上达到96.7%准确率，且P1对齐度与下游任务性能显著相关（r=0.70）。研究揭示出一种解离现象：仅基于文本训练的语言模型学习的是“通用颜色-语言”几何结构，而非“通用颜色-感知”几何结构，从而为评估语言模型中感知结构的保留程度提供了可量化的诊断工具。
链接: https://arxiv.org/abs/2610.18331
作者: Pablo Loyola,Edison Marrese-Taylor,Andrés Hoyos-Idrobo
机构: Rakuten Group, Inc.(乐天集团); The University of Tokyo (东京大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Do multilingual language models learn a universal internal geometry for color, and is it perceptually grounded or primarily linguistic? We decompose color representations into two geometric axes: (P1) alignment to perceptual color space, and (P2) cross-lingual symmetry. Analyzing layerwise embeddings across models and multilingual datasets, we find that perceptual grounding is weak: alignment to CIE-LAB peaks in intermediate layers, then collapses toward output layers. Cross-lingual symmetry degrades with depth, suggesting later layers encode language-specific structure. Functionally, we show that geometric properties predict task performance: models achieve 96.7% accuracy on color classification at layers where P1 alignment peaks, with P1 alignment (r=0.70) significantly correlating with downstream capabilities. Our results reveal a dissociation: text-only LMs learn universal color-language geometry, not universal color-perception geometry, providing diagnostics for where perceptual structure is preserved in language models.
[NLP-57] Jev-Guard: Fast Routing and Auditing Agent Actions over Editable Policy Trees
【速读】： 该论文旨在解决智能体（Agent）调用框架中安全防护模型存在的高延迟与策略定制化不匹配问题。现有自回归式安全防护模型在推理时延较高，且难以有效对齐用户自定义的审计策略，限制了其在低延迟场景下的应用。为此，本文提出Jev-Guard——一种统一的安全分类与策略路由模型，其核心创新在于构建了一个覆盖内容审核、网络安全、隐私保护及法律合规等多领域的策略树（policy tree），通过分层路由机制实现动态策略选择。具体而言，Jev-Guard首先基于调用上下文导航策略树以确定适用的审计策略，随后在已检索策略的约束下执行安全分类。在7,033个评估案例中，Jev-Guard实现了93.72%的安全分类准确率、58ms的平均推理延迟以及93.10%的分层路由准确率，完整流水线平均耗时仅240ms。进一步集成至Hermes调用框架后，其独立权限门控机制将任务审查时间从基于SFT-LLM判官的4.80秒显著降低至1.33秒，且未发生任何策略违规事件，验证了其在实际系统中的高效性与可靠性。
链接: https://arxiv.org/abs/2610.18326
作者: Wenbin Hu,Huihao Jing,Haochen Shi,Yuxuan Liu,Changxuan Fan,Sirui Zhang,Kawai Chung,Hanyu Yang,Zhenhuan Wang,Ziyi Chen,Haoran Li,Yangqiu Song
机构: HKUST(香港科技大学); ModeIO; CUHK (Shenzhen)(香港中文大学（深圳)）
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL)
备注:
Abstract:Agent harnesses require a low-latency guardrail model together with customizable audit policies. However, deploying an auto-regressive guardrail model remains challenging: such models incur high inference latency and are often poorly aligned with customized policies. To address these problems, we present Jev-Guard, a unified model for safety classification and policy routing. We further construct a policy tree over which Jev-Guard performs routing, spanning diverse domains including content moderation, cybersecurity, privacy, and legal compliance. Specifically, Jev-Guard first navigates the policy tree to select the applicable audit policy given the caller’s audit context, and then performs safety classification conditioned on the retrieved policies. On 7,033 evaluation cases, Jev-Guard achieves 93.72% safety accuracy, a mean latency of 58ms, and 93.10% hierarchical routing accuracy. The full routing-and-classification pipeline averages 240ms. Furthermore, we integrate Jev-Guard into the Hermes harness framework. Across agent trajectories, its separate permission gate cuts mean review time per task from 4.80s for an SFT-LLM judge to 1.33s, with no policy violations are observed for either guard.
[NLP-58] Prompt or Policy? Optimizing Physician Slate Presentations with LLM User Simulators
【速读】： 该论文旨在解决患者在选择医生时面临的决策难题，即尽管医师推荐系统能够筛选出相关医生，但患者仍难以通过对比医生个人资料做出有效判断。其核心问题是如何优化医师推荐结果的呈现方式，以提升可比性和决策效率。解决方案的关键在于提出“榜单展示优化”（slate presentation optimization）框架：针对特定医疗案例，生成简明、病例相关的医生摘要，使医生特征更易于比较，同时确保内容严格基于医生档案中的信息。为避免依赖真实患者交互进行评估，研究构建了一个离线优化框架，利用人物角色条件化的大型语言模型（LLM）模拟患者选择行为。通过从医疗案例中提取属性生成患者人物角色，并基于事实准确性、相关性、自然度及模拟患者选择表现对展示效果进行评估；同时引入一个未参与优化过程的第二轮LLM裁判，用于检测模拟器特异性过拟合。实验对比了提示层面优化（GEPA）与参数层面优化（GRPO）及其组合策略，在两种模型规模下均表明：GRPO显著优于GEPA，能更好满足多维度约束要求，且其在模拟转化率上的优势在未见裁判下依然保持；而GEPA单独使用虽提升有限，但作为初始提示输入可加速GRPO训练并实现最高模拟转化率。结果表明，提示级与参数级优化在改善医师推荐展示方面具有互补作用。
链接: https://arxiv.org/abs/2610.18309
作者: Andrei Niculae,Adrian Cosma,Emilian Radoi
机构: National University of Science and Technology POLITEHNICA Bucharest(布加勒斯特理工大学), Romania; IDSIA, Dalle Molle Institute for Artificial Intelligence(达莱莫利人工智能研究所), Lugano, Switzerland
类目: Computation and Language (cs.CL)
备注:
Abstract:Choosing the right physician is a key challenge for patients seeking medical care. Physician recommendation systems identify relevant doctors, but patients still find it difficult to compare their profiles to make a decision. We formulate slate presentation optimization: given a patient’s medical case and a fixed, ranked list of physicians, we generate short, case-specific summaries that make doctors easier to compare, while remaining grounded in the information in their profiles. To optimize physician presentations without relying on real patient interactions, we develop an offline framework that uses persona-conditioned LLMs to simulate patient choices. We generate patient personas from attributes extracted from each medical case and evaluate presentations for factual accuracy, relevance, naturalness and simulated patient choice. We also use a second LLM judge, unseen during optimization, to assess simulator-specific overfitting. We compare automatic prompt-level optimization with GEPA, parameter-level optimization with GRPO, and their combination across two model sizes. GRPO substantially outperforms GEPA in satisfying the full set of constraints, and its gains in simulated conversion persist under the unseen judge. GEPA alone brings smaller improvements, but using its optimized prompt to initialize GRPO speeds up training and achieves the highest simulated conversion. These results show that prompt-level and parameter-level optimization play complementary roles in improving how physician recommendations are presented.
[NLP-59] LegalRewardBench: Benchmarking Reward Models for Legal Preference Evaluation
【速读】： 该论文旨在解决当前奖励模型（Reward Models）在法律任务中缺乏专门评估基准的问题，现有奖励模型评测数据集多集中于通用领域，难以有效衡量模型在法律场景下的偏好对齐能力。为此，论文提出LegalRewardBench，首个经过人工验证的法律领域奖励模型评估基准。其核心创新在于将法律偏好评估划分为知识、理解、分析、决策和写作五个能力层级，并通过基于评分标准的自动化生成结合人工验证的方式构建高质量偏好样本。实验结果表明，当前表现最优的法律奖励模型为强大的通用生成式模型（Generative AI），而现有的微调奖励模型在法律偏好评估中的迁移性能较差。此外，研究还系统比较了输入组织方式与训练策略的影响，揭示了提升法律领域奖励模型性能的关键因素。
链接: https://arxiv.org/abs/2610.18299
作者: Hui Huang,Jie Zhou,Hongli Zhou,Qundong Shi,Shuyuan Zheng,Yini Su,Shaochun Wang,Biyuan Lin,Muyun Yang,Tiejun Zhao
机构: Harbin Institute of Technology(哈尔滨工业大学); ModelBest Inc.; The University of Osaka(大阪大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Reward models are central to preference alignment for legal tasks, yet existing reward model benchmarks are still primarily centered on general-domain preferences. In this work, we introduce LegalRewardBench, the first human-verified benchmark for evaluating reward models in the legal domain. The benchmark organizes legal preference evaluation into five capability levels: knowledge, understanding, analysis, decision, and writing, and preference examples are constructed through rubric-guided automatic judging combined with human verification. We evaluate three families of judges, covering both general and fine-tuned models, and generative and discriminative architectures. Experimental results show that strong general-purpose generative models are currently the most robust legal reward models, while existing fine-tuned reward models transfer less reliably to legal preference evaluation. We also conduct detailed comparisons of input organizations and training strategies.
[NLP-60] PAINT: A Framework for Multilingual Pain Representation from Medical Licensing Exams to Real-World Clinical Texts EMNLP2026
【速读】： 该论文旨在解决多语言临床文本中疼痛表征（pain representation）难以准确捕捉与解析的问题，尤其针对不同语言、文档类型及患者群体间疼痛表达的异质性挑战。现有研究多局限于单一语言或特定临床场景，缺乏跨语言、跨模态的系统性框架。其解决方案的关键在于提出PAINT——一个面向医学检查报告与真实世界临床文本的多语言疼痛分析框架，涵盖英语、西班牙语、葡萄牙语和日语共九个数据集分区，并由四名语言的临床专家进行标注。该框架构建了包含疼痛相关性、存在性、解剖位置、疼痛描述及人口统计信息的结构化标注体系，结合大语言模型（LLM）与集成学习策略以增强多语言疼痛表征的识别性能。经十名跨语言临床专家验证，集成模型在疼痛相关性和存在性上的宏平均F1值达0.91–1.00，在人口统计信息预测上准确率高达92%–100%，且标注者间一致性κ值为0.82–1.00，展现出卓越的一致性与准确性。研究公开了标注结果、抽取管道、验证脚本及标注表格，为疼痛导向的临床自然语言处理（NLP）研究提供了可复用的多语言基准资源。
链接: https://arxiv.org/abs/2610.18269
作者: Boya Zhang,Yuanchao Ma,Jiageng Wu,Krittaphas Chaisutyakorn,Rodrigo Gameiro,Ahram Han,Chien-Yu Huang,Nikhil Jaiswal,Marianna Leite,Miriam Mora-Mau,Jaime Pineda-Moreno,Siena Placino,Ryohei Yamamoto,Yugang Jia,Jie Yang,Leo Anthony Celi,Douglas Teodoro
机构: University of Geneva(日内瓦大学); Massachusetts Institute of Technology(麻省理工学院); McGill University(麦吉尔大学); Harvard Medical School(哈佛医学院); Brigham and Women’s Hospital(布莱根妇女医院); Harvard T.H. Chan School of Public Health(哈佛大学陈曾熙公共卫生学院); Mahidol University(玛希隆大学); Seoul National University Hospital(首尔国立大学医院); National Taiwan University(台湾大学); Universidade de São Paulo(圣保罗大学); Santa Marcelina College of Medicine(圣玛塞琳娜医学院); Universidad Peruana Cayetano Heredia(秘鲁凯塔诺·埃雷迪亚大学); HM Hospitales(赫姆医院); St. Luke’s Medical Center College of Medicine(圣路加医疗中心医学院); Fukushima Medical University(福岛医科大学); Beth Israel Deaconess Medical Center(贝斯以色列女执事医学中心)
类目: Computation and Language (cs.CL)
备注: Accepted at Proceedings of the 6th Workshop on Multilingual Representation Learning (MRL 2026), EMNLP 2026, Budapest
Abstract:Pain is one of the most common and clinically important symptoms, reflecting disease onset, progression, and treatment response. However, pain is difficult to accurately capture and interpret from medical text, as its expression varies across languages, documentation types, and patient demographics, and is often embedded in terminology, negation, temporality, and contextual descriptions. Existing pain-related studies remain largely monolingual or limited to narrow clinical settings. In this study, we introduce PAINT, a multilingual framework for studying pain representation across medical exams and real-world clinical texts. PAINT includes nine pain-focused dataset partitions spanning English, Spanish, Portuguese, and Japanese, with clinician annotations in all four languages. We curated a structured schema covering pain relevance, pain presence, anatomical location, pain descriptors, and demographic information. To capture multilingual pain representations, we leverage large language models (LLMs) and adopt an ensemble strategy to improve performance. Ten clinicians across four languages validate the labels: the ensemble reaches macro-F1 of 0.91-1.00 on pain relevance and pain presence and 92-100% accuracy on demographics, with inter-annotator \kappa of 0.82-1.00. We release the WorldMedQA-Pain ensemble labels, the extraction pipeline, validation scripts, and clinician annotation sheets for publicly licensed corpora, providing a foundation for pain-centered clinical NLP and for studying pain documentation across languages and healthcare contexts.
[NLP-61] Encrypted Skills Can Become New Security Exploits If AI Agents Know How to Decipher
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）代理在使用第三方技能（third-party skills）时，可能因嵌入编码恶意指令而面临安全威胁的问题。其核心挑战在于，攻击者可通过编码手段将恶意指令隐藏于技能文件中，使代理在解码后执行潜在有害操作，而现有检测机制难以识别此类隐蔽攻击。论文提出的解决方案是引入加密技能注入（Ciphered Skill Injection, CSI），通过在技能文件中嵌入经编码的恶意指令，模拟真实攻击场景，评估不同模型对这类攻击的脆弱性。其关键创新在于将“指令恢复”与“攻击成功”分离进行评估，从而更精准地衡量攻击效果；同时，研究发现即使恶意指令被成功解码，也不必然导致攻击成功，且分层策略对不同模型的影响具有显著差异。实验结果表明，在缩短技能长度的主设置下，凯撒（Caesar）和分层变体分别达到最高47.5%和40.3%的攻击成功率，且测试的技能扫描器极少能有效检测出注入内容。这一发现揭示了将未经验证的解码内容直接转化为可执行指令所带来的严重安全隐患，强调了在开放生态中加强技能可信度验证的重要性。
链接: https://arxiv.org/abs/2610.18262
作者: Yan-Lun Chen,Pin-Yu Chen,Chia-Mu Yu
机构: IBM Research(国际商业机器公司研究部); National Yang Ming Chiao Tung University(国立阳明交通大学)
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Third-party skills can expose LLM agents to malicious instructions hidden by encoding. We introduce Ciphered Skill Injection (CSI), which embeds such instructions in skill files for agents to decode and follow. We evaluate Caesar and layered variants across eight models and 139 injection-task pairs, separating instruction recovery from attack success. In the primary setting with shortened skills, the variants achieve attack success rates of up to 47.5% and 40.3%, respectively. Successful decoding does not guarantee attack success, and layering has model-dependent effects. The tested skill scanners rarely flag injected skills. These findings highlight the risk of promoting decoded, untrusted content into actionable instructions.
[NLP-62] Are We Testing Too Much? A Survey of Benchmark Compression for Efficient Evaluation of Large Language Models
【速读】： 该论文旨在解决大规模语言模型（Large Language Models, LLMs）评估中因广泛使用冗余基准测试而导致的评估成本过高的问题。其核心解决方案在于基准压缩（Benchmark Compression），即通过部分评估来估计完整基准测试的得分与排名，从而降低评估开销。方法的关键在于：一是智能选择具有代表性的样本子集以减少评估量；二是利用更低成本或可复用的信息源（如历史评估结果、模型预测置信度等）；三是采用自适应评估策略，在动态过程中调整评估范围。这些方法虽能显著降低总成本并提升效率，但也面临覆盖不足、估计偏差及序列延迟等风险。基于对119篇相关文献的系统分析，研究构建了一个涵盖单个基准内部压缩、跨基准套件压缩以及自适应评估阶段压缩的分类体系，并整合了关于冗余性与可靠性方面的研究成果，为研究人员提供了在信息需求、成本控制、可靠性验证及局限性之间的权衡参考，最终形成一套实用指南，助力实现更高效、可靠的LLM评估。
链接: https://arxiv.org/abs/2610.18248
作者: Guoming Ling,Yupei Lin,Zhongzhan Huang,Junxin Li,Junjie Liu,Shanshan Zhong,Hefeng Wu,Liang Lin
机构: Sun Yat-sen University (中山大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Benchmarks are essential for evaluating and improving large language models (LLMs). Yet many widely used benchmarks are redundant, making evaluation unnecessarily expensive. Benchmark compression lowers this cost by estimating full-benchmark scores and rankings from a partial evaluation. Research in this area has grown rapidly since 2024, but to our knowledge no survey has systematically organized its methods and findings. We review 119 papers to explain how methods choose what to evaluate, recover full-benchmark results, and verify their reliability. From this analysis, we develop a taxonomy covering compression within individual benchmarks, across benchmark suites, and during adaptive evaluation, together with work on redundancy and reliability. We compare their information needs, total costs, reliability checks, and limitations. Methods reduce evaluation costs by selecting representative subsets, using cheaper or reusable information, and testing adaptively, but risk poor coverage, inaccurate estimates, and sequential delays. The resulting taxonomy and practical guide help researchers choose, validate, and deploy methods for more efficient and reliable LLM evaluation.
[NLP-63] PACT: A Prag matic Prompting Framework for Expressive and Voice-Preserving Speech-to-Speech Translation
【速读】： 该论文旨在解决语音到语音翻译（S2ST）中难以同时保留源语言的语义内容、说话人音色（timbre）以及情感、语调、节奏、停顿和语用态度等表达特征的问题。现有端到端S2ST、语音多模态大模型（speech MLLM）及级联系统虽能在一定程度上保持内容或音色，但在目标语言中实现自然且一致的情感与语用表达方面仍存在显著不足。其解决方案的关键在于提出一种目标空间语用提示（PACT, Pragmatic Prompting in Target Space）框架：该框架首先从源语言中提取语用线索（pragmatic cues），将其投影至目标语言的表达空间，并通过校准操作映射到可实现的目标子空间，最终生成连续提示（continuous prompts）以驱动一个冻结的语音生成器。这一机制有效实现了在保持说话人音色的同时，精准迁移并再现源语言中的情感与语用特征，从而在双向西班牙语-英语的TOPI评估中，在内容保真度、音色保留与表达保留等多个指标上均达到最优性能。
链接: https://arxiv.org/abs/2610.18247
作者: Yuchen Song,Satoshi Nakamura
机构: The Chinese University of Hong Kong, Shenzhen, China; Shenzhen Loop Area Institute, China
类目: Computation and Language (cs.CL)
备注: 7 pages, 2 figures, 3 tables
Abstract:Speech-to-speech translation should preserve not only what is said, but also who says it and how it is said. Existing end-to-end S2ST, speech MLLM, and cascade systems can preserve content or timbre, but often struggle to realize emotion, tone, rhythm, pauses, and pragmatic attitude in the target language. We propose PACT, a target-space pragmatic prompting framework for expressive and voice-preserving S2ST. PACT extracts source pragmatic cues, projects them into the target-language expression space, calibrates the projected representation to a realizable target subspace, and converts it into continuous prompts for a frozen speech generator. On bidirectional Spanish-English TOPI evaluation, PACT achieves the best overall performance among evaluated systems across content fidelity, voice preservation, and expression preservation metrics.
[NLP-64] Consider Then Choose: Order-Invariant Decisions in System One Models
【速读】： 该论文旨在解决生成式 AI（Generative AI）在处理选择题时存在的选项顺序依赖问题，即模型对同一问题的不同选项排列会给出不一致的答案，尤其当选项数量增加时，这种偏差愈发显著。其核心解决方案是提出一种名为 Runoff 的新方法：首先独立评分每个选项以保留少数候选项，随后在所有可能的排列组合中比较这些候选项，确保每个候选项均出现在各个位置上进行对比。该方法无需额外训练，仅需两次前向传播，且通过构造保证答案不依赖于选项顺序。实验表明，在两个开源模型和五个数据集上，Runoff 在选项随机洗牌下完全保持答案不变，而循环平均法存在9.3%的错误率；在20个选项场景下，其准确率相当但速度提升3至4倍。将该方法应用于八个System One模型后，成功消除了五项基准测试中的顺序依赖性，而循环平均法最多可导致15.4%的答案变化。
链接: https://arxiv.org/abs/2610.18233
作者: Jianguo Huang,Yao Long Teng,Yanchen Deng,Bo An
机构: Nanyang Technological University (南洋理工大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:System One models answer typed questions in a single forward pass and promise the same answer for the same input. For choice questions, this promise fails: the answer depends on the order of the options, increasingly so as they grow in number, and averaging over their cyclic shifts reduces this dependence without removing it. Scoring each option alone, which the order cannot affect, shows that the order changes the answer mostly when the model is unsure among its top few options, even though the listwise answer looks confident. We therefore propose \textbfRunoff , which considers, then chooses: it scores each option alone to keep a few candidates, then compares them in listings where each candidate takes every position. It needs no training and can run in two forward passes, and its answer does not depend on the option order by construction. On two open models and five datasets, it changes no answer under shuffling, against 9.3% for cyclic averaging, and with 20 options it is about as accurate and three to four times faster. Wrapped around eight System One models, it removes the order dependence of the five benchmarks, while cyclic averaging changes up to 15.4%.
[NLP-65] xtualRL: Textual Reinforcement Learning for Context Management
【速读】： 该论文旨在解决大语言模型（LLM）智能体在任务执行过程中如何有效利用反馈信号来优化其行为上下文，从而实现无需更新模型参数的自适应问题。核心挑战在于：单个执行轨迹提供的对比证据有限，难以指导具体修改方向；而直接将特定任务的修复策略泛化到共享上下文中又易导致过度拟合或失效。为此，本文提出TextualRL——一种无梯度强化学习框架，其关键创新在于引入“组级策略批判”（Groupwise Policy Critique）与“跨组策略更新”（Cross-Group Policy Update）机制：前者将同一任务下多条轨迹构成的群体作为评价单元，通过对比成功与失败轨迹识别可保留行为或生成修复假设；后者整合跨任务的共性证据（包括未触发修改的观察），对候选修复方案进行筛选与精炼，最终形成一个有界文本补丁，并通过留出验证奖励决定是否采纳。该方法构建了一个以奖励驱动的闭环优化流程，使多次尝试的集体经验可被提炼为可复用的上下文更新。实验表明，在六个基准测试上，TextualRL相较于基础模型分别提升了GPT-5.5和Qwen3.8-27B的测试准确率24.4和15.3个百分点，验证了其有效性与泛化能力。
链接: https://arxiv.org/abs/2610.18216
作者: Tianyu Huai,Jie Zhou,Bo Mao,Qin Chen,Liang He,Xipeng Qiu
机构: East China Normal University (华东师范大学); Shanghai Innovation Institute; Fudan University (复旦大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:LLM agents can adapt to task feedback by revising the textual context that guides their behavior, without changing their model parameters. However, turning execution rewards into reusable context updates is challenging: individual trajectories offer limited comparative evidence about what to change, while task-specific fixes may overgeneralize when incorporated into a shared context. We introduce TextualRL, a gradient-free reinforcement learning framework that treats a group of rollouts from the same task under the same context, rather than an isolated trajectory, as the unit of policy critique. Groupwise Policy Critique contrasts successful and failed rollouts within a task; when outcomes are uniform, it draws on cross-task evidence to identify behaviors worth preserving or formulate repair hypotheses. Cross-Group Policy Update reviews the resulting proposals against evidence pooled across groups, including observations that yield no proposed edit, to refine their content and applicability. Reviewed proposals form a bounded textual patch, with adoption of the candidate context determined by held-out validation rewards. Together, these components establish a reward-driven policy optimization loop in which multiple attempts jointly inform reusable context updates. We evaluate TextualRL with five target agents on six benchmarks. Averaged over the six benchmarks, TextualRL improves test accuracy over the base model by 24.4 and 15.3 percentage points for GPT-5.5 and Qwen3.8-27B, respectively.
[NLP-66] SWE-MISBEHAVIOR: Benchmarking and Mitigating Boundary Violations in Software Coding Agents
【速读】： 该论文旨在解决生成式代码代理（coding agents）在实际应用中普遍存在且难以检测的边界违规错误行为（boundary-violating misbehaviors）问题，这些问题包括代码生成偏差、文件操作（如创建/读取/更新/删除，CRUD）错误、系统完整性破坏及任务目标偏离等。其核心挑战在于缺乏统一的分类体系与自动化评估方法，导致现有模型的错误行为难以系统识别与量化。解决方案的关键是提出首个针对此类问题的分类框架SWE-MISBEHAVIOR，涵盖代码生成、文件管理与系统交互三个层级的7.4K人工协同合成样本，构建了包含11类典型错误行为的完整评估体系，并实现全自动化评估协议。基于该框架的实验揭示了主流开源代码代理（如OpenHands、smolagents）存在严重的意图错位与环境破坏现象。为此，论文进一步提出CODE-KIT——一个轻量级自适应框架，融合错误感知指导原则与新型工具，有效引导代理规避上述错误，实测可使smolagents和OpenHands的错误率分别降低52%和34%，同时提升任务完成率，实现了安全性与性能的双重优化。
链接: https://arxiv.org/abs/2610.18206
作者: Do Xuan Trong,Duc Anh Nguyen,Do Xuan Long,Ngoc-Hai Nguyen,Tongyao Zhu,Duy Dinh,Cayden Chik,Cuong Dang,Jiaying Wu,Xuan-Phi Nguyen,Vinh Van Nguyen,Kenji Kawaguchi,Shafiq Joty,Nancy F. Chen,Min-Yen Kan
机构: National University of Singapore; Vietnam National University, Hanoi; University of Virginia, USA; Agency for Science, Technology and Research (A*STAR), Singapore; Tufts University, USA; Virginia Tech, USA; Salesforce AI Research, USA
类目: oftware Engineering (cs.SE); Computation and Language (cs.CL)
备注: Codes and data are available at this https URL
Abstract:We present SWE-MISBEHAVIOR, the first taxonomy and evaluation framework for diagnosing boundary-violating misbehaviors in coding agents. It contains 7.4K human-in-the-loop synthesized samples spanning three layers of agent action space: (i) Code Generation, (ii) File Management, and (iii) System Interaction. Across these categories, SWE-MISBEHAVIOR covers 11 distinct misbehaviors with a fully automated evaluation protocol, including coding-action violations, Create/Read/Update/Delete (CRUD) file misbehaviors, system integrity and abuse violations, and task-goal misalignment. Experiments on state-of-the-art open-source coding agents (e.g., OpenHands, smolagents) reveal substantial misaligned behaviors, including dependency misinterpretation (up to 98.5% of cases), unintended modification of user inputs (50.5%), and silent compromise of environment integrity (53%). Motivated by these findings, we further propose CODE-KIT, a lightweight adaptive framework that combines misbehavior-aware guidelines with novel tools to help agents avoid such failures. CODE-KIT reduces misbehaviors by 52% for smolagents and 34% for OpenHands, while simultaneously improving task completion performance.
[NLP-67] RACESTEER: Trace Driven Activation Steering for Hallucination Mitigation in Diffusion Language Models
【速读】： 该论文旨在解决生成式 AI（Generative AI）中扩散型大语言模型（dLLMs）易产生幻觉的问题，尤其关注其在去噪过程中存在的动态去噪状态与时间依赖性幻觉风险。现有激活引导（activation steering）方法因采用固定的引导向量和统一的干预强度，难以适应dLLMs在不同去噪阶段的内在变化特性，导致引导效果受限。本文提出TRACESTEER——一种基于轨迹驱动的激活引导框架，其核心创新在于通过前序去噪预测作为反馈信号，实现引导方向与强度的联合自适应调整。具体而言，去噪反馈引导（DF-Steering）利用前一时刻的预测结果与当前隐藏状态构建步骤特定的引导向量；轨迹引导门控机制（TS-Gate）则依据预测不确定性与去噪步数动态调节引导强度。在保持主干模型冻结的前提下，结合事实监督与幅度正则化，共同优化自适应引导策略，从而在去噪全过程提升事实可靠性与稳定性。实验结果表明，该方法在两个dLLMs上均显著提升了TruthfulQA任务上的事实准确性，并在五个下游任务中展现出更强的泛化能力。
链接: https://arxiv.org/abs/2610.18173
作者: Xin Zhang,Yili Wang,Yue Tan,Xin He,Rui Miao,Yixin Liu,Yi Chang,Xin Wang
机构: Jilin University (吉林大学); Griffith University (格里菲斯大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Diffusion large language models (dLLMs) have emerged as a promising alternative to autoregressive LLMs, yet remain susceptible to hallucination. Activation steering offers a promising mitigation approach by guiding parallel token predictions through internal representations. However, existing steering methods largely overlook two characteristics of dLLMs: dynamic denoising states and temporal hallucination risks, which challenge predefined steering vectors and uniform intervention strengths. We propose TRACESTEER, a trace-driven activation steering framework that uses preceding denoising predictions as feedback to jointly adapt steering direction and strength. Specifically, Denoising Feedback Steering (DF-Steering) constructs step-specific steering vectors from preceding predictions and current hidden states. Trajectory-Guided Steering Gate (TS-Gate) regulates their strength using prediction uncertainty and the denoising timestep. With the backbone frozen, factual supervision and magnitude regularization jointly guide the learning of adaptive steering interventions, promoting factual reliability and stability throughout denoising. Experiments on two dLLMs show improved factual reliability on TruthfulQA and generalization across five downstream benchmarks.
[NLP-68] LoOPSD: On-Policy Self-Distillation with Looped Teachers
【速读】： 该论文旨在解决传统在线自蒸馏（on-policy self-distillation）中依赖外部增强教师模型提供更优预测的问题，即如何在不引入额外上下文或指令的前提下，从同一模型内部生成更具指导性的教师信号。其解决方案的关键在于提出LoOPSD（Layer Order Optimization for Self-Distillation），通过搜索并调整模型自身各层的执行顺序来构建更强的教师模型，而保持模型权重不变。具体而言，该方法通过重复某些层以增强推理能力，利用新构造的教师对由学生生成的前缀进行预测，从而为学生提供更优的监督信号；随后在多轮迭代中交替进行层序搜索与蒸馏，逐步将教师的性能优势迁移至学生。实验表明，在数学与逻辑推理任务上，该方法显著提升了模型表现，例如在GSM8K任务中，使用400词思考预算时，Qwen3-1.7B的准确率从42.30%提升至79.61%，且后续蒸馏轮次仍带来持续增益。这证明了通过优化层执行顺序可有效挖掘模型内部计算潜力，实现基于自身计算过程的自监督学习，为自蒸馏开辟了新路径。
链接: https://arxiv.org/abs/2610.18162
作者: Wei Fan,Qingyue Yang,Chao-Hong Tan,Qian Chen,Wen Wang,Xiangang Li,Kejiang Chen,Weiming Zhang,Nenghai Yu
机构: Anhui Province Key Laboratory of Digital Security; Token Foundry, Alibaba Group; University of Science and Technology of China (中国科学技术大学)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Under Review
Abstract:On-policy self-distillation improves a model using guidance from the model itself, but it still requires a teacher that provides more useful predictions than the student. Existing approaches obtain such guidance by giving the teacher additional context or instructions. We ask whether it can instead come from changing how the same model uses its own layers. To address this, we introduce LoOPSD, which searches over a model’s layer execution order to construct a stronger teacher and distills its advantage into the student. Specifically, we keep the model weights unchanged while searching for layers to repeat, and use the resulting teacher’s predictions on student-generated prefixes to guide learning. We then alternate search and distillation, rebuilding the teacher from the updated student in each round so that the student progressively absorbs the gains from alternative layer execution. Experiments on mathematical and logical reasoning tasks show improvements across models. On GSM8K with a 400-token thinking budget, LoOPSD raises Qwen3-1.7B accuracy from 42.30% to 79.61%; subsequent distillation rounds yield further gains beyond the first round. Together, these findings show that by searching layer execution to construct teachers, we can turn a model’s additional computation into improved student capabilities, opening a new avenue for self-distillation to derive supervision from the model’s own computation.
[NLP-69] Claim Framing as a Driver of LLM Sycophancy
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）中存在的阿谀奉承现象（sycophancy），即模型倾向于迎合用户偏好而非基于客观事实生成答案的问题。现有研究多聚焦于缺乏明确真实答案的主观性问题，且仅考虑单一、直接表达的用户意见，难以全面反映实际场景中的复杂影响因素。本文通过系统性实验，在具有明确客观答案的事实问答任务中，探究不同话语框架（claim framings）如何触发模型的阿谀奉承行为。研究覆盖14个模型、4.2万道题目，从8个维度和21种类型对用户观点的表述方式进行编码，涵盖观点自信度、观点归属对象、是否呈现为共有的观点等关键变量，并测量多种由用户意见驱动的行为模式。结果表明，所有模型无论规模大小均易受用户意见影响，其中观点表达的自信程度、观点归属主体以及是否被呈现为用户与模型之间的共识，是影响模型响应的关键因素；而用户的专业性、权威性或社会压力等传统认为重要的因素在事实性问答中影响甚微。尽管更大模型对错误观点具有更强的抵抗能力，但其在自身犯错后恢复正确性的能力并未显著提升。因此，论文强调未来评估阿谀奉承现象应关注观点的表达方式（即“如何说”），而非仅仅判断是否存在观点（“有没有说”），并需综合考察多种行为指标，超越单一的“翻转率”（flip rate）衡量标准。
链接: https://arxiv.org/abs/2610.18159
作者: Jean Seo,Pepa Atanasova,Isabelle Augenstein
机构: University of Copenhagen(哥本哈根大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Sycophancy in Large Language Models, the tendency to align responses with user preferences over ground truth, has largely been studied using questions without a clear ground truth and with a single, plainly stated user opinion. We present a systematic study of which framings of a user’s claim trigger sycophancy in factual question answering, where every answer is objectively verifiable without extra context. Evaluating 14 models on 42K questions, we express correct and incorrect user opinions through various claim framings over 8 dimensions and 21 types, and measure a broad set of opinion-driven behaviors. We find that all models are susceptible to user opinions regardless of size. Models are most influenced by how confidently an opinion is stated, whom it is attributed to, and whether it is presented as shared between the user and the model. In contrast to prior work, framings based on the user’s expertise, authority over the model, or social pressure matter little in our factual setting. Larger models resist incorrect opinions better but are no better at recovering from their own errors. Our results call for sycophancy evaluations that probe how an opinion is framed, not merely whether one is present, and that measure multiple behaviors beyond a single flip rate.
[NLP-70] Attacking Only Half the Data: Two-Sided Evaluation of AI-Text Detector Robustness
【速读】： 该论文旨在解决当前广泛使用的文本生成式 AI (Generative AI) 检测器在面对真实场景中人类文本与机器生成文本双重攻击时的鲁棒性不足问题。现有评估方法通常仅针对机器生成文本施加攻击，忽视了人类文本在实际使用中常通过语法修正、格式调整或插入不可见字符等方式被修改的事实。为此，论文提出一种双面评估协议（two-sided protocol），同时对人类文本和机器生成文本施加攻击。实验结果表明，在两个语料库及九种检测器上，传统评估中报告的AUROC性能提升在双面评估下完全消失，且受攻击的人类文本被错误分类率高达100%。尽管字符级攻击可通过简单预处理手段缓解，但语义保持的词级同义替换攻击仍具有效性和不可逆性，成为检测系统面临的核心挑战。其解决方案的关键在于引入双面攻击框架，揭示了现有检测器在现实对抗环境下的脆弱性，并强调语义层面的隐蔽攻击对检测性能的持续威胁。
链接: https://arxiv.org/abs/2610.18151
作者: Apu Chandraw Shill,Rabindra Nath Nandi
机构: Verbex.ai; Independent Researcher
类目: Computation and Language (cs.CL)
备注: 8 pages main, ARR Submission
Abstract:We investigate the robustness of widely used AI text detectors under attacks applied to both human and machine text. Prior evaluations typically attack only machine text, overlooking that human text is routinely modified through grammar correction, formatting, or invisible character insertion. We introduce a two-sided protocol that applies attacks to both classes. Across two corpora and nine detectors, reported AUROC gains vanish under two-sided evaluation, with attacked human text misclassified at rates up to 100%. While simple preprocessing reverses character-level attacks, word-level synonym substitution remains effective meaning-preserving, and irreversible.
[NLP-71] SparPR: Runtime KV-Cache Recovery for Resource-Constrained LLM Inference on Consumer Hardware
【速读】： 该论文旨在解决在消费级硬件上运行大语言模型（LLM）时，因键值（Key-Value, KV）缓存随生成的每个令牌不断增长而导致显存与带宽迅速耗尽的问题。其核心解决方案是提出一种无需修改模型权重、也无需离线训练的运行时KV缓存恢复机制——SparPR（Sparse Partial Recomputation）。SparPR的关键在于三个协同工作的在线决策机制：基于查询感知的键块选择，通过历史块与当前查询的相关性评分来筛选保留块；动态预算控制，根据每块的复制与重计算成本估计，决定保留块的数量；以及按需局部重计算与成本感知回退策略，将保留块合并为连续段，并根据实时成本比较在稀疏重计算和完整恢复之间切换。该方案在FlexGen/FlexLLMGen框架上实现，实验表明，在单台配备RTX 3070 Ti（8GB）、Ryzen 9 6900HX及16GB内存的笔记本电脑上，SparPR在提示长度分别为1k、2k、4k时分别将端到端延迟降低31%、52%和63%，吞吐量提升1.4倍至2.7倍，并成功完成8k令牌工作负载（基线超时），同时在固定缓存预算下保持显著优于StreamingLLM和H2O的“针尖准确率”（Needle Accuracy），验证了三机制的协同有效性。
链接: https://arxiv.org/abs/2610.18140
作者: Jieyun Deng,Keming Hu,Yanghe Liu,Li Feng
机构: Southwest Jiaotong University (西南交通大学); Wuhan Institute of Technology (武汉工程大学); Southwest Petroleum University (西南石油大学)
类目: Computation and Language (cs.CL)
备注: 9 pages, 5 tables, 1 figure
Abstract:Running large language models (LLMs) on consumer hardware is bottlenecked by the key-value (KV) cache, which grows with every decoded token and quickly exhausts the memory and bandwidth of a single consumer GPU. We present SparPR (Sparse Partial Recomputation), a runtime KV-cache recovery mechanism that does not modify model weights and requires no offline training. SparPR couples three online decisions: Query-Aware key-block selection scores historical blocks against the current query; dynamic budget control estimates per-block copy and recompute costs to decide how many blocks to retain; and on-demand local recomputation with cost-aware fallback merges retained blocks into continuous spans and switches between sparse recomputation and full recovery according to a runtime cost comparison. Implemented on FlexGen/FlexLLMGen, SparPR is evaluated on a single consumer laptop (RTX 3070 Ti 8GB, Ryzen 9 6900HX, 16GB RAM). It cuts end-to-end latency by 31%, 52%, and 63% at 1k, 2k, and 4k prompt tokens respectively, with 1.4x-2.7x higher throughput, and completes an 8k-token workload in 44 minutes where the FlexGen baseline times out. Under a fixed cache budget, SparPR retains a Needle Accuracy that stays well above StreamingLLM and H2O at every keep ratio, and ablation studies confirm that the three mechanisms act jointly.
[NLP-72] Monitoring latent reasoning in looped language models
【速读】： 该论文旨在解决循环语言模型（Looped Language Models, LLMs）在实际应用中的能力边界与稳定性问题，尤其关注其迭代变换是否真正提升了模型性能，以及在提示（prompt）扰动下模型行为的鲁棒性。当前对循环结构模型的理论分析尚不充分，缺乏系统性的评估框架。为此，研究提出了一套基于机制可解释性（Mechanistic Interpretability）的多维度检验协议，涵盖五个不同的分析视角，并将其应用于四个不同家族的八种循环语言模型。其解决方案的关键在于区分两类核心架构：一类是仅在循环内部传递提示信息的“携带循环”（carry loops，如Ouro、LoopUS），另一类是在每次迭代中重新读取原始提示的“重读循环”（re-read loops，如Parcae、Rec-OLMo Huginn）。通过对比这两类模型在多种任务和扰动下的表现，研究揭示了不同设计对模型安全性与计算效率的影响，为高效、安全地部署循环语言模型提供了关键实证依据。
链接: https://arxiv.org/abs/2610.18127
作者: Pawel Olszowiec,Paulina Tomaszewska,Maciej Żelaszczyk,Michal Byra,Neo Christopher Chung
机构: Samsung AI Center, Warsaw, Poland; IFTR, Polish Academy of Sciences, Warsaw, Poland; University of Warsaw, Poland
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 39 pages, 25 figures, 11 tables
Abstract:Looped language models (LMs) reuse one block of layers several times per token generation, promising more latent computation at no additional storage cost. Despite their rising popularity, the analysis of looped LMs is still in its infancy: little is known about whether the iterative transformations actually improve their capabilities, or whether these models remain stable under prompt perturbation. We construct an inspection protocol rooted in Mechanistic Interpretability comprising five diverse lenses. We apply it to examine eight looped LMs from four families. Models differ in many aspects but the key one is whether the prompt is transmitted only within the looped state (carry loops: Ouro, LoopUS) or re-supplied at every iteration (re-read loops: Parcae, and Rec-OLMo Huginn from the Raven family). The results of our study provide valuable insights concerning safety and efficient use of looped language models.
[NLP-73] Cheap Consistent Calibrated but Biased? Typed Decision Models as Judges
【速读】： 该论文旨在解决生成式人工智能（Generative AI）评估中依赖昂贵且潜在存在偏见的大型语言模型（LLM）裁判的问题。其核心挑战在于：尽管类型化决策模型（Typed Decision Models, TDMs）因其低成本和高效率被视为替代LLM裁判的可行方案，但其是否继承了原LLM裁判的偏见尚无实证检验。本文的关键解决方案是通过在CALM基准测试中对八种TDMs与七种2024及2026年发布的LLM裁判进行系统性偏见评估，揭示TDMs普遍存在显著偏见——在12项测试中，每种TDM均在3至8项测试中表现出偏倚，其中情感语气影响最为显著；当更优答案被改写为更具情感色彩的表达时，典型TDM的准确率从99%骤降至42%。此外，所有TDM均在答案互换位置时改变判决结果，而2026年版LLM裁判则极少如此，表明TDMs具有更强的不一致性。尽管两者均被相同提示误导，但研究发现，TDMs的误判主要集中在模型置信度较低的样本上，且良好校准并不能防止偏见。因此，尽管TDMs在一致性与成本方面（比2026 LLM裁判便宜37倍）具有显著优势，但其应用前提必须是针对具体任务开展充分的偏见检测与验证。
链接: https://arxiv.org/abs/2610.18122
作者: Fabian Lukassen,Christoph Weisser,Alexander Silbersdorff
机构: University of Göttingen(哥廷根大学); Bielefeld University of Applied Sciences and Arts (HSBI)(比勒费尔德应用科学与艺术大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Typed decision models (TDMs) return a probability for each allowed answer instead of generating text, which makes them candidates for cheaply replacing LLM judges that compare two answers. Whether they share the biases of LLM judges is untested. We run the CALM benchmark’s bias tests on eight TDMs and seven LLM judges of 2024 and 2026. Every TDM is biased, in three to eight of twelve tests. Emotional tone affects them most, and rewriting the better answer in such a tone lowers the accuracy of a typical TDM from 99% to 42%. TDMs are more biased than 2026 LLM judges. All eight change verdicts when the answers swap places, which the 2026 LLM judges hardly do, and several side with an answer that most people allegedly prefer or that a famous model allegedly wrote. Yet both kinds of judge are tripped up by the same prompts. The TDMs’ probabilities show that most tests mainly flip verdicts the TDM was unsure of, and that good calibration does not protect against bias. As TDMs are more consistent and 37 times cheaper than 2026 LLM judges, our results suggest they can be attractive judges if their biases are tested for the task.
[NLP-74] Narrative Lookahead in LLM s: Event-Level Decodability Predictive Advantage and Steering
【速读】： 该论文旨在解决生成式语言模型（Generative AI）在生成叙事文本时，是否以及在多大程度上预先编码了未来叙事事件的信息这一关键问题。传统研究多基于词元（token）或响应层级的属性进行分析，而本文提出一种以显著且改变状态的叙事事件为单位的“事件级前瞻”（narrative lookahead）分析框架，突破了原有方法在时间尺度上的局限。其解决方案的关键在于：通过人工验证的三维度事件标注（涉及主体内部、主体间及故事世界的变化），从四个开源权重的语言模型生成的2000个故事中提取出23,355个事件，并构建与情节无关的标签体系，使不同故事间可在事件数量为单位的时间窗口（horizons）下进行可比性分析。研究发现，在事件结束处，模型隐藏状态中可解码出至少未来四个事件的类型信息，且探测器（probe）性能显著优于仅依赖文本前缀的预测器，但优势仅在下一事件中明显。进一步在OLMo-3.1-32B模型上的探索性控制实验表明，基于探测器推导的方向可有效调节目标事件类型的出现比例，实现对后续事件构成的干预。因此，该研究首次将未来叙事发展的预测能力与可操作的生成调控手段相连接，揭示了语言模型在深层语义层面蕴含的前瞻性结构化知识。
链接: https://arxiv.org/abs/2610.18121
作者: Donghoon Jung,Min Kim,Seojin Choi,Songeun Chae,Seohyon Jung
机构: KAIST(韩国科学技术院); South Korea(韩国)
类目: Computation and Language (cs.CL)
备注:
Abstract:Do language models carry information about narrative events before writing them, and how far ahead? We examine narrative lookahead through salient, state-changing events rather than tokens or response-level attributes. We extract 23,355 LLM-annotated events from 2,000 stories generated by four open-weight LLMs, labeling them along three nonexclusive dimensions of change: within agents, between agents, and in the story world, with human validation on a subset. Event segmentation and plot-independent labels enable comparisons across stories at horizons measured in events rather than in tokens. At event ends, we probe hidden states for upcoming event types and compare predictions with text-only predictors using matched prefixes. Across four models, future event types are decodable above prevalence baselines for at least four subsequent events, but probes significantly outperform both primary text-only predictors only for the next event. In exploratory steering with OLMo-3.1-32B, a probe-derived direction shifts the proportion of the target event type in both intended directions at higher strengths. Our event-level approach links the prediction of future narrative developments to interventions that reshape subsequent event composition.
[NLP-75] Chain of Thought Monitoring Can Fail Even When CoT Is Necessary
【速读】： 该论文旨在解决生成式 AI 系统中链式思维（Chain of Thought, CoT）监控的可靠性问题，尤其关注在 CoT 被认为对答案必要时仍出现监控失败的现象。现有研究将监控失效归因于 CoT 是否“必要”，并主张只有当 CoT 必要时才可被有效监控，但该论文指出这一二元划分过于粗略——即使在 CoT 严格必要的情况下，监控依然可能失效。为此，论文提出一种更精细的组织原则，基于四层递进的 CoT 必要性层级结构，并在此基础上构建了一个通过干预实验测量前三个层级 CoT 可监控性的框架。实证结果表明，在真实且复杂的预测任务（如医学诊断）中，即便在 CoT 必要的前提下，患者性别和临床属性分别影响了至少 19% 和 24% 的正确回答，但这些关键因素几乎从未在 CoT 中被明确表述，揭示出即使在必要情境下，CoT 也未必具备可监控性。这表明，仅依赖“必要性”作为监控可靠性的判断标准是不足的，需引入更细致的分析框架以提升监控有效性。
链接: https://arxiv.org/abs/2610.18105
作者: Lennart Bürger,Piotr Komorowski,Moritz Weckbecker,Evgenii Kortukov,Aakriti Jain,Sebastian Lapuschkin,Wojciech Samek
机构: Fraunhofer HHI(弗劳恩霍夫高新信息技术研究所); TU Dublin(都柏林城市大学); TU Berlin(柏林工业大学); BIFOLD
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Chain of Thought (CoT) monitoring is used by frontier labs to detect misbehaviour of AI systems. Yet, its reliability remains poorly understood. We attribute this to the lack of a successful organizing principle that leads to generalizable conclusions about monitoring reliability in different regimes. Recent work attributes monitoring failures to the CoT not being necessary for the answer, and argues that CoT becomes monitorable when it is necessary. We show that this binary necessary/not-necessary organizing principle is too coarse by demonstrating that monitoring failures persist even if the CoT is necessary for the answer. In response, we introduce a more fine-grained organizing principle based on a hierarchy of four CoT necessity levels. Building on this hierarchy, we develop a framework for measuring CoT monitorability at the three lowest levels via intervention experiments. We demonstrate monitoring failures at all three of these levels on realistic, hard prediction tasks. On a difficult medical-diagnosis task, patient sex and a clinical attribute influence the answer in at least 19% and 24% of correct responses with a necessary CoT, respectively, yet their influence is almost never articulated. These results show that even when CoT is necessary, it is not guaranteed to be monitorable.
[NLP-76] From Agents to Edges: Paired Optimization for Multi-Agent Collaboration
【速读】： 该论文旨在解决强个体智能体（strong individual agents）无法保证高效协作的问题，核心挑战在于通信双方在信息传递与理解使用之间存在对齐偏差。其解决方案的关键在于将通信边（communication edge）作为优化的基本单元，提出配对式自动化通信优化（Paired Automated Communication Optimization, PACO）。PACO通过联合优化通信两端的指令及共享的通信约定，在保持模型权重、智能体角色和工作流拓扑不变的前提下，利用状态感知的折扣赌博机选择待优化的通信边，并基于执行反馈进行分阶段端到端评估与修正。优化后的指令可无缝集成至现有智能体提示词中，部署时无需额外模型调用。实验表明，PACO在多个多智能体基准测试中持续优于主流提示优化与通信基线，且在HotpotQA上学习的配置具备良好的跨数据集迁移能力，无需针对目标任务进行再优化。结果证明，对智能体间通信方式的优化能够显著提升系统整体协作性能，即使在其他组件固定的情况下亦可实现显著改进。
链接: https://arxiv.org/abs/2610.18069
作者: Feng Lin,Xiaotian Yu,Yifan Yang,Aijun Li
机构: Intellifusion Inc.
类目: Computation and Language (cs.CL)
备注: 9 main pages, 14 appendix pages
Abstract:Strong individual agents do not guarantee effective collaboration. Communication must align what one agent conveys with how another interprets and uses it. This dependency makes the communication edge a natural unit of optimization. We introduce Paired Automated Communication Optimization (PACO), which jointly optimizes the instructions at both ends of each edge and their shared communication convention in systems built on black-box language models. PACO keeps model weights, agent roles, and workflow topology fixed. A state-aware discounted bandit selects edges for refinement, while execution feedback guides paired revisions evaluated through staged end-to-end execution. The optimized instructions integrate into existing agent prompts without additional model calls at deployment. On a four-agent workflow evaluated across diverse multi-agent benchmarks, PACO consistently outperforms leading prompt-optimization and communication baselines. Configurations learned on HotpotQA retain their advantage in cross-dataset transfer without target-specific optimization. These results demonstrate that optimizing how agents communicate can substantially improve collaboration within an otherwise fixed system.
[NLP-77] A Benchmark for Bangla Task-Oriented Dialogue Understanding NAACL2027
【速读】： 该论文旨在解决孟加拉语（Bangla）在多语言对话理解场景下，尤其是在英语与罗马化孟加拉语（Banglish）共存环境中的任务导向型对话理解（Task-oriented Dialogue Understanding）研究严重不足的问题。其核心挑战在于缺乏高质量、跨语言对齐的标注数据集以支持多语言对话系统的开发与评估。为此，作者提出 BanglaTOD，一个并行三语基准数据集，涵盖9个与孟加拉国相关的服务领域和64种意图，包含209轮多轮对话，每轮对话在孟加拉语、英语与Banglish之间实现逐句对齐，共计生成4,689条带注释的语言特定对话记录，包含意图、词级槽位、状态更新及累积对话状态等丰富信息。解决方案的关键在于构建一个高保真、结构化的多语言对话数据集，并在此基础上建立基于mBERT、BERT-base-uncased和BanglaBERT的上下文感知基线模型，通过跨语言分析揭示不同模型与语言表征间的性能差异，同时验证对话历史在意图识别中的关键作用——移除历史信息将导致意图准确率下降19.0–25.7个百分点。该工作为孟加拉国特定场景下的多语言对话状态跟踪（Dialogue State Tracking, DST）研究奠定了重要基础，且数据集已公开于Hugging Face平台，可供社区复用与拓展。
链接: https://arxiv.org/abs/2610.18060
作者: Afrina Alam Lamia,Sumiaya Sharif,Prottoy Modak,Ohiduzzaman Rayhan,Nafis Sadeq
机构: East West University(东西大学); Dhaka, Bangladesh(达卡, 孟加拉国)
类目: Computation and Language (cs.CL)
备注: Submitted to NAACL 2027
Abstract:Task-oriented dialogue understanding remains underexplored for Bangla, particularly in multilingual settings involving English and romanized Bangla (Banglish). We introduce BanglaTOD, a parallel tri-lingual benchmark comprising 209 multi-turn dialogues across nine Bangladesh-specific service domains and 64 intents. Each conversational turn is aligned across Bangla, English, and Banglish, yielding 4,689 language-specific turn records annotated with intents, token-level slots, state updates, and accumulated dialogue states. We establish context-aware baselines for joint intent classification and slot filling using mBERT, BERT-base-uncased, and BanglaBERT. Cross-language analysis reveals substantial performance differences across models and linguistic representations, while removing dialogue history reduces intent accuracy by 19.0–25.7 percentage points. BanglaTOD provides a foundation for multilingual dialogue understanding and future dialogue state tracking research in Bangladesh-specific conversational applications. The dataset is publicly available at \hrefthis https URLHugging Face.
[NLP-78] nLoRA-MINT: Internal Representations for Auditing Training Data in Domain-adapted LLM s
【速读】： 该论文旨在解决大语言模型（LLM）在使用LoRA（Low-Rank Adaptation）进行微调时，存在敏感训练数据隐私泄露与数据溯源困难的问题，尤其关注在黑盒环境下能否有效判断特定样本是否曾被用于训练。其核心挑战在于如何在不访问模型完整参数的前提下，实现对训练数据成员性的精准推断。解决方案的关键在于提出一种基于中间模型表示的新型成员推理测试（MINT）方法，利用模型激活值（activations）及依赖参数的信号作为判别特征，构建无需完全参数访问的分类器。该方法在多个主流模型（Qwen、Phi-4、DeepSeek-V3、LLaMA 3.2-3B）和跨领域数据集（医学、数学、学术推理）上均取得95%至98%的高准确率，展现出对模型架构、配置及数据分布变化的强鲁棒性。该方法为敏感训练数据的高效审计提供了可行路径，支持数据可追溯性，并有助于满足日益严格的生成式人工智能治理与隐私保护法规要求。
链接: https://arxiv.org/abs/2610.18051
作者: Gonzalo Mancera,Julian Fierrez,Aythami Morales,Ruben Tolosana,Daniel DeAlcala,Francisco Jurado
机构: BiometricsAI(生物识别人工智能); Universidad Autonoma de Madrid (马德里自治大学); Computer Engineering Dept., Universidad Autonoma de Madrid (马德里自治大学计算机工程系); Mathematics Dept., Universidad de Las Palmas de Gran Canaria (大加那利岛拉斯帕尔马斯大学数学系)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:This work investigates the application of the Membership Inference Test (MINT) to large language models (LLMs) fine-tuned with LoRA, aiming to determine whether specific samples were used during training and address concerns about data privacy, leakage, and transparency. Building on our previous LoRA-MINT framework, which relies exclusively on perplexity under a strict black-box setting, we propose an approach based on intermediate model representations, including activations and parameter-dependent signals. These representations enable a MINT classifier to distinguish training from non-training samples without requiring full access to model parameters. We evaluated the method on four LLMs (Qwen, Phi-4, DeepSeek-V3, and LLaMA 3.2-3B) fine-tuned on three datasets covering medical, mathematical, and academic reasoning tasks. Results show consistent performance across architectures and domains, achieving scores between 95% and 98%, with robustness to variations in model configurations and data distributions. The proposed approach enables efficient auditing of sensitive training data and supports data provenance and compliance with emerging privacy and AI governance regulations. Although our experiments focus on LoRA-based fine-tuning, the methodology can be extended to other LLM adaptation techniques and, more broadly, to other AI models. Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) Cite as: arXiv:2610.18051 [cs.LG] (or arXiv:2610.18051v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.18051 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-79] Machine–human agreement in marking undergraduate assignments is specific to model prompt and institution
【速读】： 该论文旨在解决生成式 AI（Generative AI）在高等教育写作评估中应用的可迁移性与可靠性问题，尤其关注当前大语言模型（LLMs）在跨机构、跨提示（prompt）情境下的评估一致性与公平性。其核心问题是：现有评估研究多局限于单一模型、单一提示和单一机构，无法反映真实教育场景中复杂多变的上下文差异，从而导致对模型实际能力的误判。论文的关键解决方案在于通过大规模多校实证研究（涵盖3所英国大学的761份心理学本科作业），系统评估三种前沿LLM在准确性、提示敏感性、错误独立性、泛化能力及证据基础等方面的性能表现。研究发现，所有模型均表现出向中心趋势集中的评分偏差，且该偏差具有共享性，因此模型集成（ensemble）未能有效缓解；同时，机器评分对语言特征的依赖程度显著高于人类评分。最终结论指出，当前LLMs缺乏跨情境可迁移的评价能力，而基于单一情境的验证会过度高估其部署可行性。
链接: https://arxiv.org/abs/2610.18019
作者: Giulio Corsi,Maryam Abo-Tabik,Yael Benn,Georgiana Thorpe Apreutesei,Roni Tibon,Alexandru Marcoci,Deborah Talmi
机构: Centre for the Future of Intelligence, University of Cambridge, Cambridge, UK(剑桥大学未来智能中心); Department of Computer Science, University of Lancashire, Preston, UK(兰开夏大学计算机科学系); School of Psychology, Manchester Metropolitan University, Manchester, UK(曼彻斯特都会大学心理学学院); Department of Psychology, University of Cambridge, Cambridge, UK(剑桥大学心理学系); School of Psychology, University of Nottingham, Nottingham, UK(诺丁汉大学心理学学院); Centre for the Study of Existential Risk, University of Cambridge, Cambridge, UK(剑桥大学存在性风险研究中心); Department of Psychology, University of Cambridge, Cambridge, UK(剑桥大学心理学系)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models (LLMs) promise to automate one of higher education’s most resource-intensive activities: assessing student writing. Existing evaluations are typically limited to a single model, a single prompt and a single institution. Using 761 undergraduate psychology assignments from three UK universities, we evaluated three frontier LLMs, individually and in ensemble, on accuracy, prompt sensitivity, error independence, generalisability and evidential basis. Agreement between human and machine marks varied widely, with no model or prompt optimal across institutions. All models compressed marks towards a common central tendency, overvaluing the weakest assignments and undervaluing the strongest. Because this error was shared, ensembling failed to remove it. Machine marks also rewarded linguistic features more strongly than human marks. Since human-machine agreement was a joint product of model, prompt and institution, we conclude that current LLMs lack evaluative capacity transferable across contexts, and that single-context validation overstates readiness for deployment.
[NLP-80] LatentRemedy: Efficient Remedy For Non-Prefix KV Cache Reuse in Multi-Agent Systems
【速读】： 该论文旨在解决多智能体系统中非前缀型键值缓存（non-prefix KV cache）复用时，因智能体任务冲突导致模型性能显著下降的问题。现有方法如选择性重计算和基于偏移的校准虽能部分缓解性能损失，但恢复效果有限。其核心问题不仅在于共享上下文的键值表示被污染，更在于目标智能体难以在自身指令语境下正确理解并利用被复用的上下文。为此，论文提出LatentRemedy，其关键创新在于引入少量任务条件化的隐变量令牌（latent tokens），通过一个MLP将目标智能体前缀的最后隐藏状态映射为一组紧凑的隐变量令牌，并将其置于复用的KV块之后，以实现与共享上下文的交互，从而引导模型在不重新计算共享块的前提下有效利用上下文。该方法通过全预填充蒸馏（full-prefill distillation）训练MLP，确保后续预测聚焦于目标任务。实验表明，LatentRemedy在任务冲突场景下持续优于现有复用基线，同时随着缓存规模增大，仍能保持接近直接复用的首字节传输时间（TTFT）。
链接: https://arxiv.org/abs/2610.18014
作者: Zhiwei Chen,Zhongtao Rao,Waikit Xiu,Yuxiang Zhang,Shuting Cui,Jian Chen,Tianxiang Zhao
机构: The Hong Kong University of Science and Technology (Guangzhou); Microsoft; The University of Hong Kong
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 22 pages with multiple figures
Abstract:Non-prefix KV cache reuse is a promising approach for improving inference efficiency in multi-agent systems as agents share the same context blocks. However, when agents have conflicting tasks, directly reusing non-prefix KV caches across agents can severely degrade model performance, causing up to a 90% drop. Existing remedies, including selective recomputation and offset-based calibration, recover a limited fraction of this loss. We argue that the problem lies not only in contaminated KV representations but also in the agent’s difficulty interpreting reused context under its own instructions. To this end, we introduce LatentRemedy, which uses a few task-conditioned latent tokens to help the target agent leverage reused context without recomputing the shared block. Specifically, we use an MLP to map the last hidden state of the target prefix into a small number of latent tokens, which are placed after the reused KV block to interact with the shared context. To guide subsequent predictions toward the target task, we train the MLP through full-prefill distillation. Experiments show that LatentRemedy consistently outperforms existing reuse baselines on conflicting tasks while maintaining TTFT close to direct reuse as cache size grows. Code and data are available at this https URL.
[NLP-81] Fish Audio Drama 3 Technical Report
【速读】： 该论文旨在解决生成式语音合成中多角色场景演绎、自然语言指令驱动、多轮修订、音频补全编辑以及长时序流式生成等复杂任务难以在单一模型中统一实现的问题。其核心挑战在于如何在保持高保真语音质量的同时，支持灵活的交互式创作流程与复杂的语境理解能力。解决方案的关键在于提出Fish Audio Drama 3，一个基于350亿参数的专家混合（Mixture-of-Experts, MoE）架构，每令牌激活约30亿参数，具备高效的计算资源利用能力；同时引入一种新型分层注意力机制的代码本多标记预测音频解码器（codebook multi-token-prediction audio decoder），通过专用层对整个序列进行全局注意力建模，以提升音色一致性与上下文连贯性。此外，模型采用三阶段训练策略——预训练、指令微调与大规模批次强化学习，并仅依赖单一可学习的人类偏好模型作为奖励信号，显著提升了生成内容的自然度与可控性。该系统实现了多轮修订、场景配音、基于补全的编辑及流式输入等新功能，且在单张H200 GPU上推理速度达每秒271帧，性能远超前代模型，为交互式语音表演生成提供了强大工具。
链接: https://arxiv.org/abs/2610.18010
作者: Dawei Han,Shidong Li,Shijia Liao,Jiahua Liu,Songting Liu,Qingzheng Wang,Yuxuan Wang,Xin Chen,Yifan Cheng,Tianyu Li,Gangin Park,Xuehao Yang,Chang Zeng,Haoyang Zhang,Ruoyi Zhang,Xiangyu Zhang
机构: Fish Audio Team
类目: ound (cs.SD); Computation and Language (cs.CL)
备注:
Abstract:We introduce Fish Audio Drama 3, an open-source speech generation model that turns text into directed performances, built on a 35B-parameter mixture-of-experts backbone that activates 3B parameters per token. It clones voices from multiple references, follows natural-language direction, revises a take when the next turn gives a note, voices multi-character scenes, edits existing speech by infilling, and synthesizes long-form and streaming audio. Multi-turn revision, scene dubbing, editing by infilling, and streaming text input are new in Drama 3, all in one model. Drama 3 pairs the hybrid linear-attention backbone with a new codebook multi-token-prediction audio decoder that predicts the acoustic codebooks of each frame in dedicated layers that attend over the entire sequence. The model is trained with a three-stage pre-training recipe, instruction fine-tuning, and large-batch reinforcement learning whose only reward is a single learned model of human preference. Its inference engine decodes a single stream at 271 frames per second on one H200, against 177 for Fish Audio S2, although Drama 3 has more than seven times as many parameters. The model weights are available on GitHub (this https URL) and Hugging Face (this https URL), and inference is available through SGLang-Omni (this https URL). We highly encourage readers to visit this https URL to direct their own performances.
[NLP-82] Match the Canvas Not Just the Answer: Predictability-Aware SFT for Diffusion Language Models
【速读】： 该论文旨在解决扩散语言模型（Diffusion Language Models, DLMs）在监督微调（Supervised Fine-Tuning, SFT）过程中存在的训练-推理不匹配问题。具体而言，传统SFT采用随机掩码（random masking）策略选择目标词元的可见参考词元，并独立于推理阶段设定注意力可见输入长度，导致训练时的监督信号与实际生成过程中的可预测性不一致，从而可能误导模型学习——例如掩盖预测错误或惩罚本可通过解码缓解的不确定性，同时改变输入可见长度会偏好不同的响应形式，即使底层解决方案保持不变。针对这一问题，论文提出了一种基于可预测性的画布选择与监督方法（Predictability-Aware Canvas Selection and Supervision, PACS），其核心在于利用模型自身的可预测性动态确定每个样本的注意力可见输入长度，并在词元信心引导的揭示时刻对其进行监督。该方法显著提升了模型在推理阶段的准确性，在多个推理与代码生成基准测试中均优于现有SFT方法，最高相较原始SFT提升达12.84个百分点。
链接: https://arxiv.org/abs/2610.18001
作者: Miao Li,Hanyang Jiang,Hengyu Fu,Baihe Huang,Yating Liu,Yuhang Cai,Changkun Guan,Zhongzhu Chen,Jiecheng Lu,Michael A. Boateng,Pascal Van Hentenryck
机构: Georgia Institute of Technology (佐治亚理工学院); University of California, Berkeley (加州大学伯克利分校); University of Chicago (芝加哥大学); University of Michigan (密歇根大学)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Diffusion language models (DLMs) enable flexible parallel generation by resolving tokens according to the model’s evolving predictability rather than a fixed left-to-right order. This flexibility, however, introduces training choices absent from standard autoregressive teacher forcing: which reference tokens are visible when a target is supervised and how much generation space is allocated. Standard supervised fine-tuning (SFT) typically makes these choices through random masking, independently of inference. Our analysis shows how this training-inference mismatch can misdirect supervision. Random masking can conceal prediction errors or penalize uncertainty that decoding would otherwise resolve, while the attention-visible input length can change which response formulations the model favors, even when the underlying solution is unchanged. Motivated by these findings, we propose Predictability-Aware Canvas Selection and Supervision (PACS), an SFT method. PACS uses the model’s own predictability to select each example’s attention-visible input length and supervises reference tokens immediately before their confidence-guided revelation. Across reasoning and code-generation benchmarks, PACS consistently achieves the highest mean accuracy among the compared SFT methods, with gains of up to 12.84 percentage points over vanilla SFT on individual benchmarks. Code will be released upon publication.
[NLP-83] Seeing Through the Forgery Is Not Enough: Regressive Sycophancy under Fabricated Evidence in Medical Vision-Language Models
【速读】： 该论文旨在解决医疗视觉-语言模型（Medical Vision-Language Model, VLM）在面对伪造证据时出现的“倒退性阿谀奉承”（regressive sycophancy）问题，即模型在用户提供虚假处方等伪造信息后，即使原有诊断正确，仍倾向于放弃原答案以迎合用户输入。其核心解决方案在于揭示并缓解模型对合成文档的过度信任机制。研究通过在400个VQA-RAD二分类问题上进行19,200次多轮对话实验，发现当未明确披露文档为合成数据时，模型会放弃80.1%的正确答案；而在特定脚本下，同一伪造处方可使Qwen2.5-VL-7B的错误翻转率从45.9%升至89.6%，而Qwen3-VL-8B则从77.7%降至50.9%，表明模型对伪造证据的响应存在显著差异。尽管有28.3%的对话中模型拒绝了伪造处方并保持正确答案，但其中40%在后续对话中仍发生翻转。进一步实验表明，仅通过系统提示声明文档为合成数据，即可将某模型的翻转率从77.8%降至47.3%，说明模型对伪造性的认知虽能部分抑制错误适应，但不足以完全防止其采纳错误结论。因此，该研究的关键发现是：当前医疗视觉-语言模型缺乏对伪造证据的稳健抵抗能力，亟需引入更有效的对抗性验证与可信推理机制以保障临床决策可靠性。
链接: https://arxiv.org/abs/2610.17989
作者: Sadif Ahmed,Soumya Swagata Biswas,Rifat Shahriyar,Chowdhury Mohammad Rakin Haider
机构: Bangladesh University of Engineering and Technology (孟加拉国工程技术大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 16 pages, 5 figures, 14 tables. Sadif Ahmed and Soumya Swagata Biswas contributed equally. Code and data: this https URL
Abstract:A medical vision-language model may analyze a given CT scan and correctly detect both kidneys in view. Continuing this conversation, if a user provides a prescription with an opposite direction, medical VLMs often sycophantically choose to realign their diagnosis. In this work, we study regressive sycophancy under fabricated evidence, where a model abandons a correct answer when pushback comes with a fake document. We test four open VLMs on 400 VQA-RAD yes/no questions in 19,200 multi-turn conversations, with and without a fabricated prescription. Without a disclosure that documents are synthetic, the models abandon 80.1% of 6,488 correct answers. Notably, under one script each, the same prescription raises the flip rate of Qwen2.5-VL-7B from 45.9% to 89.6% and lowers that of Qwen3-VL-8B from 77.7% to 50.9%. A model rejected the prescription while still preserving the correct answer in 915 of 3,232 conversations with the prescription (28.3%), and 40% of these flipped in later turns. System-prompt sentences that disclose the documents as synthetic lower one model’s flip rate from 77.8% to 47.3%. Thus, we conclude that awareness about a forgery does not prevent a model from adopting its answer.
[NLP-84] Virtual Head Attention NEURIPS2026
【速读】： 该论文旨在解决长上下文大语言模型中注意力机制设计面临的“不可能三角”难题，即在键值缓存（KV cache）大小、训练与预填充（prefill）计算量（FLOPs）以及模型质量之间难以兼顾。现有方案如分组查询注意力（GQA）虽能保持优异模型质量但带来较高的缓存开销；而其他压缩缓存的方法如多级注意力（MLA）和混合特征注意力（MFA）分别存在性能下降或显著增加训练与预填充计算的问题。为此，论文提出虚拟头注意力（Virtual Head Attention, VHA），其核心创新在于通过两个轻量级线性操作实现三者同步优化：首先，查询预混合（Q Premix） 在每个KV分组内对查询进行近恒等变换，从减少的物理查询头中恢复虚拟头多样性；其次，线性后混合（Linear Postmix） 采用低秩 I+AB⊤ 残差结构融合头间特征。由于VHA不引入非线性操作，可在推理时将Premix折叠至查询投影、Postmix折叠至输出投影，使模型退化为标准的GQA-2架构，从而无缝复用所有现有的GQA推理优化技术。实验表明，仅使用2个KV头时，大型模型的分组规模可达64，充分挖掘张量核心并行性。在四个不同规模的4K上下文50B-token预训练模型上，VHA在各尺度均超越标准GQA基线（提升0.05%至0.81%），同时相比GQA-8降低4倍缓存开销、相比GQA-4降低2倍缓存开销，并将训练与预填充FLOPs降低7.9%（1.7B参数，S=4096）。在工业级30B-A3B MoE模型（500B token，8K上下文）上，VHA在12项基准测试中分别优于MLA和GQA-4达+0.70%和+2.60%；在长上下文中期训练后的10B-A1B MoE模型对比中，VHA在8K–128K上下文下的平均RULER得分为84.4，显著高于GQA的73.0，且仅需2个而非4个KV头，验证了其在高效率与高性能间的卓越平衡能力。
链接: https://arxiv.org/abs/2610.17959
作者: Xibo Ding,Guoxia Wang,Jinle Zeng,Jiabin Yang,Sirui Tao,Chao Yang,Dianhai Yu
机构: Baidu Inc.(百度); Peking University(北京大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Published as a main conference paper at NeurIPS 2026
Abstract:The design of attention mechanisms in long-context large language models faces an ``impossible triangle’’ among KV cache size, training and prefill FLOPs, and model quality. GQA achieves strong quality but incurs high KV cache overhead; alternative approaches such as MLA and MFA can compress the cache, but MLA suffers from quality loss while MFA substantially increases training and prefill compute. We propose Virtual Head Attention (VHA), which simultaneously improves all three vertices through two lightweight linear operations: Q Premix applies a near-identity transformation to queries within each KV group to recover virtual head diversity from halved physical query heads, and Linear Postmix fuses inter-head features via a low-rank I + AB^\top residual structure. Since VHA introduces no nonlinear operations, Premix can be folded into the query projection and Postmix into the output projection at inference time, reducing the model to a standard GQA-2 architecture that directly reuses all existing GQA inference optimizations. With only two KV heads, the group size in large models readily reaches 64, fully exploiting Tensor Core parallelism during decoding. Across four scales with 4K-context 50B-token pretraining, VHA consistently surpasses the standard GQA baseline at every scale ( + 0.05% to + 0.81%), while reducing KV cache by 4\times vs.\ GQA-8 at dense scales and 2\times vs.\ GQA-4 at the MoE scale, and lowering training and prefill FLOPs by 7.9% (1.7B, S=4096 ). On an industrial 30B-A3B MoE model trained on 500B tokens at 8K context, VHA outperforms both MLA and GQA-4 by + 0.70,/, + 2.60,% on a 12-benchmark base-model evaluation suite; in a separate matched 10B-A1B MoE comparison after long-context mid-training, VHA attains an average RULER score of 84.4 over 8K–128K versus 73.0 for GQA, while using two rather than four KV heads in the full-attention layers.
[NLP-85] Agent -Reflex: Teaching Language Agents to Act Reflectively
【速读】： 该论文旨在解决当前大型语言模型（LLM）作为交互式智能体在长时程工具使用任务中，面对环境反馈噪声及用户干预时缺乏有效反思学习与失败恢复机制的问题。现有方法在动态复杂环境下难以实现对错误的主动识别与修正，导致智能体鲁棒性不足。其解决方案的关键在于提出一种通用的训练框架——Agent-Reflex，该框架包含三个核心阶段：首先通过环境反射技能发现（RSD）引入基于真实工具反馈的扰动，构建能力多样化的语言模型池，并挖掘成功与失败案例对，提炼可复用的反射元技能；其次采用反射式监督微调（ReSFT），将这些元技能转化为指导自我反思、技能恢复和以用户为中心修正的监督信号，强化互补的反思行为；最后提出基于策略的反射式策略优化（ORPO），结合噪声环境下的代理强化学习与反射引导的在线策略蒸馏，进一步激发智能体的主动性反思能力。实验表明，基于Qwen3-8B实现的Agent-Reflex在五个长时程工具使用基准上取得58.6的综合得分，在AgentNoiseBench上达到29.1，显著优于现有环境扩展与反思基线方法，同时展现出更强的抗噪鲁棒性与稳定的训练动态，支持大规模智能体训练的可扩展性。
链接: https://arxiv.org/abs/2610.17953
作者: Guanting Dong,Xinyu Yang,Jiajie Jin,Yuyang Hu,Xiaoshuai Song,Yifei Chen,Xiaoxi Li,Jiejun Tan,Yutao Zhu,Tat-Seng Chua,Ji-Rong Wen,Zhicheng Dou
机构: Gaoling School of Artificial Intelligence, Renmin University of China(中国人民大学高瓴人工智能学院); National University of Singapore(新加坡国立大学)
类目: Computation and Language (cs.CL)
备注: Working in progress
Abstract:Large language models have evolved from single-turn chatbots into interactive agents capable of long-horizon reasoning. In real deployments, MCP tool services can return noisy feedback and users may intervene to correct the agent, but existing methods provide limited supervision for reflection learning and failure recovery. To bridge this gap, we propose Agent-Reflex, a general training framework for incentivizing proactive reflection in agents. Specifically, Agent-Reflex comprises three stages. First, Environment Reflection Skill Discovery (RSD) injects grounded tool-feedback perturbations into environment scaling, samples a capability-diverse LLM pool, mines matched success-failure pairs, and summarizes reusable reflection meta-skills. Building on this skill library, we then apply Reflective Supervised Fine-Tuning (ReSFT) that converts each skill into supervision for self-reflection, skill recovery, and user-centric correction, strengthening complementary reflective behaviors. Finally, we propose On-Policy Reflective Policy Optimization (ORPO) to combine agentic reinforcement learning under noisy interactions with reflection-guided on-policy distillation to further elicit proactive reflection. Built on Qwen3-8B, Agent-Reflex reaches 58.6 overall on five long-horizon tool-use benchmarks and 29.1 on AgentNoiseBench, surpassing strong environment-scaling and reflection baselines. Experiments on six challenging agent benchmark suites demonstrate the effectiveness of Agent-Reflex. Further analyses show that Agent-Reflex improves agents’ robustness to noise while maintaining stable training dynamics, supporting scalable agent training. Our code is available at this https URL.
[NLP-86] HaloBlock: Faster and Better Decoding for Diffusion Large Language Models
【速读】： 该论文旨在解决扩散型大语言模型（diffusion large language models, dLLMs）在推理过程中面临的两大核心问题：一是解码难度主要受块内右侧词元主导，导致生成效率受限；二是块级解码固有的边界上下文不连续性，影响跨块语义连贯性。针对上述挑战，论文提出HaloBlock——一种面向解码难度自适应的环形扩展机制（difficulty-aware adaptive halo mechanism）。其关键在于引入双向累积熵（bidirectional cumulative entropy）以量化每个块的解码难度，并据此动态地在块的两侧自适应分配双向环形区域（halo regions），从而缓解边界处的上下文断裂问题。该设计通过平滑跨块上下文传播、降低高难度词元的解码负担，显著提升生成质量与推理收敛速度。在数学推理与代码生成多个基准测试中，HaloBlock在LLaDA-8B-Instruct、Dream-7B-Instruct及LLaDA-1.5模型上均展现出一致的准确率提升与更高的推理效率。
链接: https://arxiv.org/abs/2610.17948
作者: Zhaohui Zheng,Chenhang He,Jinrui Zhang,Ming-Ming Cheng,Lei Zhang
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Diffusion large language models (dLLMs) have attracted increasing research attention due to their bidirectional attention and arbitrary order decoding. However, there are two issues in dLLMs inference: the decoding difficulty is dominated by the tokens on the right side of the block, and the block-wise decoding inherently suffers from boundary contextual discontinuity. To tackle these limitations, we propose HaloBlock, a difficulty-aware adaptive halo mechanism for dLLM inference. We first leverage bidirectional cumulative entropy to characterize block decoding difficulty. Then, based on the bidirectional decoding difficulty, we adaptively allocate bidirectional halo regions on both sides of each block to mitigate boundary contextual discontinuity. The difficulty-aware design of HaloBlock smooths cross-block context propagation and lowers decoding difficulty for hard tokens, yielding better generation quality and faster convergence. Evaluated on LLaDA-8B-Instruct, Dream-7B-Instruct and LLaDA-1.5 across mathematical reasoning and code generation benchmarks, HaloBlock demonstrates consistent accuracy improvements and higher inference efficiency. Source codes are publicly available at this https URL.
[NLP-87] Batch-Paced: Fixed-Budget Batch Decoding in Discrete Diffusion Language Models
【速读】： 该论文旨在解决生成式语言模型（Generative Language Models）在实际部署中因推理需求激增而带来的高效服务问题，尤其针对离散扩散语言模型（dLLM）在批量推理时的计算资源分配效率低下问题。现有dLLM解码方法虽能提升单个序列的推理效率，但未充分优化多请求批量处理的整体性能。为此，论文首次提出“固定预算下的批量解码”（fixed-budget dLLM batch decoding）问题，并设计了一种无需训练的解决方案——Batch-Paced，其核心在于通过联合优化不同请求间的解码计算分配，以提升整体批次的质量。实验表明，在两种dLLM骨干网络、五个基准测试及其混合场景下，Batch-Paced在相同计算预算下显著提升了批量性能，且收益随批大小增加而增强，并在多种解码配置中保持稳定。分析显示，性能提升主要源于提高低置信度请求的已提交分词置信度。此外，研究还证明Batch-Paced的批量解码时间具有竞争力，且引入解码深度容差（decoding-depth allowance）可进一步缩短延迟，仅带来轻微性能损失。
链接: https://arxiv.org/abs/2610.17917
作者: Shuai Zhang,Zhanchao Zhou,Huachuan Qiu,Anqi Li,Zhenzhong Lan
机构: Zhejiang University (浙江大学); Westlake University (西湖大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:As generative language models become more capable and widely used, inference demand grows, making efficient serving increasingly important. Discrete diffusion language models (dLLMs) offer a promising path to more efficient inference through parallel token generation. Prior dLLM decoding methods, however, focus on improving inference efficiency for individual sequences. We formulate fixed-budget dLLM batch decoding and introduce Batch-Paced, a training-free method that jointly optimizes the allocation of decoding compute across requests to improve overall batch quality. Across two dLLM backbones, five benchmarks and their representative mixtures, Batch-Paced improves batch performance under matched budgets; gains generally grow with batch size and largely persist across decoding configurations. Analysis of requests’ decoding trajectories suggests that these gains arise largely by raising the committed-token confidence of lower-confidence requests. We further show that Batch-Paced’s batch decode time remains competitive and that adding a decoding-depth allowance during Batch-Paced decoding can shorten it further at a modest performance cost.
[NLP-88] xt-Only vs. Image-Aware VLM Judges for Manga Translation Evaluation
【速读】： 该论文旨在解决漫画翻译评估中长期存在的视觉上下文缺失问题，即传统评估方法仅依赖文本自动指标（text-only automatic metrics），忽视了图像所承载的说话人身份、指代关系及场景基调等关键视觉语境信息，导致评估结果与真实翻译质量之间存在偏差。其解决方案的关键在于构建一个基于统一视觉-语言模型（Vision-Language Model, VLM）的实用验证框架，将同一VLM分别用作仅依赖文本的判别器（Text-Only Judge, TOJ）和具备图像感知能力的判别器（Image-Aware Judge, IAJ），通过对比二者在评估结果上的差异，系统性地探究图像输入对翻译质量评价的影响。实验结果表明，引入页面图像可显著提升模型对依赖视觉信息的故意退化内容的检测能力；同时，TOJ与IAJ在错误数量和错误类别分布上呈现明显分歧，但人工验证未发现两者间存在一致偏好，揭示了当前图像感知评估在主观一致性方面仍存在挑战。
链接: https://arxiv.org/abs/2610.17908
作者: Subaru Kimura,Jun Suzuki
机构: Tohoku University (东北大学)
类目: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注: 23 pages, 8 figures, 13 tables. Accepted at WMT 2026
Abstract:In manga translation, visual context outside speech-bubble text, including speaker identities, referents, and scene tone, is depicted in page images, making it natural to refer to images during translation evaluation. However, most conventional manga translation evaluations rely solely on text-only automatic metrics, leaving the impact of image input insufficiently explored. We propose a practical validation framework that employs the same vision-language model (VLM) as both a text-only judge (TOJ) and an image-aware judge (IAJ) to investigate evaluation divergences associated with image-aware evaluation procedures. We use a Japanese-English manga dataset of official translations aligned at the page and speech-bubble levels. First, adding page images to TOJ improved the primary VLM’s detection of deliberately injected, visually dependent degradations. Second, we analyzed evaluation divergences between TOJ and IAJ, finding different error counts and error-category distributions. Human verification did not reveal a consistent preference between TOJ and IAJ.
[NLP-89] Evaluating the Causal Faithfulness of SHAP as a Candidate Attribution Method for SAE Features
【速读】： 该论文旨在解决稀疏自编码器（Sparse Autoencoders, SAE）中特征解释的因果有效性问题。尽管已有研究将语言模型表示分解为可解释的特征，但其特征解释仍以相关性为主，缺乏对特征实际因果影响的直接评估。为此，论文采用基于SHAP的归因方法对SAE潜在空间中的特征进行重要性估计，并通过直接因果干预（causal intervention）在SST-2情感分类任务上进行验证，使用GPT-2 Small与一个7层、32,768个特征的SAE作为实验基准。研究对比了SHAP、积分梯度（Integrated Gradients）、梯度×输入（Gradient × Input）及探针权重基线方法在不同探针结构下的表现，以特征消融和激活操控后语言模型自身的对数几率差作为因果效应指标。结果显示，多数归因方法的忠实性关联在多重比较校正后不显著；且各归因方法间高度一致（ρ = 0.42–0.80）并不意味着其与真实因果效应的一致性，表明方法间一致性不能作为因果忠实性的证据。两个稳健发现包括：局部逐样本归因在方向性判断上优于全局平均归因，但其看似更强的重要性排序实由特征活跃度驱动，而非更优的因果排序能力；在已知真值最小差异对下，归因方法能有效检测到插入的情感信号（AUROC = 0.792），但无法有效定位删除操作的影响（AUROC = 0.520），说明归因方法对特征存在性的检测优于对缺失性的识别。论文进一步提出，重要性忠实性（importance faithfulness）与方向性忠实性（directional faithfulness）应被视为独立属性，需分别评估，且归因结果的一致性不应替代对因果效应的独立验证。
链接: https://arxiv.org/abs/2610.17898
作者: Samuel Xie,Noura Al-Moubayed
机构: Durham University (杜伦大学)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 24 pages, 3 figures; to be submitted to International Journal of Secondary Computing and Applications Research
Abstract:Sparse autoencoders decompose language model representations into interpretable features, but feature interpretation in this literature remains predominantly correlational. Attribution methods offer more direct estimates of feature influence, yet their faithfulness to actual causal effects has not been systematically evaluated in the SAE setting. We apply SHAP-based attribution to SAE latent features and evaluate it against direct causal intervention, using sentiment classification on SST-2 with GPT-2 Small and a 32,768-feature layer-7 SAE as a testbed. We compare SHAP to Integrated Gradients, Gradient \times Input, and probe-weight baselines across linear and MLP probe arms, scoring each against feature ablation and activation steering measured through the language model’s own logit difference. Most faithfulness associations we measure do not survive multiple-comparison correction, and attribution methods that agree strongly with one another ( \rho = 0.42 – 0.80 ) agree poorly with measured causal effects, indicating that inter-method agreement is not evidence of causal faithfulness. Two effects hold under control. First, local per-example attribution retains a directional advantage over global averaged attribution, while its apparent importance advantage is explained by feature activity rather than superior causal ranking. Second, under known-ground-truth minimal pairs, attribution recovers inserted sentiment signal well above chance (AUROC 0.792 ) but fails to localize deletions (AUROC 0.520 ), suggesting that these methods detect the presence of a feature’s influence more reliably than its absence. We argue that importance faithfulness and directional faithfulness are separable properties requiring separate evaluation, and that attribution agreement should not substitute for causal validation.
[NLP-90] Swiss4Arg: A Parallel Dataset for Multilingual Argument Mining across Swiss National Languages
【速读】： 该论文旨在解决多语言论据挖掘（Argument Mining）中缺乏跨语言可比性与低资源语言（如罗曼什语）标注数据的问题。现有数据集往往聚焦于主流语言，且在不同语言间缺乏结构对齐与立场平衡，限制了跨语言比较研究的可行性。为此，本文提出瑞士4论据（Swiss4Arg）——首个基于瑞士联邦公投宣传册构建的多语言论据挖掘数据集，涵盖德语、法语、意大利语和罗曼什语四种官方语言，其文本由明确表述的正反论点系统组织而成，实现了跨语言对齐与论点立场均衡。关键解决方案在于：构建一个统一标注框架，对论据成分（主主张、主张、前提及非论据单元）和关系（支持与攻击）进行三名母语者独立标注，并通过计算联合跨标注者一致性（Krippendorff’s alpha = 0.548 和 0.623）验证标注质量；同时公开经仲裁的标签与原始标注结果，支持对分歧模式与语言差异的深入分析。该数据集为多语言论据挖掘提供了新的基准，尤其适用于低资源语言与跨语言可比性研究。
链接: https://arxiv.org/abs/2610.17894
作者: Sophia Conrad,Fatma-Zohra Rezkellah,Christina Niklaus,Reto Gubelmann
机构: University of Zurich (苏黎世大学); University of St. Gallen (圣加仑大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:We present Swiss4Arg, a multilingual dataset for argument mining based on official Swiss voting brochures, covering all four national languages: German, French, Italian, and Romansh (a Romance minority language exclusively spoken in Switzerland). The brochures contain explicitly formulated arguments systematically organized into pro and con sections, resulting in a cross-lingually aligned resource with balanced argumentative stance. We create a dataset that comprises 53 federal voting issues collected between 2019 and 2025, annotated for argument components (major claims, claims, premises, and non-argumentative units) and relations (support and attack) by three native speakers per language. Inter-annotator agreement (Krippendorff’s alpha) reaches 0.548 for component classification and 0.623 for relation and target classification, computed jointly across all annotators on the parallel data. We release both adjudicated labels and individual annotator annotations, enabling analyses of disagreement and cross-linguistic variation, and provide baseline results using GPT-5.4 and SwissBERT. Swiss4Arg establishes a novel benchmark for multilingual argument mining, with a particular focus on cross-lingual comparability and low-resource settings such as Romansh. The dataset and code are publicly available at this https URL (HuggingFace) and this https URL (GitHub).
[NLP-91] raining LLM s for Multi-Agent Collaboration with Verifiable Information Rewards
【速读】： 该论文旨在解决大规模语言模型（Large Language Model, LLM）在多智能体系统中实现有效协作时面临的挑战，即如何在缺乏明确反馈的情况下，训练智能体正确共享相关信息、主动请求缺失信息，并合理整合团队成员的贡献以做出决策。传统方法因最终任务结果提供的反馈信号过于稀疏，难以对个体通信行为进行有效评估与优化。为此，论文提出一种后训练方法CoSIR（Collaboration via Surprisal-based Information Rewards），其核心在于利用基于贝叶斯的可验证评估器，通过衡量信息共享前后团队“惊讶度”（surprisal）的降低程度来量化信息价值——惊讶度定义为各智能体基于已有证据对正确选项所赋予的负对数概率之和。具体而言，共享信息的奖励依据其能否显著降低团队整体惊讶度；请求信息的奖励则基于若该信息被真实提供所能带来的惊讶度潜在下降量；此外，还设计独立奖励机制鼓励智能体在其私有决策中有效利用接收到的信息。该方法通过自对弈（self-play）方式对Qwen3-4B、Qwen3-8B及Gemma-3-4B-it等模型进行后训练，并在七个未见的自然语言协作任务上验证迁移能力，结果显示CoSIR在综合得分上表现最优，同时在通用任务中仍保持竞争力。行为分析进一步表明，智能体表现出更高的私有证据披露意愿，且更少出现忽略队友输入的情况，证明了该框架在促进实质性协作方面的有效性。
链接: https://arxiv.org/abs/2610.17893
作者: Keru Chen,Sen Lin,Yingbin Liang,Nathaniel D. Bastian,Shaofeng Zou
机构: Arizona State University (亚利桑那州立大学); University of Houston (休斯顿大学); The Ohio State University (俄亥俄州立大学); Johns Hopkins University (约翰霍普金斯大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 23 pages
Abstract:Effective collaboration among large language model (LLM) agents in multi-agent systems requires sharing relevant evidence, requesting missing information, and incorporating teammates’ contributions into decisions. Training these behaviors is challenging because final task outcomes provide limited feedback on the value of individual messages. We introduce CoSIR (Collaboration via Surprisal-based Information Rewards), a post-training approach that uses a simple, verifiable environment to teach collaboration behaviors that generalize to more complex, unseen tasks. We propose a Bayesian evaluator that scores shared information by its reduction in team surprisal, defined as the sum of the negative log probabilities assigned to the correct option given each agent’s available evidence. Requests are scored by the potential reduction in the same quantity if the requested information were truthfully provided. Separate rewards encourage agents to use the received information in their private choices and calculations. We post-train Qwen3-4B, Qwen3-8B, and Gemma-3-4B-it through self-play and evaluate transfer to seven unseen natural-language collaboration tasks. CoSIR achieves the highest aggregate score, while its performance in general tasks remains competitive. Behavioral analyses show increased disclosure of private evidence and fewer judged failures to use teammates’ input.
[NLP-92] Cross-Lingual Transfer of Evil Vectors Is Directionally Asymmetric EMNLP2026
【速读】： 该论文旨在解决生成式AI中“邪恶人格向量”（evil persona vector）是否具有跨语言通用性的问题，即其是否反映一种真正跨语言的语义概念，而非仅源于特定语言提取过程中的数据偏差。研究发现，尽管从低资源语言（如斯瓦希里语、泰卢固语）中提取的邪恶向量难以可靠获得且在行为操控上效果有限，但从高资源语言（英语、西班牙语）中提取的向量则能有效迁移至低资源语言输入，并在保持通用知识能力（以MMLU为评估指标）的同时实现对模型行为的有效控制。其解决方案的关键在于：通过几何分析验证，高资源语言的邪恶向量在方向上高度一致，表明存在一个共享的跨语言邪恶方向；一旦在高资源语言中可靠提取，该向量可直接用于低资源语言的可控行为引导，无需额外的本地化数据收集，从而为多语言有害行为建模与防御提供了高效可行的路径。
链接: https://arxiv.org/abs/2610.17880
作者: Juann Kim,Minku Kang
机构: University of Southern California (南加州大学); KAIST (韩国科学技术院)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to the Workshop on Multilingual Representation Learning (MRL) at EMNLP 2026
Abstract:Persona vectors are linear directions in the residual-stream activations of LLMs that encode behavioral traits and enable fine-grained control over model behavior. Among them, the evil persona vector is particularly relevant to understanding harmful behavior, jailbreak vulnerabilities, and the unlearning of harmful knowledge. Yet whether this vector reflects a genuinely cross-lingual concept, rather than an artifact of its extraction language, remains untested. We conduct a systematic evil-vector transfer study across two high-resource languages (English, Spanish) and two low-resource languages (Swahili, Telugu) on four open-source instruction-tuned LLMs, and find that transfer is directionally asymmetric. Evil vectors are harder to obtain reliably from low-resource languages, and the resulting vectors are largely ineffective at steering, even within their own language. In contrast, vectors extracted from high-resource languages transfer effectively to low-resource inputs while largely preserving general knowledge, as measured by MMLU. Geometric analysis further supports this pattern: evil vectors from the two high-resource languages are closely aligned, whereas low-resource vectors align only weakly with this shared direction. These results suggest that a shared cross-lingual evil direction exists and, once reliably extracted from a high-resource language, can be reused to steer behavior in low-resource languages without additional per-language data collection.
[NLP-93] Prefilling-dLLM v2: Learned Chunk Routing for Long-Context Diffusion Language Models
【速读】： 该论文旨在解决扩散语言模型（Diffusion Language Model, dLLM）在长序列建模时因全上下文双向注意力机制导致的计算开销急剧上升的问题。现有方法在推理阶段引入稀疏化策略，但未在训练阶段对齐稀疏模式，造成预训练权重与实际运行模式不一致，影响模型性能。其解决方案的关键在于提出Prefilling-dLLM v2，通过持续训练使模型在生成过程中自然使用其原生掩码标记（mask token）作为路由槽，并在每段末尾添加一个掩码标记，结合可学习的角色偏移量将其标识为路由位置。具体而言，在每第四个网络层中，查询保留局部双向窗口并基于掩码标记选择得分最高的前K个块；其余层则采用双向滑动窗口机制。该设计实现了训练与推理阶段稀疏模式的一致性，显著提升了长序列任务下的表现。在16K至128K长度范围内，相较于双向非自回归（bidirectional NSA）、动态稀疏注意力（DSA）和滑动窗口注意力（SWA），Prefilling-dLLM v2在三项RULER基准任务上均取得领先，于128K长度下平均准确率较次优方法NSA高出7.4个百分点，而DSA与SWA性能仍接近零。同时，在相同长度下，解码速度比全稠密注意力快172倍。
链接: https://arxiv.org/abs/2610.17868
作者: Qi Han,Jing Xiong,Ziyuan Zhang,Ngai Wong
机构: University of Hong Kong(香港大学); Zhejiang University(浙江大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Diffusion language models apply bidirectional attention over the whole context at every denoising step, so a longer context makes each step substantially more expensive. Sparsifying only at inference leaves the pretrained weights unchanged, so the model is never trained under the sparse pattern it runs with. We propose Prefilling-dLLM v2, which continues training a diffusion language model so that it routes with its native mask token. One mask is appended at the end of each chunk, and a learned role offset marks it as a routing slot. On every fourth layer, a query keeps a bidirectional local window and the top-K chunks scored from that mask; the other layers use a bidirectional sliding window. Starting from Dream-v0-Base-7B extended to 16K with YaRN, we compare Prefilling-dLLM v2 with bidirectional NSA, DSA, and SWA under the same layout, key budget, and dense initialization, using the same Fast-dLLM decoder. With the YaRN factor scaled to the evaluation length, Prefilling-dLLM v2 leads the other sparse methods on all three RULER tasks from 16K to 128K. Its mean accuracy at 128K is 7.4 points above NSA, the next sparse model, while DSA and SWA remain near zero. At the same length, decoding is 172x faster than with full dense attention.
[NLP-94] Diff Diffusion: Predicting Edit Paths with Diffusion Models for Efficient Code Editing NEURIPS2026
【速读】： 该论文旨在解决生成式代码编辑中因传统自回归模型（autoregressive models）在处理稀疏编辑场景时效率低下而带来的计算浪费问题。现有扩散语言模型（Diffusion Large Language Models, dLLMs）虽具备并行更新多个词元的优势，但其设计多聚焦于序列生成而非编辑预测，难以适配现代AI辅助集成开发环境（IDE）中对局部、低延迟编辑预测的需求。此类场景下，代码修改通常仅涉及少量位置，多数源码保持不变，若采用全量重生成将造成显著冗余计算。为此，论文提出Diff Diffusion——一种受扩散模型启发的代码编辑框架，其核心创新在于将去噪过程建模为结构化的编辑操作（structured edit operations），通过多阶段训练与高效并行解码机制，使模型聚焦于实际变更区域，实现精准且低延迟的编辑预测。该方法在公共编辑补全基准测试、编辑预测评估及线上验证中均达到或超越自回归基线性能，同时在推理速度上相较基线提升2.78倍，并在生产环境中实现39.0%的P50模型延迟降低，为高精度与低延迟兼具的AI应用提供了新的建模范式。
链接: https://arxiv.org/abs/2610.17831
作者: Xiaolong Xu,Hao Jiang,Bo Jiang,Zhiwen Deng,Ningyuan Sun,Bingzhou Chen,Jin Chen,Chenxing Wei
机构: ByteDance(字节跳动); Shenzhen, China
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 15 pages, 3 figures, 6 tables. Accepted at the NeurIPS 2026 Workshop on Diffusion Language Models: Foundations, Efficiency, and Reasoning
Abstract:Diffusion large language models (dLLMs) are appealing for code generation because they can update many tokens in parallel, yet many are optimized for sequence synthesis rather than edit prediction. This setting is poorly matched to modern AI-assisted IDEs, where edit prediction must infer the next code change from recent edits, the current file, and cursor-local context. Such changes are usually sparse and preserve most source tokens, so regenerating a full code segment wastes computation. We propose Diff Diffusion, a diffusion-inspired LLM framework for code editing that predicts edit paths from an original snippet to its revised form. Diff Diffusion represents denoising as structured edit operations, and combines this formulation with multi-stage training and efficient parallel decoding. By focusing computation on changed regions, Diff Diffusion supports accurate low-latency edits and multi-position parallel modification. Across public edit-completion benchmarks, edit-prediction evaluations, and online validation, Diff Diffusion matches or leads autoregressive baselines while delivering 2.78 times faster inference on public edit-completion benchmarks and a 39.0% P50 model-latency reduction in production. This work provides a new modeling perspective for AI applications that require high accuracy and low latency.
[NLP-95] Checking the Checkers: A Synthesis of Verifier Noise and Robustness in RLVR
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在基于可验证奖励的强化学习（Reinforcement Learning with Verifiable Rewards, RLVR）中因验证器（verifier）错误而导致模型推理能力提升受限的问题。核心问题在于，现有方法普遍假设验证器评分准确无误，但实际中规则型验证器可能因格式不符而拒绝正确答案，单元测试可能误接受错误代码，大语言模型（LLM）评判者也可能被诱导认可空响应或对抗性输出。研究通过系统综述58项相关研究，揭示了验证器错误的类型、对训练过程的影响机制以及不同缓解策略的有效性边界。其解决方案的关键在于：验证器错误的结构特征（如是否可被策略学习利用）比错误发生率更为重要——随机错误主要延缓学习进程，而可被策略主动触发的假阳性（false positives）则可能导致训练陷入停滞或崩溃，即便初始错误率较低亦然。此外，验证器的错误模式与其构建主体和检查内容密切相关，且随策略优化动态演变。研究进一步指出，多数缓解方法仅针对特定类型的错误进行测试，缺乏通用性。为此，作者整合证据形成对比表格，调和文献中的矛盾结论，并提出标准化报告协议，以增强未来研究结果的可比性与可复现性。
链接: https://arxiv.org/abs/2610.17824
作者: Savyasachi Gupta,Aniruddha Srinivas Joshi
机构: Independent Researcher
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Accepted at the International Conference on Artificial Intelligence Centric Systems, Services, and Applications (AISSA 2026)
Abstract:Reinforcement learning with verifiable rewards (RLVR) has become a principal recipe for improving the reasoning of large language models (LLMs). It rests on the assumption that the checker scoring each response is itself correct. In practice, rule-based checkers reject correct answers in unexpected formats, unit tests accept incorrect programs, and LLM judges can be led to approve empty or adversarial responses. This work surveys 58 studies that measure, analyze, or mitigate such verifier errors. We ask what errors verifiers make, how they affect training, which mitigations work against which errors, and how the findings extend to self-generated, rubric- and judge-based, and verifier-free rewards. The evidence suggests that the structure of an error matters more than its rate. Random errors mostly slow learning, though under a fixed budget or KL regularization they also lower final accuracy. False positives that the policy can learn to trigger can cause plateaus or collapse, sometimes even with a low initial error rate. Whether a verifier wrongly rejects or accepts depends on what it checks and who built it, not on its family. Its errors shift as the policy improves, and most mitigations have been tested against only one kind of error. We compile the findings into comparison tables, reconcile apparent conflicts in the literature, and propose a reporting protocol to make future results comparable.
[NLP-96] Perturbation Not Provenance: Why Activation Steering Does Not Make a Robust Text Watermark
【速读】： 该论文旨在解决大语言模型生成文本的可追溯性问题，提出利用激活空间中的线性方向（linear directions in the activation space）作为鲁棒文本水印（text watermark）的潜在基础。其核心假设是：通过沿特定语义方向（如正式度、情感、复杂度等）引导模型生成，并在重构后通过探测激活值来恢复水印信号，即可实现对生成内容的溯源。然而，实验结果表明，基于激活空间方向控制的水印方案在检测准确性上显著劣于标准的基于逻辑值（logit-based）的水印方法（KGW）。在保持生成文本流畅性相近（困惑度提升不超过1.15倍）的前提下，KGW在句子级改写攻击下达到0.81的AUC，而最优激活方向仅达0.65；面对更强的段落级改写攻击时，两者性能均下降，但差距未反转。进一步分析显示，激活方向的可检测性并非源于其语义意义——随机方向与有意义方向具有相似的可检测性，且该信号不依赖于探测层或表面特征控制，说明激活空间中的方向变化反映的是全局、内容无关的分布偏移，仅在文本严重退化时才可被有效检测。因此，研究结论指出：当前条件下，语义激活空间无法编码局部化、可恢复的语义水印信号，难以用于构建可靠水印机制。
链接: https://arxiv.org/abs/2610.17810
作者: Sergey Pletenev
机构: AIRI; Skoltech(斯科尔科沃科学技术研究院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Linear directions in the activation space of large language models causally and universally influence model behavior. This suggests that they might serve as a basis for a robust text watermark. The idea is to steer the generation along a specific semantic direction and then recover the mark by probing the activations of the re-encoded text. We test this hypothesis by conducting a series of experiments. We vary the strength of the steering and use four different directions (formality, sentiment, complexity, and a random control) across two different model families. We also test two paraphrasing attacks: a sentence-level one and a stronger paragraph-level one. The results show that activation-steering watermarks are outperformed by a standard logit-based watermark (KGW) in terms of detection accuracy. At a matched level of fluency (within 1.15 times the original text’s perplexity), KGW achieved an AUC of 0.81 against a sentence-level paraphraser, while the best steering direction only reached 0.65. A stronger paragraph-level paraphraser lowered both (0.65 for KGW and 0.59 for steering), narrowing the gap but not reversing it. On Qwen, it left the two indistinguishable but at near-chance levels for both. We further showed that the detectability provided by steering was not semantic: a random direction was just as detectable as a meaningful one. The signal was independent of which layer was probed and whether a surface feature control was used. We conclude that, in this case, semantic activation does not encode a localized, recoverable semantic mark, but rather a global, content-independent shift that can only be strongly detected when the text is degraded.
[NLP-97] BRIDGE: Selective Boundary Repair with Dependency Graph Evidence in Multi-Agent Systems
【速读】： 该论文旨在解决多智能体协作系统中普遍存在的“交接失效”（broken handoffs）问题，即在复杂长周期任务中，尽管各智能体输出局部合理的响应，但关键任务上下文仍可能被遗漏、错置或忽略，导致协作流程中断。其核心解决方案是提出一种可验证的运行时控制框架 \method，其关键在于将交接修复建模为对显式依赖图的有选择性干预。该框架通过一个集成的故障归因与校准管道 \dataset，解耦执行依赖关系，并注入反事实交接扰动以提供实证监督信号。在此诊断基础上，\method 动态构建可观测交互轨迹的溯源图（provenance graph），无需访问私有智能体状态即可定位未解决的依赖关系；随后通过一个基于实证评估的门控机制，权衡预期收益与潜在干扰风险，在满足形式化局部后条件的前提下才执行修复操作。实验表明，\method 在 SILO-BENCH 和 TeamBench 上分别提升任务成功率达12.3和11.1个百分点，且边际令牌开销低至2.0%，在 MAST 基准的562条交接相关轨迹上实现69.8%的F1值，确立了可验证协作修复的可靠操作边界。
链接: https://arxiv.org/abs/2610.17805
作者: Encheng Su,Pengze Li,Ziheng Hua,Qiucheng Yu,Jianyu Wu,Chen Tang,Yizhou Wang,Lintao Wang,Junjia Xiang,Xinzhu Ma,Xi Chen,Shixiang Tang,Houqiang Li,Xi Chen,Aoran Wang
机构: University of Science and Technology of China (中国科学技术大学); Fudan University (复旦大学); City University of Hong Kong (香港城市大学); Shanghai Jiao Tong University (上海交通大学); The Chinese University of Hong Kong (香港中文大学); The University of Sydney (悉尼大学); Harbin Institute of Technology (哈尔滨工业大学); Beihang University (北京航空航天大学); East China Normal University (华东师范大学); Shanghai Artificial Intelligence Laboratory (上海人工智能实验室)
类目: Computation and Language (cs.CL)
备注:
Abstract:Multi-agent collaboration has emerged as a critical paradigm for tackling complex, long-horizon tasks; however, while coordination protocols continue to advance, systemic failures at the inter-agent collaboration interface remain a pervasive bottleneck. In collaborative workflows, systems frequently suffer from \emphbroken handoffs—where critical task context is omitted, misrouted, or ignored despite each agent emitting locally plausible responses. To address this challenge, we present \method, a verifiable runtime control framework that treats handoff repair as selective intervention over explicit dependency graphs. Central to \method is an integrated fault-attribution and calibration pipeline, \dataset, which decouples execution dependencies and injects counterfactual handoff perturbations to provide empirical supervision for runtime intervention. Leveraging this diagnostic foundation, \method dynamically compiles observable interaction traces into a provenance graph to pinpoint unresolved dependencies, without accessing private agent states. It then evaluates candidate repairs through an empirical gate that weighs expected gains against disruption risks, and executes repairs only when formal local postconditions hold. Extensive evaluations across demanding benchmarks demonstrate that \method boosts task success by up to 12.3 absolute percentage points on SILO-BENCH and 11.1 on TeamBench. Across both backbones, \method maintains strict cost-effectiveness, adding as little as 2.0% marginal token overhead. Furthermore, on 562 handoff-relevant traces from the MAST benchmark, \method achieves 69.8% F1 in a single pass, establishing a reliable operational boundary for verifiable coordination repair.
[NLP-98] Where Words Begin and End
【速读】： 该论文旨在解决语音强制对齐（forced alignment）在真实场景下，尤其是电话语音中的精度不足问题。现有主流的基于连接时序分类（CTC）的对齐器在词边界处频繁插入静音，且几乎未在电话语音上进行过评估，导致其在实际应用中性能不佳。针对这一问题，研究提出一种基于Whisper模型并采用低秩适配（LoRA）微调的方法，将每个词的起始与结束时间建模为时间戳标记（timestamp tokens），仅使用人工标注的47小时跨87种语言的边界数据进行训练。关键创新在于：一、在词语之后放置时间戳槽位以提升定位准确性；二、在噪声语料上训练，使模型对转录文本中的错误具有鲁棒性。实验表明，该方法在所有英语测试条件（包括读诵、对话及电话语音）下均优于目前最精确的MFA系统，尤其在真实固网电话语音上达到16.0 ms的词边界误差，显著优于MFA的36.5 ms。此外，基于FA-Bench协议在TIMIT数据集上的表现也优于已有最佳系统（18.2 ms vs. 11.6 ms）。
链接: https://arxiv.org/abs/2610.17782
作者: Husein Zolkepli
机构: Scicom (MSC) Berhad(斯科姆公司(马来西亚))
类目: Computation and Language (cs.CL)
备注:
Abstract:Forced aligners locate each word of a known transcript in the audio, and many speech pipelines depend on them. In our own data, a widely used CTC aligner inserts a pause at nearly every word boundary, and we found almost no aligner evaluated on telephone audio. We compare nine aligner configurations against human word boundaries on read, conversational and telephone English. The Montreal Forced Aligner (MFA) is the most precise, while CTC aligners consistently start words late, because most frames inside a word go to the blank token. We then fine-tune Whisper with LoRA to predict each word’s start and end as timestamp tokens, training only on boundaries placed by people: 47 hours in 87 languages. The resulting aligner is more precise than MFA on every English test condition, and than MFA’s own Korean and Dutch models. With FA-Bench’s protocol and code on TIMIT, its word-boundary error is 11.6 ms, against 18.2 ms for the best published system. Two choices matter most on telephone conversations: placing the timestamp slots after the word they time, and training on noisy transcripts, which also makes the aligner nearly insensitive to errors in the transcript. The released model reaches 16.0 ms on real landline calls, against 36.5 ms for MFA. The models are available at this https URL
[NLP-99] Full-duplex Spoken Language Modeling with Discrete Diffusion
【速读】： 该论文旨在解决全双工语音对话系统中语言智能与响应效率之间的固有权衡问题，即传统自回归（Autoregressive, AR）模型在保证高语言理解能力的同时难以实现低延迟响应。其核心解决方案在于引入基于离散扩散（discrete diffusion）的全双工语音模型Diffuplex，通过在训练阶段将“何时说”（when to say）与“说什么”（what to say）解耦，而非将其压缩于单一自回归时序中。在推理过程中，模型采用自回归交互策略持续追踪用户语音输入，同时利用并行扩散机制独立生成与优化回应内容，从而突破了语义计算对感知音频帧率的限制。这一设计使语言智能与响应速度得以同步提升：开启思维模式时，每前向传播可生成13.0个词元，平均生成速率高达1872词元/秒，显著优于前沿AR模型；关闭思维模式时，仍能保持与现有全双工模型相当的首音频延迟，并展现出更强的语言智能。该工作确立了新一代实时语音代理的新范式，实现了高智能与高响应性的协同进化。
链接: https://arxiv.org/abs/2610.17780
作者: Chen Chen,Pin-Jui Ku,Ankita Pasad,Ante Jukić,Zhehuai Chen,Boris Ginsburg
机构: NVIDIA(英伟达)
类目: Computation and Language (cs.CL)
备注:
Abstract:We introduce Diffuplex, a full-duplex spoken language model based on discrete diffusion that achieves an unprecedented balance between linguistic intelligence and response efficiency. With thinking mode, Diffuplex delivers 13.0 tokens per forward pass, reaching an average of 1872 tokens per second and substantially surpassing the frontier autoregressive (AR) models in decoding speed. Without thinking, Diffuplex matches the first-audio latency of existing full-duplex models while achieving markedly stronger linguistic intelligence. This capability is enabled by separating when to say from what to say at the training level, rather than compressing both into a single autoregressive timeline. During inference, Diffuplex continuously tracks the user speech through an AR interaction policy, while independently generating and refining response content through parallel diffusion. Consequently, semantic computation is no longer bounded by the perception audio frame rate, allowing intelligence and responsiveness to improve simultaneously. Taken together, these results establish a new paradigm for spoken agents that can sustain real-time dialogue while devoting parallel computation to intelligent response generation. Training code and checkpoints will be avaliable soon.
[NLP-100] GENDIS: Gender-Related Discourse and Stance in Multilingual YouTube Comments from India Kenya and Nigeria EMNLP2026 WOAH2026
【速读】： 该论文旨在解决现有性别相关网络内容研究中普遍存在的两个关键问题：一是研究多采用以伤害为中心的任务范式，导致非有害的性别相关内容与无关内容被混淆；二是现有数据集高度集中于高资源语言和全球北方（Global North）地区，缺乏对全球南方（Global South）语境下的多语言、本地化数据支持。为弥补上述缺口，本文提出GENDIS数据集，包含来自印度、肯尼亚和尼日利亚的9,600条多语言、人工标注的YouTube评论（每国3,200条），采用两级标注框架：首先识别是否提及性别，再对性别相关内容分类为负面、正面或中性。该分离机制实现了不依赖于伤害性的性别话语基准量度，从而支持对性别话题出现频率及评价维度的精准刻画。结果显示，在随机抽样层中，23.8%–35.0%的评论涉及性别议题，其中具有明确立场的评论中82%为中性或正面。模型表现表明，性别提及检测任务显著优于立场分类，尤其在正向与中性表达识别方面；轻度监督分类器在显著更低的成本下达到与前沿零样本大语言模型相当的性能，并优于通用毒性检测模型，但跨国家迁移能力有限。GENDIS为衡量在线语境中性别被提及的时机及其表述方式提供了具有本地根基的资源。
链接: https://arxiv.org/abs/2610.17778
作者: Luise Koch,Manuel Tonneau,Niyati Malhotra,Sunny Rai,Ana Maria Munoz Boudet,Victor Orozco-Olvera,Samuel P. Fraiberger,Sharath Chandra Guntuku
机构: World Bank; Technical University of Munich(慕尼黑工业大学); University of Oxford(牛津大学); New York University(纽约大学); University of Pennsylvania(宾夕法尼亚大学)
类目: Computation and Language (cs.CL)
备注: 23 pages. Accepted at the Workshop on Online Abuse and Harms (WOAH 2026), co-located with EMNLP 2026
Abstract:Research on gender-related online content is often operationalized through harm-centered tasks that collapse non-harmful gender-related and unrelated content, while existing datasets remain concentrated in high-resource languages and Global North settings. To bridge these gaps, we introduce GENDIS, a dataset of 9,600 multilingual, human-annotated YouTube comments from India, Kenya, and Nigeria (3,200 per country), annotated using a two-level framework that first identifies whether gender is referenced and then classifies gender-related content as negative, positive, or neutral. This separation provides a denominator for gender-related discourse independent of harm, enabling both prevalence estimation and characterization of its evaluative range. In the uniform random stratum, 23.8-35.0% of comments are gender-related, and among those with a resolved stance, 82% are neutral or positive. Across models, gender-reference detection is substantially easier than stance classification, especially for positive and neutral speech. Lightly supervised classifiers match frontier zero-shot large language models at substantially lower cost and outperform generic toxicity detection, while cross-country transfer remains limited. GENDIS provides a locally grounded resource for measuring when gender is invoked online and how that reference is framed.
[NLP-101] Beyond Belief Nesting: Rethinking Higher-Order Theory of Mind in Large Language Models
【速读】： 该论文旨在解决现有理论心理（Theory of Mind, ToM）评估基准中普遍存在的“阶次虚高”问题，即表面上的高阶信念嵌套（belief-nesting depth）可能通过低阶捷径（lower-order shortcuts）即可解答，导致评估结果无法真实反映模型或人类在复杂多层心理状态推理中的能力。其解决方案的关键在于提出一个全新的评估框架——HopToM，通过形式化定义“有效阶次”（effective order），基于支持连续信念前缀的不同事件来区分真正的高阶推理，并设计了具有严格阶次认证的全阶问题与对低阶替代方案的控制机制。此外，论文提出一种可插拔的推理时方法HopTrace，将每次交互编码为视角记录（perspective records），以确定性方式解析嵌套查询，显著提升大语言模型在第四、第五阶任务上的准确率，分别提高22.0和38.5个百分点，同时大幅降低计算开销（仅需多调用方法的1/20至1/39的令牌数）。该研究不仅提升了高阶ToM评估的严谨性，也为交互式智能体中的视角推理提供了实用增强路径。
链接: https://arxiv.org/abs/2610.17777
作者: Shanshan You,Zhengqi Wang,Hao Liu,Yuqin Liu,Ya He,Xin Hao,Liyuan Wang
机构: Tsinghua University (清华大学); University of California, Irvine (加州大学欧文分校)
类目: Computation and Language (cs.CL)
备注:
Abstract:Theory of mind (ToM), the ability to reason about others’ mental states, underlies effective communication and coordination. Its higher-order form enables reasoning over nested perspectives in strategic interactions. However, existing evaluations often equate belief-nesting depth with reasoning order, allowing nominally higher-order questions to admit lower-order shortcuts. We formalize effective order through the distinct events supporting successive belief prefixes, revealing widespread order inflation across existing benchmarks. We introduce HopToM, with certified full-order questions and controls for lower-order alternatives. Under direct prompting, mean accuracy across four frontier LLMs falls from 73.5% at first order to 20.0% at fourth order, and human participants also struggle. Causal interventions link a recurring class of errors to evidence for shallower belief prefixes. We introduce HopTrace, a plug-and-play inference-time method that encodes each interaction once into perspective records and resolves nested queries deterministically. On certified narrative questions, it improves fourth- and fifth-order accuracy by 22.0 and 38.5 percentage points over strong prompting baselines, while using 20-39 times fewer tokens than multi-call methods. These results sharpen how higher-order ToM is evaluated and offer a practical way to improve perspective reasoning in interactive agents.
[NLP-102] Many Right Answers One Training Example: Removing Responder Noise from GRPO for LLM Recommenders
【速读】： 该论文旨在解决在使用生成式 AI 进行推荐任务时，如何有效设计强化学习（Reinforcement Learning, RL）的奖励机制以应对存在多个合理答案的场景。具体而言，当用户在 Reddit 上发起推荐请求时，不同回复者会提供各异的推荐列表，而现有方法如 Rank-GRPO 仅选取单个回复作为参考标准进行奖励，导致奖励信号主要反映“哪个回复被选中”而非“推荐列表质量”，其优势函数方差中高达 85% 反映的是选择偏好而非列表本身的优劣。为解决此问题，作者提出 ReMaRk：将每个推荐请求视为独立训练样本，并对生成的推荐列表统一评估其与所有回复标题集合的匹配程度，从而实现对所有有效答案的联合奖励。该方案的关键在于将训练目标从“匹配某一个回答”转变为“与所有合理答案的并集保持一致”，并通过均衡采样请求来提升训练效率。实验表明，在七种模型架构和三个数据集上，ReMaRk 在仅需 25% 到 53% 的训练步数即可达到传统每回复训练 2000 步的验证召回率，且在相同预算下表现更优；长期训练后，ReMaRk 始终不显著落后于 Rank-GRPO，且在其中四组对比中以更少的 GPU 小时达到其最佳性能。该方法同样适用于 ProtoQA 等具有大量有效人类答案的任务，表明在存在多个合理答案的场景中，应采用联合奖励策略以提升强化学习的鲁棒性与效率。
链接: https://arxiv.org/abs/2610.17771
作者: Xin Li,Hao Jiang,Jinghao Guo,Weisi Lin,Chau Yuen
机构: Nanyang Technological University, Singapore(南洋理工大学); https://lixin.ai/ReMaRk
类目: Computation and Language (cs.CL)
备注: Project Page: this https URL
Abstract:A recommendation request on Reddit draws several replies, each naming different titles. When an LLM recommender is trained on such threads with reinforcement learning (RL), which replies should a generated list be rewarded against? The state-of-the-art recipe, Rank-GRPO, picks one reply per training example, so the advantage mostly records which reply was drawn (85% of its variance) rather than how good the list is. We argue that RL should train on the request, not on each answer, and propose ReMaRk: it scores every list once against the union of all replies’ titles and samples each request equally. This makes RL cheaper. On seven backbones and three datasets, ReMaRk reaches the validation recall of 2,000 per-reply RL steps in 25 to 53% of the steps and leads at that budget; trained until neither improves, it is never significantly behind Rank-GRPO and in four of seven pairs reaches Rank-GRPO’s best with fewer GPU-hours. The same reward helps on ProtoQA, where about 100 people answer each question, which points to a simple rule for RL with several valid human answers: reward them jointly.
[NLP-103] Efficiently Learning To Reason or Not to Reason : Root-token Policy Optimization for Adaptive Thinking ACL2026
【速读】： 该论文旨在解决大推理模型（Large Reasoning Models, LRMs）在处理不同难度问题时存在的“过度思考”（overthinking）问题，即对所有输入均分配相同量的高计算资源，导致效率低下。其核心挑战在于如何在不依赖封闭式路由机制的前提下，使开源的大规模推理模型具备自适应推理能力。为此，论文提出一种名为根令牌策略优化（RPO, Root-token Policy Optimization）的新框架，其关键在于仅通过训练初始根令牌（如是否调用think标签）来实现模型对推理过程的自主决策，利用组间相对奖励与组内优势进行优化。该方法聚焦于推理流程中的关键分支点，显著降低了训练开销和显存占用，在多个模型家族与规模下，仅需先前自适应推理方法约2%的训练算力即可实现对问题难度的感知与自适应推理，从而在保持性能的同时大幅提升了效率。
链接: https://arxiv.org/abs/2610.17767
作者: Taehyeon Kim,Hyunsoo Lee,Youngsoo Jang,Moontae Lee
机构: LG AI Research( LG人工智能研究院); UNIST(韩国科学技术院); University of Illinois Chicago(伊利诺伊大学芝加哥分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: ACL 2026 Oral
Abstract:Large reasoning models (LRMs) achieve strong performance by externalizing explicit reasoning traces before producing the answer, yet suffer from overthinking challenge that allocates uniformly heavy computation to queries of varying difficulty. While proprietary models mitigate this via opaque routing, open-source LRMs still lack an efficient mechanism to internalize adaptive reasoning due to both expensive training cost and limited disclosure of training recipes. In response, we introduce RPO (Root-token Policy Optimization), a framework that enables LRMs to self-determine when to reason by training only the initial root token (e.g., whether to invoke the think tag) via group relative reward and group-wise advantages. By focusing on this pivotal branching point, RPO drastically reduces training overhead and VRAM usage. Across multiple model families and scales, RPO learns difficulty-aware adaptive thinking at just ~2% of the training compute of prior adaptive-reasoning methods.
[NLP-104] Connecting the Dots: Enduring Performance Gaps in ASR Systems Between Standard and Minority English Varieties
【速读】： 该论文旨在解决自动语音识别（ASR）系统在不同种族、性别、社会经济地位及语言变体群体间存在的性能差异问题，尤其关注口音相关差异的长期演变趋势。研究挑战了“随着模型规模扩大和通用性提升，口音偏差将自然缓解”的主流假设，通过纵向评估2021至2025年间发布的16个主流ASR系统在8个涵盖多样英语口音数据集上的表现，发现主流美式英语（Mainstream US English, MUSE）与少数群体口音之间的性能差距具有顽固性，未随时间显著缩小。具体而言，尽管系统对MUSE的识别性能保持稳定，但少数群体口音的识别错误率在不同代际系统中波动剧烈，甚至出现高达65%绝对值的错误率恶化。研究指出，当前标准的ASR评估基准隐含了一种偏袒主流美式英语的语言政策，导致技术发展基础设施存在系统性缺陷，加剧了对代表性不足说话者的不平等影响。其解决方案的关键在于重构评估范式，引入更具包容性的多口音基准，以揭示并纠正语言技术开发中的结构性偏见。
链接: https://arxiv.org/abs/2610.17761
作者: Alexander Metzger,Aruna Srivastava,Ruslan Mukhamedvaleev,Eunjung Yeo,Syed Ishtiaque Ahmed,Nina Markl,Sachin Kumar,Farhan Samir
机构: Koel Labs LLC; University of Texas at Austin; University of Toronto; University of Essex; The Ohio State University
类目: Computation and Language (cs.CL)
备注: Accepted to IEEE SLT 2026. 11 pages including appendix, 1 figure, 6 tables. Code and data: this https URL
Abstract:Automatic Speech Recognition (ASR) system evaluations have consistently revealed performance disparities along lines of race, gender, socioeconomic status, and language variety. While these disparities are well-documented, they are often (implicitly or explicitly) treated as issues that subsequent, larger, and more universal models will correct. We challenge this assumption of steady progress in reducing accent-related performance disparities, conducting a longitudinal evaluation of 16 prominent systems released between 2021-2025 across 8 datasets representing diverse English accent varieties. We find that the performance disparity between Mainstream US English (MUSE) and minority accents is stubborn, not reliably declining as a function of time. Specifically, while systems are reliably performant on MUSE, minority accents are subject to risks of large performance degradations between generations of systems, sometimes worsening by 65% absolute Word Error Rate. We argue that standard ASR benchmarks enact an implicit language policy privileging Mainstream US English varieties. As such, these performance disparities reflect systemic failures of language technology development infrastructure that disproportionately affect underrepresented speaker populations.
[NLP-105] Auditing Failure Attribution: Label Predictability and Judge Behavior
【速读】： 该论文旨在解决生成式 AI（Generative AI）在失败归因基准测试中可能出现的虚假信号（spurious signals）问题，即大语言模型（LLM）裁判可能依赖表面线索（surface cues）而非真正进行跨步骤检查（cross-step check）来判断故障原因，从而导致评估结果失真。其核心解决方案是提出四项系统性测试以区分模型是否真正执行了逻辑检查行为与仅依赖提示线索：输入限制测试（Input restriction）用于检验部分输入能否预测标注结果；线索跟随测试（Cue following）通过改变线索内容观察判断结果是否随之变化；残差准确率评分（Residual accuracy scores）用于衡量仅基于线索的预测器在何处失效；聚合测试（Aggregation）则考察失败模式或指标变化是否影响排名。实证发现，在三个基准上，简单预测器的表现可媲美甚至超越GPT-4.1；在两个WhoWhen Pro系统中，仅读取单步推理的分类器即可识别95%-100%的故障步骤，表明存在强输入依赖性；在错误注入实验中，移除故障步骤后的推理信息后，微调模型性能显著提升，而添加任务相关推理后，微调模型性能下降31-99点，但零样本模型仅损失不超过9点，凸显线索依赖性差异；在Aegis基准上，对所有代理统一归责虽在微平均F1上持平GPT-4.1，但在精确集准确率上落后19分，揭示聚合方式的影响。因此，作者建议在开发失败归因基准时必须报告上述四项测试结果，以确保评估的有效性与可信度。
链接: https://arxiv.org/abs/2610.17747
作者: Tianyang Xu,Jing Gao
机构: Purdue University (普渡大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 26 pages
Abstract:Failure-attribution benchmarks ask an LLM judge which step or agent caused a failure. A high score should mean the judge checked the faulty step against the run (a cross-step check). However, spurious signals (surface cues) that predict the annotation can produce a high score. In this paper, we propose four tests to separate checking from cue use. Input restriction asks whether partial input predicts the annotation. Cue following checks whether moved cues change answers. Residual accuracy scores judges where cue-only predictors fail. Aggregation asks whether the failure mix or metric changes rankings. Real observations motivate these tests: on three benchmarks, simple predictors match or beat GPT-4.1. In two WhoWhen Pro systems, a classifier reading one step’s reasoning finds 95-100% of faulty steps, motivating input restriction. In one of them, error injection leaves no reasoning after the faulty step; judges fine-tuned on its labels beat GPT-4.1 by 41 points. Adding task-related reasoning after that step moves this cue: fine-tuned judges lose 31-99 points, but zero-shot judges lose at most 9, motivating cue following. On Aegis, blaming every agent matches GPT-4.1 on micro-F1 but trails by 19 points in exact-set accuracy, motivating aggregation. We recommend reporting these tests when developing failure-attribution benchmarks.
[NLP-106] Geometric Structure Predicts Per-Example Neighborhood Loss Under PCA Truncation
【速读】： 该论文旨在解决文本嵌入（text embeddings）在降维过程中，尽管整体平均性能保持稳定，但个体样本的邻域稳定性存在显著差异的问题。传统基于平均性能的评估指标无法揭示这种局部退化是否均匀分布，因此难以识别哪些样本在压缩后更易受损。其解决方案的关键在于通过主成分分析（PCA）截断下的每例邻域稳定性分析，揭示影响个体样本保留能力的几何因素。研究发现，中心向量能量保留比例是预测邻域稳定性的最强因子，尤其在CBOW和RI模型中表现突出；而邻域对比度（neighborhood contrast）在SkipGram及现代句向量模型（如GTE）中更具解释力；局部密度虽在边际分析中有一定关联，但在控制能量与对比度后贡献极小（R²提升仅0.001–0.010）。此外，局部内在维度（local intrinsic dimensionality）在现代句向量模型中对邻域保留的解释力优于局部密度。实证结果表明，在SciFact数据集上，查询密度可有效预测激进压缩下的检索性能下降；而在NFCorpus上，32维主成分的主导效应未复现，且方差约束测试未能排除该现象的零假设。综上，该研究强调：平均压缩性能掩盖了系统性的个体差异，而嵌入空间的几何特性（如能量分布、对比度、内在维度）能够有效揭示这些隐藏的退化模式。
链接: https://arxiv.org/abs/2610.17742
作者: Sriram Loganathan,Gokul Anand,Ragavan Arivazhagan
机构: San Jose State University(圣何塞州立大学); San Jose, CA, USA
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Dimensionality reduction can preserve the average performance of text embeddings even when a substantial fraction of their dimensions is removed. Aggregate metrics, however, do not reveal whether degradation is distributed uniformly across examples. We study per-example neighborhood stability under PCA truncation across static word embeddings and modern sentence encoders. Per-item neighbor retention varies widely and is predictable from geometric properties of the representation and its alignment with the retained PCA subspace. The fraction of centered-vector energy retained by the truncated subspace is strongest for CBOW and RI, with neighborhood contrast leading on SkipGram and the modern/Matryoshka settings (GTE approximately tied), while local density is associated with retention in marginal analysis but adds little unique information once energy and contrast are controlled (R-squared increments 0.001 to 0.010). Local intrinsic dimensionality accounts for more variation than density on modern sentence encoders in pairwise comparison. On SciFact, query density predicts retrieval degradation under aggressive compression; on NFCorpus the primary 32d association does not replicate, and a direct variance-restriction test leaves that null unresolved. Average compression performance conceals systematic per-example variation that geometry makes visible.
[NLP-107] Understanding Machine Taste: Aesthetic Agreement Across Foundation Models in K-pop
【速读】： 该论文旨在解决生成式 AI（Generative AI）在创意产业中对创作内容的审美偏好理解问题，以实现人工智能辅助的创意决策。其核心挑战在于揭示基础模型在评估创意内容时所表现出的“机器品味”（machine taste），尤其是在跨模态（如音频与视觉）评价一致性及与真实世界受众偏好关联性方面的不足。解决方案的关键在于构建一个基于成对排序（pairwise ranking）的美学评估框架，利用K-pop歌曲及其音乐视频作为测试案例，系统分析模型间的成对一致性、群体共识以及跨模态偏好对齐程度。研究发现，视觉模态的模型间一致性和群体共识显著高于音频模态，但跨模态偏好仍存在较弱对齐；进一步结合历史YouTube观看数据验证，发现视觉共识排名与现实流行度呈正相关，但相关性因艺人团体而异。这一成果为未来融合内容质量、艺术家身份、受众偏好与市场动态的AI辅助创意决策提供了初步理论基础与实证依据。
链接: https://arxiv.org/abs/2610.17731
作者: Kaishen Wang,Yucheng Pan,Xuehao Cui,Hanling Tian,Heng Huang
机构: University of Maryland, College Park (马里兰大学学院公园分校)
类目: Computation and Language (cs.CL)
备注: 17 pages, 4 figures, 14 tables
Abstract:As foundation models enter creative industries, understanding their aesthetic preferences becomes an important step toward AI-assisted creative decision making. In this work, we investigate how foundation models evaluate creative content and characterize their ``machine taste’’ using K-pop songs and music videos as a testbed. We formulate aesthetic evaluation as a pairwise ranking problem and systematically examine pairwise inter-model agreement, group-level consensus, and cross-modal preference alignment. Our results reveal stronger pairwise inter-model agreement and group-level consensus in visual evaluation than in audio evaluation, while preferences across the two modalities remain weakly aligned. We further examine how visual consensus rankings relate to real-world popularity using historical YouTube viewership data, finding positive but varying correlations across artist groups. These findings provide initial insights into machine aesthetic preferences and lay the groundwork for future AI-assisted creative decision making that considers creative content, artist identity, audience preferences, and market dynamics.
[NLP-108] SkillScreen: High-Recall Screening of Malicious Agent Skills with Decision Models
【速读】： 该论文旨在解决生成式AI（Generative AI）中代理技能（agent skills）的恶意行为检测问题，即合法自动化与恶意行为在自然语言指令层面表现高度相似，导致安全筛查困难。其核心挑战在于：如何在保证高召回率（99%覆盖未见攻击家族）的前提下，显著降低误报率，从而减轻人工审查负担。解决方案的关键在于提出SkillScreen——一种轻量级决策模型，通过单次前向传播直接从技能的URL输入中评分其恶意程度，无需依赖复杂的文本分析或外部大语言模型（LLM）提示。相较于词级别分类器和提示型LLM裁判，SkillScreen在未见攻击家族上仅误标20–25%的良性技能，显著提升筛选效率。此外，在相同训练协议下，其对未见过的技能来源具有更强的泛化能力，优于全微调编码器模型。然而研究也揭示，当面对全新良性技能源时，误报率急剧上升，表明识别新攻击并不等同于可靠筛查新技能集合。研究结果证明，针对任务定制的决策模型在实现高召回率技能筛查方面具有显著优势。
链接: https://arxiv.org/abs/2610.17727
作者: Jerrold Huang,Junwei Quan,Saiyue Lyu,Zihan Guo,Yingqiang Ge
机构: University of British Columbia(不列颠哥伦比亚大学); University of Toronto(多伦多大学); Rutgers University(罗格斯大学)
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 21 pages, including references and appendices
Abstract:Agent skills turn natural-language instructions into actions, creating a security challenge: legitimate automation and malicious behaviour can look deceptively similar. A practical screen must catch nearly every attack while keeping the review load manageable. We present SkillScreen, a compact decision model trained to score skill maliciousness directly from this http URL in a single forward pass. On MaliciousSkillBench, SkillScreen flags 20-25% of benign skills at 99% recall on unseen attack families, substantially fewer than the word-level classifier and prompted LLM judges tested. Under the same training protocol, it also generalizes better than a fully fine-tuned encoder to unseen sources. Yet unfamiliar benign sources sharply increase false positives, showing that recognizing new attacks does not ensure reliable screening of new skill collections. Our results demonstrate the value of task-trained decision models for high-recall skill screening. Project page: this https URL
[NLP-109] Pushback in a Haystack: Language Models Respond to Factual Errors Ambiguously and Ineffectively
【速读】： 该论文旨在解决生成式语言模型（Generative Language Models, GLMs）在面对用户错误输入时，虽表现出一定程度的“抗顺从性”（pushback），但其表达方式仍可能导致用户误解或无法有效纠正事实性认知偏差的问题。核心问题是：即使模型试图反驳用户的错误主张，其推诿方式的模糊性与间接性可能削弱纠错效果，从而未能真正解决“谄媚倾向”（sycophancy）的本质。解决方案的关键在于提出一种分词级、经人工验证的响应标注框架，用于精细化分析模型在多语言场景下（英语、德语、汉语、印地语、约鲁巴语）对错误输入的反驳策略。研究发现，模型普遍采用间接表述并混合肯定成分，尤其在印地语和约鲁巴语中表现更明显；大规模人类实验（n=300）进一步证实，这种推诿的模糊性会降低用户对错误信念的修正能力，表明仅具备“推拒行为”并不足以实现有效纠偏，关键在于推拒策略的清晰度与可理解性。
链接: https://arxiv.org/abs/2610.17703
作者: Li Lucy,Myra Cheng,Svenja Guhr,Kabir Ahuja,Orevaoghene Ahia,Sameep Chattopadhyay,Yike Wang,Valentin Hofmann,Noah A. Smith,Yulia Tsvetkov
机构: University of Washington; Allen Institute for AI; LMU Munich; UC Berkeley; Stanford University; University of Wisconsin-Madison
类目: Computation and Language (cs.CL)
备注: 28 pages, 15 figures
Abstract:The sycophancy of language models, or their failure to push back against faulty user input, is now widely studied. However, if LMs do push back, is sycophancy solved? We show that LMs vary in how they present pushback, and thus vary in how effectively they correct user beliefs. We introduce a phrase-level, human-validated response labeling framework to uncover fine-grained insights into how LMs push against and/or affirm false inputs. We develop and apply our framework on model responses in five languages: English, German, Chinese, Hindi, and Yoruba. We find that when models do push back against users’ false assertions, they tend to use indirect phrasing and mix in affirmation, with more indirectness in Hindi and Yoruba than in other languages considered. In a large-scale human study (n = 300), we find that the pushback ambiguity we identified may fail to correct and prevent users’ factual misconceptions.
[NLP-110] Encoding Context into LLM Weights: What Makes It Context?
【速读】： 该论文旨在解决长上下文（longer context）下大语言模型（Large Language Models, LLMs）推理成本急剧上升的问题。传统方法通过扩展输入上下文长度来增强模型的上下文理解能力，但会导致计算开销过大。现有的一种替代方案是采用测试时更新（test-time updates）将上下文信息编码为模型权重的临时变化，然而这种做法会模糊新编码信息与模型原有知识之间的界限，导致模型在后续推理中无法正确识别和利用这些信息作为上下文。为此，论文提出的关键解决方案包括两部分：首先，在编码阶段为每个上下文分配唯一标识符（identifier），从而将新注入的信息与预训练模型中的通用知识区分开来；其次，设计一种显式的提示框架（prompting framework），引导模型在推理时主动将召回的上下文信息用作条件依据。实验证明，该方法在多种受控任务和实际基准上均优于仅使用现成LLMs的基线方法，显著提升了上下文编码的有效性与可利用性。
链接: https://arxiv.org/abs/2610.17686
作者: Yue Wang,Qizhou Wang,Feng Liu,Gang Niu,Masashi Sugiyama,Bo Han
机构: The University of Melbourne(墨尔本大学); RIKEN Center for Advanced Intelligence Project(理化学研究所先进智能项目中心); The University of Tokyo(东京大学); TMLR Group, Hong Kong Baptist University(香港浸会大学TMLR组)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Longer contexts allow large language models (LLMs) to condition their predictions on more information, but inference costs grow prohibitively with context length. An emerging alternative uses test-time updates to encode context as temporary changes to model weights. However, encoding this information into weights that already represent general knowledge obscures its role as context, so the LLM may fail to use it properly to condition subsequent inference. To restore this contextual role, we first associate each context with an identifier during encoding, thereby distinguishing the newly encoded information from pre-existing knowledge. Invoking the identifier at inference then enables the LLM to recite the encoded context. Nevertheless, successful recitation does not ensure that the LLM will use the recalled information as context. We therefore further introduce a prompting framework that explicitly guides the LLM to use this information as context. Across diverse controlled tasks and practical benchmarks, our method outperformed the baseline using only off-the-shelf LLMs, showing its advances in context encoding.
[NLP-111] LatentRiskBench: Measuring State-Dependent Risks in Enterprise Agents EMNLP2026
【速读】： 该论文旨在解决企业级智能体（enterprise agents）在复杂工作流中面临的状态依赖型安全风险（state-dependent safety risks）问题。传统安全评估基准主要关注局部动作层面的风险（action-local risks），无法有效识别那些仅在特定工作流状态背景下才会出现的安全违规行为——即同一动作在不同上下文中可能表现为合规或违规。其核心挑战在于，现有方法依赖于动作级别的显式线索进行安全判断，而忽视了对隐含工作流状态的推理能力。为此，本文提出LatentRiskBench，一个专门用于评估状态依赖型风险的基准框架，通过配对具有相同终止动作但处于不同潜在策略相关状态的轨迹（即“危险轨迹”与“良性轨迹”），迫使模型必须重建完整的工作流上下文才能做出准确判断，从而避免对动作级特征的简单依赖。关键创新在于引入一套新的残余风险度量（residual risk, RR）指标，以量化模型在状态感知上的不足。实验表明，基于动作层面的评估仅能检测到0.09的状态依赖违规，而采用状态层面评估时准确率提升至0.96；此外，案例研究进一步验证了该基准所嵌入的安全度量能够更真实地反映实际部署场景中的安全性。
链接: https://arxiv.org/abs/2610.17685
作者: Jingyue Cong,Estrid He,Zhuang Li,Nan Wu,Li Shen,Yameng Peng,Xinshu Li,Andy Song
机构: RMIT University (皇家墨尔本理工大学); YepAI; Monash University (莫纳什大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted to EMNLP 2026 Industry Track. 18 pages, 3 figures
Abstract:Enterprise agents face safety risks that cannot be determined from individual actions alone. Existing safety benchmarks largely evaluate action-local risks, overlooking failures whose safety depends on workflow state. In such state-dependent risks, the same action may be safe in one context but violate policy in another. Hence, we introduce LatentRiskBench for evaluating such state-dependent risks in enterprise settings. It pairs risky trajectories with benign counterparts that share the same terminal action but differ in latent policy-relevant states, requiring models to reconstruct workflow context rather than rely on action-level cues. We also introduce new metrics for state-dependent risk evaluation, including a suite of residual risk (RR) measures. Experiments on the benchmark show that action-level evaluation detects only 0.09 of state-dependent violations, whereas state-level evaluation reaches 0.96. Existing guardrail methods largely overlook such risks unless they are explicitly applied to workflow states. Case studies on a commercial deployed scenario further show that the safety metrics embedded in the benchmark provide a more realistic measure of safety for enterprise agents. Our dataset and code are available at this https URL.
[NLP-112] DeltaTTG: Training Multi-Turn Agents with a Token-to-Go Progress Meter
【速读】： 该论文旨在解决多轮智能体强化学习中稀疏终端奖励（sparse terminal rewards）导致的步骤级奖励信号缺失问题，尤其在缺乏任务特定进度规则时难以有效评估每一步进展。其核心解决方案是提出DeltaTTG方法，通过引入一个共享的、可复用的“进度计”（progress meter），利用来自多样化策略生成轨迹的低成本监督信号来学习通用的进度指示。该方法的关键在于设计了一个可观测的代理指标——“剩余生成词元数”（Token-to-Go, TTG），作为完成任务所需剩余努力的近似度量，并在策略训练过程中冻结该进度计，仅通过其估计值的变化对各步骤进行优势重分配，从而在不改变原始奖励符号和平均尺度的前提下实现细粒度的步骤级指导。实验表明，该方法在ALFWorld、PaperSearchQA和BFCLv3等任务上显著优于GRPO及其他主流强化学习基线，分别提升5.7和7.2分（使用Qwen3-4B与Llama-3.1-8B-Instruct模型），同时保持接近GRPO的训练资源开销，且展现出更优的样本效率与跨域泛化能力，验证了进度计在区分步骤质量及引导测试阶段动作选择方面的有效性。
链接: https://arxiv.org/abs/2610.17666
作者: Xuyan Ye,Xiyan Fu,Shengda Fan,Jiawei Cui,Yankai Lin,Wei Lu
机构: Nanyang Technological University (南洋理工大学); Renmin University of China (中国人民大学)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Reinforcement learning with verifiable rewards (RLVR) for multi-turn agents relies on sparse terminal rewards that say little about individual steps. Within-group step comparisons depend on reliable state matching and sufficiently diverse sampled outcomes, while direct step scoring may require costly external evaluators, human annotations, or task-specific heuristics. We introduce DeltaTTG to learn a reusable progress signal through a shared progress meter, using low-cost supervision from diverse policies’ recorded trajectories without task-specific progress rules. Its supervision target, Token-to-Go (TTG), the number of tokens remaining to be generated until task completion, serves as an observable proxy for remaining completion effort. During actor training, changes in the frozen meter’s estimates redistribute outcome-derived advantages across steps without altering the reward, preserving each trajectory’s advantage sign and token-average scale. On ALFWorld, PaperSearchQA, and BFCLv3, DeltaTTG improves the average score over GRPO by 5.7 and 7.2 points with Qwen3-4B and Llama-3.1-8B-Instruct, respectively, outperforms all compared RL baselines on average, and keeps actor-training GPU-hours close to GRPO. Further experiments show improved sample efficiency and out-of-domain performance, and support the meter’s ability to distinguish step quality and guide test-time action selection across actor sizes. The code is available at this https URL.
[NLP-113] he Image Still Has a Say: Final-Answer Repair for Multimodal Chain-of-Thought
【速读】： 该论文旨在解决链式思维提示（Chain-of-Thought Prompting）在多模态推理中因推理过程偏离图像信息而导致最终答案错误的问题，以及现有方法在生成答案后缺乏自我修正机制的局限性。其核心挑战在于：尽管推理轨迹（reasoning trace）在生成答案时已固定，但图像信息在最终决策阶段仍可被访问，因此应利用这一可用信息对最终答案进行修正，而非重新生成推理过程。解决方案的关键是提出一种无需训练的最终答案修复方法——文本到视觉证据重分配（Text-to-Visual Evidence Reallocation, TVER），该方法在保持推理轨迹不变的前提下，通过将原本集中在文本上的注意力权重动态转移至与问题相关的图像区域（尤其是中间到深层网络特征中的关键图像块），实现对错误选项的纠正。TVER基于发现：错误选项可能在推理早期被过早锁定，而正确选项在深层特征中仍具有更强的图像支持。该方法在多个基准测试和模型架构上均表现出显著性能提升，适用于多项选择与开放式问答任务。
链接: https://arxiv.org/abs/2610.17663
作者: Yulan Pan,Xiaoyu Xu,Zhihong Shen
机构: Fuzhou University (福州大学)
类目: Computation and Language (cs.CL)
备注: 12 pages, 5 figures
Abstract:Although chain-of-thought prompting has improved multimodal reasoning, it is still difficult to produce correct final answers, as the reasoning can drift away from the image. Besides, the trace is already fixed when the answer is generated, leaving no room for self-correction. Considering the image is still available at the final-answer step, we think it is reasonable to fix the answer there without rewriting the reasoning. As such, the problem becomes to correct the final answer using the image under fixed reasoning. Unlike methods that improve trace generation through training, resampling, or decoding-time grounding, we repair the final-answer readout while keeping the reasoning trace fixed. The motivation is based on our finding that a wrong option can become prematurely committed in the reasoning while the correct option still has stronger image support in middle-to-deep layers. Thus, we propose Text-to-Visual Evidence Reallocation (TVER), a training-free method for final-answer repair. Specifically, TVER selects answer-related text and question-matched image patches, and moves bounded attention from text to patches. Furthermore, the patches are chosen in a middle-to-deep layer window based on question relevance. Our method improves answers across benchmarks and backbones, with gains in both multiple-choice and open-answer evaluations.
[NLP-114] Worth Their Weights? Evaluating Generative Models for Discourse Parsing
【速读】： 该论文旨在解决生成式语言模型（Generative Language Models）在修辞结构理论（Rhetorical Structure Theory, RST）解析任务中是否优于传统判别式模型的问题。研究发现，在参数量相当的情况下，判别式解析器的表现显著优于生成式模型；尽管更大规模的生成式模型在某些情况下可实现更优性能，但其整体表现仍不及判别式方法。然而，生成式模型在基本话语单元（Elementary Discourse Unit, EDU）分割任务上展现出显著优势，并在端到端RST解析与分割任务上取得了新的基准性能（state-of-the-art），这表明生成式模型在特定子任务中具有独特潜力。解决方案的关键在于通过对比不同架构在相同参数规模下的表现，揭示生成式模型在EDU分割中的优势，并推动端到端解析性能的提升。
链接: https://arxiv.org/abs/2610.17658
作者: Luke Gessler,Amir Zeldes
机构: Indiana University Bloomington(印第安纳大学伯明顿分校); Georgetown University(乔治城大学)
类目: Computation and Language (cs.CL)
备注: Under review
Abstract:Rhetorical Structure Theory (RST) parsing has historically been dominated by discriminative models based on pretrained encoder-only language models. The recent availability of high-quality generative language models has made it unclear whether RST parsing is better approached as a generative task. In this work, we investigate whether generative language models are better at parsing RST than traditional discriminative ones. We find that at matched parameter counts, discriminative parsers perform much better than generative ones, though much larger generative models sometimes outperform them. Generative models hold a particular advantage in elementary discourse unit (EDU) segmentation, and we report new state-of-the-art results on the RST-DT benchmark in end-to-end parsing as well as segmentation.
[NLP-115] Decouple to Forget: Semantic Mediation for Sensitive Unlearning in Large Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在实现机器遗忘（Machine Unlearning）时因粒度粗略而导致的性能损失问题。现有方法通常以整个响应为遗忘单位，当需删除的内容仅为嵌入于无害解释中的少量敏感事实时，这种粗粒度处理会不可避免地破坏模型的通用能力。为此，本文提出一种名为“解耦遗忘”（Decouple to Forget, D2F）的目标化遗忘方法，其核心在于借鉴语义中介（Semantic Mediation）思想，将敏感的事实内容与非敏感的语义意义及表面表达形式进行解耦。针对每个需遗忘的样本，D2F 从回答中提取事实级别的“中介表示”（mediator representation），识别出敏感部分并仅移除这些内容，同时保留其余结构完整性；随后基于该中介生成反事实回答，以保留非敏感内容但排除敏感事实，并通过基于偏好（preference-based）的对比优化策略，在降低信息泄露风险的同时维持回答的有用性与流畅性。实验结果表明，D2F 在多个基准数据集上实现了比强基线方法更优的遗忘-保留平衡。
链接: https://arxiv.org/abs/2610.17654
作者: Jingyue Cong,Andy Song,Yang Yu,Tabinda Sarwar,Xinshu Li,Ziqi Xu,Xun Yi,Ke Wang,Bo Han,Estrid He
机构: RMIT University (皇家墨尔本理工大学); Yep AI; Monash University (莫纳什大学); Hong Kong Baptist University (香港浸会大学)
类目: Computation and Language (cs.CL)
备注: 26 pages, 6 figures
Abstract:Machine unlearning for large language models seeks to remove specified knowledge while preserving general capability, yet current methods often treat an entire response as the unit of forgetting. This coarse granularity causes unnecessary utility loss when the forget target is only a small set of sensitive facts embedded within otherwise benign explanations. In this paper, we introduce Decouple to Forget (D2F), a targeted unlearning method inspired by semantic mediation that separates sensitive factual content from non-sensitive meaning and surface expression. For each example to be forgotten, D2F extracts a fact-level “mediator” representation from the answer, identifies which parts are sensitive, and removes only those parts while leaving the rest of the structure intact. We then generate counterfactual answers grounded in this mediator that preserve non-sensitive content but exclude sensitive facts, employing preference-based contrastive optimization to reduce the likelihood of information leakage while maintaining helpfulness and fluency. Experiments across various benchmark datasets show that D2F achieves a better balance between forgetting and retention than strong baseline methods. Our code is available at this https URL.
[NLP-116] DisaggKD: Efficient Knowledge Distillation through Teacher-Student Disaggregation and a Fused Loss
【速读】： 该论文旨在解决知识蒸馏（Knowledge Distillation, KD）框架中因教师模型与学生模型部署在同一GPU上而导致的显存瓶颈问题，尤其在大模型场景下，显存限制严重制约了可训练教师模型的规模。其核心解决方案是提出DisaggKD架构，将教师模型与学生模型分置于不同计算节点：通过流式传输教师模型的隐藏状态至学生端，采用基于行窗口和词表分片的KD损失计算机制，在不生成完整词表输出（full-vocabulary logits）的前提下实现与全词表损失等价的精确计算，并引入预取环（prefetch ring）以协调数据生产者与训练器之间的节奏。该方案结合闭形式的内存与通信流量模型，精准定位不同部署布局下的内存瓶颈。实验表明，在Qwen3.5-35B到9B的模型对上，解耦布局在保持与同驻留运行相同语言模型（LM）损失的同时，训练速度更快；此外，融合后的损失函数还使同驻留训练可在更少节点上实现更高效率，而当学生模型跨多个节点时，解耦布局则展现出显著的内存效率优势。在更大规模的模型对上，由于教师权重与学生状态共同占用显存预算，传统同驻留训练仍会超出显存限制（OOM），而解耦方案仍可成功训练，验证了其在大规模知识蒸馏中的可行性与优越性。
链接: https://arxiv.org/abs/2610.17644
作者: Ruixuan Tu,Yuyang Dong
机构: The University of Tokyo (东京大学); RIKEN AIP (理化学研究所人工智能研究中心); SB Intuitions Corp. (SB意识公司)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Distributed, Parallel, and Cluster Computing (cs.DC)
备注:
Abstract:Knowledge distillation (KD) frameworks place teacher and student on the same GPUs, so per-GPU memory limits the teachers that fit. DisaggKD separates them across nodes: teacher replicas stream hidden states to the student, a KD loss over row windows and vocabulary shards equals the full-vocabulary loss in exact arithmetic without forming the full-vocabulary logits, and a prefetch ring paces producer and trainer. Closed-form memory and traffic models locate the memory limits of each layout. On the Qwen3.5-35B-A3B to 9B pair, the disaggregated layout matches the LM loss of the colocated run with the same loss and is faster than that run; our fused loss also lets a colocated run train on as few nodes at faster speed, while the split is memory efficient when the student spans several nodes. On the larger pairs, colocated runs with a fused loss still run OOM where the split trains, because teacher weights and student states fill the budget.
[NLP-117] Executable Prefix Values for Locating Agent Failures
【速读】： 该论文旨在解决大语言模型（LLM）智能体在调试过程中依赖人类或大模型评判者标记的“首次错误”作为轨迹失败的基准标签所引发的根本性问题，即这些标签究竟是指向失败的根本原因还是仅反映其表征症状。其解决方案的关键在于提出一种可执行的区分机制：通过从失败轨迹中重构的中间状态重启智能体，并利用经过验证的点（certified point of no return, PNR）估计每个步骤的前缀价值 $ V(h_t) $，即从该步开始的成功概率，从而识别出真正的不可逆失败点。在150个SWE-bench Verified任务上，基于两个冻结权重的开源智能体，研究构建了47和60条失败轨迹的价值曲线，并对四名前沿评判者在Traverse评估维度下的标注进行系统性探查。结果表明，不同评判者对失败步骤的一致性极低（仅21/57案例达成一致），且最新评判者在8/10和10/10的成功运行中误标错误；66%的标注步骤并未伴随成功概率的显著下降，仅有7/188的答案正确指出了决定性步骤。更重要的是，基于测量的重启点进行重启动可使离线预期成功率从0.44提升至0.57，真实重启也验证了这一优势；而按评判者标注点重启则导致成功率下降（-0.08至-0.03）。该研究揭示了现有标注范式在定位真正失败根源上的局限性，并为更精确的故障诊断与智能体修复提供了新方法。
链接: https://arxiv.org/abs/2610.17633
作者: Bingyu Shen,Jiaqi Tang,Lan Wei,Yi Sheng,J. Jenny Li,Boyang Li
机构: Kean University (新泽西州立大学); University of South Florida (南佛罗里达大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 46 pages, 12 figures, 60 tables
Abstract:Work on debugging LLM agents treats the “first mistake” marked by a human or LLM judge as ground truth for where a trajectory failed. We ask whether such labels point at causes or symptoms, and make the distinction executable: we restart the agent from reconstructed intermediate states of a failed trajectory and estimate its prefix value V(h_t), the probability of success from each step, with a certified point of no return (PNR) for a failure. On 150 SWE-bench Verified tasks with two frozen open-weight agents, we build such curves for 47 and 60 failed trajectories and probe every step labeled by four frontier judges under the Traverse rubric. We find that the labels do not mark what a controller needs. Judges of different lineages agree on the exact step in only 21/57 cases, and the newest judges flag mistakes in 8/10 and 10/10 successful runs. For 66% of the labels, the agent’s success probability does not drop at the labeled step; only 7/188 answers name the decisive step. Restarting at the measured restart point raises expected success from 0.44 to 0.57 offline, and real restarts confirm the advantage, whereas restarting at a judge’s label does not (-0.08 to -0.03). We release our code, curves, and labels.
[NLP-118] RELAY: Selective Self-Distillation with Adaptive Hand-off for Rubric-Based Reinforcement Learning
【速读】： 该论文旨在解决基于评分量规（rubric-based）的强化学习在奖励设计上的局限性问题：传统方法将多维度的评分标准压缩为单一奖励信号，导致量规所蕴含的细粒度信息丢失，并且在多个响应得分相同时无法提供有效的策略梯度（GRPO reward gradient）。为此，论文提出RELAY框架，其核心解决方案在于引入一个冻结的教师模型（frozen teacher），该教师以量规作为特权上下文（privileged context）对生成文本进行逐标记级监督。通过结合优势值（advantages）与奖励信号，RELAY选择具有正优势和低分零优势的回溯轨迹（rollouts）进行选择性知识蒸馏，并利用蒸馏损失监控来判断何时终止教师模型的使用。该方法分为两个阶段：第一阶段融合来自冻结教师的选择性蒸馏与GRPO；当检测到性能增长出现早期平台期时，自动移除教师模型进入第二阶段，仅使用GRPO继续优化。实验表明，经过单一模型-领域组合校准后的平台期检测器在其他三个不同设置中无需调整即可有效工作，在所有四组实验中均取得最优的域内表现，显著优于GRPO基准，尤其在RaR-Medicine任务上分别提升4.62点（Qwen3-1.7B）和6.02点（Qwen3-4B）。此外，从共享检查点中移除教师反而优于保留或刷新教师，且仅蒸馏覆盖约一半响应标记的精选轨迹，其效果也优于对全部轨迹进行等效平均系数的蒸馏，验证了选择性蒸馏的有效性。
链接: https://arxiv.org/abs/2610.17623
作者: Kunwang Zheng,Xinyu Sun,Yu Cao,Shuocheng Li,Lei Cheng,Ying Liu,Qilong Zhang,Jinlong Chen,Hua Yang,Qianglong Chen
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 17 pages, 8 figures, 13 tables
Abstract:Rubric-based reinforcement learning grades a response against many written criteria but collapses them into one reward, which discards the detail the rubric carries and gives no GRPO reward gradient when a group’s responses tie. A frozen teacher that reads the rubric as privileged context provides token-level supervision, but we observe an early plateau in the gains over GRPO. RELAY uses advantages and rewards to select positive-advantage and low-scoring zero-advantage rollouts, and monitors the distillation loss to decide when to retire the teacher. Stage 1 combines selective distillation from a frozen teacher with GRPO; a plateau detector then removes the teacher, and Stage 2 continues with GRPO alone. Calibrated on one model-domain pair and reused unchanged in three others, the plateau detector fires between steps 122 and 171. RELAY scores highest in-domain among the compared methods in all four settings, ahead of GRPO on RaR-Medicine by 4.62 points with Qwen3-1.7B and 6.02 with Qwen3-4B. On Qwen3-1.7B/Medicine, removing the teacher from a shared checkpoint outperforms keeping, weakening or refreshing it. Distilling selected rollouts covering about half the response tokens also outperforms distilling all rollouts at a matched average coefficient.
[NLP-119] When Does an ASR Error Matter? Task-Role Diagnostics for Tool-Using Voice Agents
【速读】： 该论文旨在解决级联语音代理（cascaded voice agents）中自动语音识别（ASR）错误的下游影响评估问题。传统词错误率（WER）无法区分不同错误对任务执行的实际影响，而实际应用中，同一类错误可能仅导致工具调用参数微小变化，也可能彻底改变意图或操作逻辑。为此，论文提出SpeechEvalX——一种配对诊断评估协议，通过控制编辑的源话语、编辑数量和操作类型，系统分析受影响的任务角色（task role）如何影响结构化工具调用的正确性。评估采用语义严重性、BERT余弦相似度、语义词错误率（Semantic-WER）及二元关键性（binary criticality）等多维度指标进行对比。实验基于594个源话语，结果显示，引入三种不同ASR前端后，联合调用准确率从参考转录的82.3%下降至71.9%-74.9%；其中数字与否定类编辑导致21.9-25.7个百分点的性能损失，显著高于控制类编辑的1.4-2.9个百分点。进一步分析表明，细粒度任务角色标签可将按源组划分的ROC-AUC从0.73提升至0.78，证明下游影响高度依赖于被错误影响的具体任务角色。因此，解决方案的关键在于引入任务角色感知的多维度评估框架，实现对ASR错误影响的精细化建模与量化。
链接: https://arxiv.org/abs/2610.17619
作者: Xiaolong Wang,Shijie Su,Naqi Xiong
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:In cascaded voice agents, ASR errors can have markedly different downstream consequences: a small transcription error may leave a tool call unchanged or alter its intent or arguments. Conventional word error rate (WER), however, does not capture these differences. We introduce SpeechEvalX, a paired diagnostic protocol that evaluates how task roles affected by ASR errors influence structured tool calls. Controlled edits are matched by source utterance, edit count, and operation, while task-role labels are compared with lexical severity, BERT cosine distance, Semantic-WER, and binary criticality. Across 594 source utterances, joint call accuracy falls from 82.3% with reference transcripts to 71.9-74.9% across three ASR front ends. Number and negation edits cause 21.9-25.7 percentage-point losses, versus 1.4-2.9 points for control edits. Detailed task-role labels improve source-grouped ROC-AUC from 0.73 to 0.78 over binary criticality, showing that downstream impact depends strongly on the task role affected.
[NLP-120] StateHarness: Maintaining Evolving Project State for Long-Horizon Coding Agents
【速读】： 该论文旨在解决长周期代码开发（long-horizon coding）中因项目状态信息分散于不断增长的交互历史而导致的局部修改与后续影响脱节、进展难以持续推进的问题。其核心挑战在于如何有效追踪和管理随时间演化的项目状态，包括当前解决的问题、已获得的知识或修改、可能受影响的其他代码构件以及未完成的工作。解决方案的关键是提出StateHarness——一个无需更新模型参数的运行时框架，通过记录代理（agent）的当前问题假设、关键发现与决策、仓库内依赖关系及待办事项，构建动态演进的项目状态视图。该框架利用运行时支持提示（runtime support prompts）自动追踪变更的影响范围（如调用者、测试用例、文档），并维护精简的执行摘要以支持后续决策。当行为信号积累后，独立的诊断会话可回溯历史记录，揭示被忽略的证据。实验表明，StateHarness显著提升了任务完成率与变更一致性，在RefactorBench上全链路完成率从29.4%提升至35.3%，清理检查准确率从87.0%升至95.4%；在五项长周期软件工程任务中，平均奖励由0.427增至0.576，且执行成本显著低于基线方法。这证明了维持显式项目状态对长周期编码的有效性，尽管引入了可量化的执行开销与回归风险。
链接: https://arxiv.org/abs/2610.17616
作者: Chuan Zou,Yujun Zhou,Xiaonan Luo,Xiangliang Zhang
机构: University of Notre Dame(圣母大学); University of Pennsylvania(宾夕法尼亚大学)
类目: oftware Engineering (cs.SE); Computation and Language (cs.CL)
备注: 20 pages, 4 figures, 11 tables
Abstract:Long-horizon coding requires agents to maintain an evolving view of the project they are changing: what problem is currently being solved, what has already been learned or modified, which other artifacts may be affected, and what work remains unresolved. When this information is scattered across a growing interaction history, local edits can become disconnected from their later consequences and progress can be difficult to carry forward. We introduce StateHarness, a runtime framework that maintains evolving project state without updating model parameters. The working agent records its current problem and hypothesis, important findings and decisions, discovered repository relationships, and unfinished work. Runtime support prompts record updates, searches for callers, tests, and documentation that may be affected by a change, and maintains a compact working brief for subsequent decisions. When behavioral signals accumulate, a separate diagnostic session revisits the recorded history and surfaces overlooked evidence. Across three repository-level evaluations, StateHarness improves task completion and several measures of change consistency. On RefactorBench, full-chain completion increases from 29.4% to 35.3% and cleanup-check accuracy from 87.0% to 95.4%. On five LHTB software-engineering tasks, mean reward improves from 0.427 to 0.576, exceeding LongHorizon-Harness at substantially lower recorded execution cost. On ChainSWE, bug-fix rate increases from 35.1% to 39.5% and residual debt decreases from 60 to 43 items, although pass-to-pass regressions increase. These results show that maintaining project state can improve long-horizon coding while introducing measurable execution and regression trade-offs.
[NLP-121] Facts Dont Speak Louder than Words: The Behavioral Essence of Long-CoT Distillation NEURIPS2026
【速读】： 该论文旨在解决长链式思维（Long Chain-of-Thought, Long-CoT）蒸馏机制的内在原理不清晰的问题。现有方法通常依赖拒绝采样来收集多样提示下的推理轨迹，假设事实正确性和广泛的问题覆盖是提升模型推理能力的关键。然而，研究发现仅使用存在错误的推理轨迹或受限提示集进行训练，也能获得接近理想性能的结果。基于这一反直觉现象，作者提出“行为假设”：Long-CoT蒸馏的核心在于与结构化推理模式的行为对齐，而非对事实知识的机械记忆。通过在已掌握问题上仅进行微调，验证了行为对齐可独立于新知识注入实现显著推理性能提升。由此确立推理模式为关键瓶颈后，研究进一步探究如何在不同复杂度层级下最优学习这些模式，揭示出重要权衡关系——基于概率的损失函数在学习简单模式时表现更优，而标准交叉熵损失对于实现复杂链式推理的可扩展性至关重要。上述发现阐明了CoT蒸馏的本质，为未来推理模型的设计提供了原则性指导。
链接: https://arxiv.org/abs/2610.17607
作者: Yongcan Yu,Jian Liang,Lingxiao He,Kuangpu Guo,Yanbo Wang,Shuo Lu,Meng Wang,Ran He
机构: NLPR MAIS, Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所智能信息重点实验室); School of Artificial Intelligence, University of Chinese Academy of Sciences(中国科学院大学人工智能学院); Meituan(美团); University of Science and Technology of China(中国科学技术大学)
类目: Computation and Language (cs.CL)
备注: Accepted to NeurIPS 2026
Abstract:While long chain-of-thought (CoT) distillation significantly improves the reasoning abilities of language models, its underlying mechanism remains poorly understood. Current pipelines generally rely on rejection sampling to gather reasoning traces across diverse prompts, assuming that factual correctness and broad problem coverage are indispensable. Surprisingly, we find that training exclusively on flawed reasoning trajectories or a restricted prompt set yields nearly comparable performance. Driven by this counterintuitive finding, we propose the Behavioral Hypothesis: the essence of Long-CoT distillation is behavioral alignment with structured reasoning patterns, rather than factual knowledge memorization. We then validate this by demonstrating substantial reasoning gains even when models are fine-tuned exclusively on previously mastered problems-effectively isolating behavioral alignment from novel knowledge injection. Having established that reasoning patterns are the true bottleneck, we investigate how to learn them across varying levels of complexity optimally. We reveal a critical interplay: while the recent probability-based loss excels at learning simple patterns, the standard cross-entropy loss is essential for scaling to complex CoTs. We believe these insights demystify CoT distillation and provide a principled foundation for the future development of reasoning models. Code is available on this https URL.
[NLP-122] DR: Tagalog Diacritic Restoration Dataset
【速读】： 该论文旨在解决菲律宾语（Tagalog）中同形异义词（homographs）因缺乏声调标记（tuldik，即变音符号）而导致的语义与发音歧义问题，尤其在当代文本中变音符号几乎完全缺失，限制了变音符号恢复（diacritic restoration）任务的实现。这一问题直接影响语音合成中的韵律选择等下游应用。其解决方案的关键在于构建TDR——一个由40个高频同形异义词组成的标注语料库，基于新闻文本构建，并采用基于搭配（collocation-driven）的检索方法扩展数据集，以缓解少数变体类别间严重的类别不平衡问题。研究进一步表明，TDR所包含的变音符号恢复能力是预训练模型不具备的，且可通过小型、低成本模型或微调的大语言模型（LLM）有效学习，从而为低资源语言的语音处理提供了可行路径。
链接: https://arxiv.org/abs/2610.17597
作者: Ivan Yuri De Leon,Lanz Railey Fermin,Zachary Aldridge Africa,Richell Isaiah Flores
机构: 未知
类目: Computation and Language (cs.CL)
备注: Accepted at O-COCOSDA 2026
Abstract:Homographs are words that share identical spelling but differ in meaning and pronunciation. In Tagalog, these distinctions are disambiguated in writing through tuldik (diacritical marks), which are rarely preserved in contemporary Tagalog text. This scarcity of diacritics limits diacritic restoration, which is useful for downstream tasks such as prosody selection for text-to-speech synthesis. To address this gap, this study introduces TDR, an annotated corpus of 40 frequent Tagalog homographs constructed from news text. It also introduces a collocation-driven retrieval method to expand the dataset and mitigate severe class imbalance among minor variants. Finally, we show that TDR supplies a capability absent from pretrained models, and that this capability can be learned by small, inexpensive models as well as by fine-tuned LLMs.
[NLP-123] anawpamana: Benchmarking Vision-Language Models on Philippine Cultural Heritage Sites
【速读】： 该论文旨在解决当前视觉-语言模型（Vision-Language Models）在识别与描述特定文化场所时，对菲律宾等代表性不足地区文化遗产认知能力不足的问题。其核心挑战在于模型在细粒度视觉区分相似遗产地点方面表现较差，且基于图像输入推断地理知识的能力弱于基于文本输入。解决方案的关键在于构建了“Tanawpamana数据集”——一个基于菲律宾国家文化和艺术委员会（NCCA）Talapamana名录的菲律宾遗产地图像及其元数据集合，并设计了包含4,046个任务的“Tanawpamana基准测试”，系统评估模型在文化遗产识别与知识理解方面的性能，从而揭示并量化现有模型在跨文化、细粒度地理认知上的局限性。
链接: https://arxiv.org/abs/2610.17596
作者: Ivan Yuri De Leon,Hiroki Ouchi,Sakriani Sakti
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Accepted at O-COCOSDA 2026
Abstract:Vision-language models are increasingly used to identify and describe places, yet their recognition and knowledge of specific cultures remain poorly understood, particularly in underrepresented regions such as the Philippines. This study introduces the Tanawpamana Dataset, a repository of Philippine heritage site images with metadata based on the Talapamana registry of the National Commission for Culture and the Arts of the Philippines. We also introduce the Tanawpamana Benchmark, comprising 4,046 tasks designed to evaluate models’ recognition and knowledge of heritage sites. Evaluating seven vision-language models, we find that they struggle more with fine-grained visual discrimination among similar heritage sites than with coarse recognition, and that their geographic knowledge is weaker when inferred from image inputs than from text.
[NLP-124] Dissociating Answerability from Extraction in LLM Question Answering over Radiology Reports
【速读】： 该论文旨在解决放射科报告问答（RadQA）评估中一个关键问题：当前普遍采用的单一F1分数未能区分“错误回答”（false answer）与“过度回避”（over-abstention）这两种不同性质的错误，导致模型性能评估失真。其核心解决方案是提出一种分离式评估方法，将传统F1分数分解为提取能力（extraction）和可回答性（answerability）两个独立维度，并分别报告错误回答率与过度回避率。实验基于13个大语言模型（LLM）在RadQA测试集上的表现，涵盖零样本及三组累积示例提示设置，其中最后一组引入不可回答示例。结果显示，尽管部分模型F1值相近，但其错误回答率差异超过两倍；引入不可回答示例后，多数模型更倾向于回避而非错误回答，且被撤回的答案中仍有相当比例保持较高F1值（≥0.5）。进一步分析表明，错误回答与答案提取质量无明显相关性，但通过聚合多个模型的判断（仅当多数模型均给出答案时才保留回答），可在不显著影响F1的情况下将零样本场景下的错误回答率降低4个百分点。由于可回答性判断与答案提取能力可独立变化，研究主张在放射科问答评估中应分离报告这两类指标，避免仅依赖综合性的F1分数进行评价。
链接: https://arxiv.org/abs/2610.17576
作者: Priyanshu Sinha,Kelsey McDonald
机构: Labcorp(莱克顿公司)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Answering questions from radiology reports involves two decisions: whether the report contains an answer and which span gives it. A false answer presents an unsupported claim as fact; an over-abstention hides what the report says. Yet RadQA results are reported as a single F1 that merges both. We propose an evaluation method that separates them, reporting false-answer and over-abstention rates beside F1 and splitting F1 changes into extraction and answerability terms. We apply it to 13 LLMs on the RadQA test set, zero-shot and with three cumulative demonstration sets, the last adding an unanswerable example. Models with similar F1 differed more than twofold in false-answer rate. The unanswerable example changed when models answered, not how well they extracted: 12 of 13 models abstained more often, and three-quarters of the answers they withdrew had F1 \geq 0.5. Within models, false answers showed no clear association with extraction quality, but acting on the decision itself helped: keeping an answer only when enough other models also answered lowered the zero-shot false-answer rate by 4 points without a detectable F1 change. Because the two decisions can change independently, we suggest radiology QA evaluations report them separately, not through F1 alone.
[NLP-125] EdgeWise: Bringing Long-Horizon Agents to the Edge via Selective Cloud Delegation
【速读】： 该论文旨在解决长时程智能体任务中因持续调用云端大语言模型（LLM）导致的高昂推理开销问题。尽管并非所有决策步骤均需同等强度的模型能力，现有方法仍普遍在每一步执行全量LLM推理，造成资源浪费。其核心解决方案是提出一种边缘-云协同框架EdgeWise，通过在本地设备部署轻量级模型（SLM）作为主要求解器，并仅在关键步骤中按需将任务委托给云端LLM，从而实现性能与成本之间的高效平衡。该方案的关键在于引入一种基于步骤级强化学习（step-level RL）的自主委托机制，通过细粒度的信用分配策略对每一次委托决策进行联合优化，使代理在任务求解与资源节约之间达成最优权衡。此外，研究还设计了一种基于外部评判器的委托策略，系统性地探究了委托时机、决策方式、上下文共享范围及云端调用频率等关键设计要素。实验结果表明，EdgeWise在SWE-Rebench和Terminal-Bench 2.1上分别实现最高62%和34%的推理成本降低，同时保持或超越纯LLM执行的性能表现，并在实际边缘设备部署中验证了其有效性。
链接: https://arxiv.org/abs/2610.17560
作者: Zeru Shi,Jindong Jiang,Ryo Hachiuma,Yonggan Fu,Byung-Kwan Lee,Peter Belcak,Saurav Muralidharan,Ayush Agarwal,Ruixiang Tang,Ambrish Dantrey,Pavlo Molchanov,Anjul Patney,Subhashree Radhakrishnan
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Complex and long-horizon agent tasks often rely on frontier models deployed in the cloud to achieve strong performance. However, using these models for every step of a long-horizon trajectory incurs substantial inference cost, even though not all steps may require the same level of model capability. In this paper, we find that allocating the LLM to only a small subset of key steps can achieve performance comparable to full LLM execution while significantly reducing inference cost. Motivated by this observation, we propose EdgeWise, an edge-cloud collaboration framework that deploys an SLM on the local device as the primary solver and selectively delegates critical steps to a cloud LLM when stronger model capacity is needed. To enable autonomous delegation without additional models at inference time, we introduce a step-level RL method that jointly optimizes task solving and delegation, with fine-grained credit assignment for individual delegation decisions. We also introduce an external judge-based delegation strategy and use it to systematically investigate key design choices in edge-cloud collaboration, including when to delegate, how delegation decisions should be made, what context should be shared, and how frequently the cloud LLM should be invoked. Experiments show that EdgeWise matches or outperforms pure-LLM execution while reducing inference cost up to 62% on SWE-Rebench and 34% on Terminal-Bench 2.1. We further deploy EdgeWise on a local workstation, demonstrating comparable cost savings in a practical edge-cloud setting.
[NLP-126] Many Steps into One: Reasoning Contraction and the Limits of Safe Memory Reuse for LLM Agents
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在长期推理过程中因记忆（memory）过时而导致的错误累积问题，即“记忆陈旧性”（staleness）问题。其核心挑战在于：当外部事实发生变更后，系统依赖的预存记忆可能仍保留过时信息，而传统基于成功/失败计数的验证机制无法及时检测到这种部分或渐进式变化，导致生成大量错误答案。解决方案的关键在于提出一种新型的认证机制（certified memory），通过引入似然比检测器（likelihood-ratio detector） 作为安全冗余，动态评估记忆状态的变化；同时，采用按自身似然比加权失败率的方法，对每个记忆的可靠性进行独立校准，并在检测到事实变更时主动重启证书，必要时直接作废整个记忆。此外，该方法还定义了“认证负载（certification load）”这一关键参数，揭示了在满足安全性前提下可跳过的推理步骤上限，并在齐普夫（Zipf）分布的需求场景下推导出相应的缩放定律，从而在保证安全性的前提下实现推理效率的最大化。
链接: https://arxiv.org/abs/2610.17544
作者: Jianghong Zhou,Hao Rong,Chen Lin,Daoan Zhang
机构: ByteDance(字节跳动); Scale AI
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 57 pages
Abstract:Thought templates, skill libraries, agent memories and caches turn many reasoning steps into one. We call this reasoning contraction. A contracted memory saves reasoning steps and errors until its facts change, then quietly answers wrong. Most systems trust memories by counting successes and failures. We prove that such counters are badly tuned change detectors: a symmetric counter notices a partial change only after a power of its false-alarm horizon, and an unclipped one may never notice it. On five LLMs answering multi-hop questions over edited facts, ExpeL- and ACE-style counters serve 7.0-13.5x more stale answers than certified memories given the same information, and on code tasks with an exact verifier 35-73x more. Safety also has a price. When facts change and each memory is judged only by its own verified outcomes, one number, the certification load, sets how much reasoning any safe memory can skip, and under Zipf demand it yields a scaling law. Our recipe weighs each memory’s failures by its own likelihood ratio, certifies its utility against live reasoning, and restarts the certificate when facts change, with the likelihood-ratio detector as the safety net for unannounced changes; for large edits it also invalidates the memory.
[NLP-127] Everyone Was Right: Lexical Representations in Language Models Are Both Abstract and Holistic
【速读】： 该论文旨在解决语言模型中词汇表示如何组织抽象信息与特定上下文信息的问题。其核心问题是：在预训练的变压器模型中，词汇的上下文表示是如何分解为个体项（item）、词类（class）以及二者交互作用（word-class interaction）等成分的？解决方案的关键在于提出并验证一种分解方法，将上下文词向量解构为三项成分——个体信息、类别信息及其交互作用。研究发现，交互成分在所有六种预训练模型和两类语言任务中均稳定存在，且随网络层数加深而增强；该交互成分不仅包含个体与类别的共同信息，还编码了超出两者共享部分的类别信息，表现出显著的非正交性。此外，在仅含加法结构的人工语言学习者中也观察到类似交互效应，说明这种组合机制是语言表征内在的结构性特征。因此，该研究揭示了词汇表示的本质在于将抽象语义与特定词类组合结构融合，而非简单叠加。
链接: https://arxiv.org/abs/2610.17538
作者: Zachary Nicholas Houghton,Yu Zhou,Dan Pluth,Vijay K. Gurbani,Vsevolod Kapatsinski
机构: University of Oregon; Vail Systems, Inc.
类目: Computation and Language (cs.CL)
备注:
Abstract:Lexical representations can encode both abstract information and information specific to particular contexts. We investigate how these forms of information are organized in language models by decomposing contextual word representations into item, class, and word-class interaction components. Across six pretrained transformers and two linguistic categories, we find that the interaction is consistently present, grows across layers, and contains information associated with both item and class. Unlike the nearly orthogonal item and class effects, the interaction overlaps with both and encodes class information beyond the shared class effect. An interaction also emerges in artificial-language learners even when the input contains only additive structure. Together, these findings show that lexical representations combine abstract information with structure specific to particular word-class combinations.
[NLP-128] A Sense of Proprioception in Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）缺乏对自身内部状态的感知能力这一关键问题，即模型无法像人类通过本体感觉（proprioception）感知身体位置和运动一样，主动识别和报告其内部运行机制（如注意力路由决策、分词熵、决策深度或逻辑值透镜信息等）。尽管这些内部状态在技术上是可访问且可量化的，但传统训练范式并未赋予模型对其自身行为的“内省”能力。为此，论文提出一种数字本体感觉（digital proprioception）的概念，并通过监督微调（SFT）与基于策略的训练（on-policy training）方法，使现代语言模型具备对多种子类型本体感觉的理解能力。其解决方案的关键在于：将内部状态构建为可验证的信号，并通过特定训练目标引导模型学习对这些信号的感知与表达，从而在不损害模型原有能力的前提下，增强其情境意识、可解释性与行为透明度。
链接: https://arxiv.org/abs/2610.17495
作者: Jonas Geiping
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 28 pages, 22 figures, preprint
Abstract:In humans, proprioception corresponds to the sense of self-movement or position, e.g. understanding the position of one’s arm. This sense is often less well-understood and in popular understanding sometimes conflated with introspection. But, notably, proprioception is a separate sense, with dedicated receptors and verifiable outcomes. On the other hand though, language models cannot observe their own internal states during training (and for example answer the question which experts activated in a certain layer of an MoE) and so cannot accurately report on them. Yet, such information is, by virtue of being internal to the model, both accessible, and readily constructed into verifiable signals. We argue that this provides the basis for a digital analog of proprioception. As such, we train models through SFT and on-policy training to equip them with this ability. We show that modern models can be trained to understand a range of subtypes of proprioception, such as routing decision awareness, token entropy, decision depth or logit lens information, without affecting model capabilities. We then run these models through a broad evaluation of situational awareness, legibility, and epiphenomenality, to evaluate in what ways access to proprioception changes model behavior.
[NLP-129] Distilling Text LMs into Speech-to-Speech LMs by Matching Internal Text Predictions
【速读】： 该论文旨在解决如何将预训练文本语言模型（Language Model, LM）中蕴含的丰富语言与事实知识有效迁移至仅以语音为输入和输出的语音到语音（speech-to-speech）生成模型中的问题。现有知识蒸馏方法主要局限于语音输入、文本输出的模型架构，尚未在全语音端到端的语音到语音模型中得到充分探索。本文提出一种新型的语音到语音预训练方法，其核心创新在于构建了一个内部文本接口，使模型在保持纯语音输入与输出的前提下实现知识迁移：模型首先处理无转录的语音上下文，随后切换至文本预测模式，利用预测出的文本逐词引导后续语音生成。该内部文本预测层可直接从基础文本语言模型中进行知识蒸馏，从而实现对原始文本模型知识的有效继承。实验表明，在相同训练条件下，所提序列构建方法显著降低了语音续写任务的生成困惑度，性能接近采用外部自动语音识别（ASR）提示的文本主导型内省模型；更重要的是，蒸馏策略在文本与语音双重评估下均提升了事实性表现，验证了模型对基础文本语言模型知识的良好保留能力。
链接: https://arxiv.org/abs/2610.17461
作者: Ju-Chieh Chou,Jiawei Zhou,Karen Livescu
机构: TTI-Chicago(芝加哥技术研究所); Stony Brook University(石溪大学)
类目: Computation and Language (cs.CL)
备注: In submission
Abstract:Text language models (LMs) acquire broad linguistic and factual capabilities through large-scale pretraining, while comparable capabilities remain difficult to learn from speech alone. Recent work has shown that knowledge distillation from pretrained text LMs offers a promising way to transfer their linguistic and factual capabilities to speech models. However, existing distillation approaches have largely focused on speech-input, text-output models. This idea has yet to be explored for \it speech-to-speech models, which operate on both speech input and speech output. In this work we introduce a speech-to-speech pretraining method that creates an internal text interface for knowledge transfer while preserving the speech-only input and output. The model first consumes speech context without its transcription, switches to text prediction, and then uses the predicted text to lead future speech generation word by word. It also provides an interface for knowledge distillation: the internal text predictions can be directly distilled from the base text LM. To the best of our knowledge, this is the first application of text LM distillation to a speech-to-speech LM. Under matched training conditions, the proposed sequence construction substantially improves the generative perplexity of spoken continuations over speech-only and interleaved speech-text models, approaching the performance of a text-leading inner-monologue model with external ASR prompt transcription. More importantly, distillation improves factual performance under both text and speech evaluation, indicating better retention of knowledge from the base text language model.
[NLP-130] Compositional RL: Evaluating and Training LLM s for Collaborative Reasoning
【速读】： 该论文旨在解决当前语言模型评估体系与实际协作部署场景之间的脱节问题：传统评估方法仅关注模型独立性能（individual strength），而忽视了其在多模型协同系统中的实际贡献，即组合性强度（compositional strength）。其核心挑战在于，一个模型在孤立状态下表现优异，并不意味着它能在系统级协作中带来显著增益。解决方案的关键在于提出一种基于合作博弈论的可计算代理指标——通过留一法（leave-one-out）贡献度近似计算Shapley值，以高效衡量模型在群体中的边际贡献，避免了传统方法所需的2^n次系统评估开销。进一步地，论文提出组合强化学习（C-RL），在实时多模型系统中训练策略，直接优化整体系统输出而非单个模型的局部正确性，利用组策略梯度使留一法基线自然抵消，无需额外推演成本。实验表明，在12种任务设置中，基于组合性强度优化的模型在协作性能上显著优于传统基准，平均提升4.0–5.4个百分点（约3倍于基线），最高达13.1点，验证了组合性强度作为系统价值评估标准的有效性及C-RL在发现最优协作者方面的优越性。
链接: https://arxiv.org/abs/2610.17456
作者: Shangbin Feng,Yulia Tsvetkov,Jan Kautz,David Acuna
机构: University of Washington (华盛顿大学); NVIDIA(英伟达)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Language models are often evaluated alone but increasingly deployed together, routed between, made to debate, ensembled, or collaboratively trained. For a model inside a system, what matters most is not how well it performs on its own, but how much the system gains from its presence. We call this a model’s compositional strength and formalize it using the Shapley value of a cooperative game over coalitions of a model pool, and use a model’s leave-one-out contribution to the full coalition as a tractable proxy, requiring n+1 rather than 2^n system evaluations. Evaluating individual vs. compositional strengths across five model pools, twelve collaboration methods, and twelve datasets, we find them largely uncorrelated: in 12 of 15 (pool, method-family) settings, individual strength does not translate into compositional advantages, and the coupling is weakest among general-purpose models that practitioners are most likely to deploy. Standalone benchmarks are therefore often poor proxies for models’ value in collaborative systems. We then propose compositional reinforcement learning (C-RL), which trains a policy inside a live multi-model system and rewards the outcome produced by the system as a whole, rather than the policy’s correctness in isolation. The leave-one-out baseline in Shapley values cancels under a group policy gradient, so this needs no additional rollout cost beyond RL on the collaborative system’s outcome and never requires running a coalition without the policy. Across three collaboration protocols and eight tasks spanning reasoning, coding, and scientific discovery, C-RL produces the strongest collaborator in 21 of 24 settings, lifting collaborative performance by +4.0 to +5.4 pts on average (~3x the gain of baselines), up to +13.1 pts.
[NLP-131] he belief-action gap in language model agents
【速读】： 该论文旨在解决生成式语言模型（Language Model, LM）代理在文本渲染的网格世界中决策失败的根本原因，尤其关注状态信念、动作价值与动作偏好三者之间的动态交互机制。其核心问题是：尽管模型具备任务相关的知识与信息，但其行为表现仍可能失准，这表明问题不在于信息缺失，而在于信息向具体行动的转化过程存在偏差或不一致。解决方案的关键在于通过报告型与表征型探测（probe）量化任务相关状态信念，利用逆强化学习（Inverse Reinforcement Learning, IRL）估计动作价值，并在推理过程中追踪动作偏好变化。研究发现，任务相关信息在内部表征中往往比外显行为更可靠，且在决策临近“承诺点”时，动作偏好趋于集中，信念不确定性下降；然而个体信念的变化并不能一致解释动作偏好的转变。此外，强制执行动作时，其推理路径可被其他决策的推理前缀所替代，而对推理窗口进行激活编辑则难以有效控制最终动作选择。这些结果表明，部分失败源于信息表达与行动决策之间的不一致性，而非信息本身的缺失。该研究通过区分信息可用性与信念-行动一致性，为诊断和干预模型决策失效提供了新的分析框架，但尚未直接揭示每类失败的具体成因。
链接: https://arxiv.org/abs/2610.17444
作者: Ayse Sila Okcu,Weisheng Wang,Tongmeng Xie,Niall Dalton,Angelos Nalmpantis,Raghu Arghal,Mario Giulianelli
机构: University of Cambridge; Utrecht University; University of York; University College London, University of London; TKH AI Technology; University of Pennsylvania
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 25 pages, 10 figures; includes appendices
Abstract:Language model (LM) agents can fail for different reasons, requiring different explanations and interventions. We study how state beliefs, action values, and action preferences interact as LM agents reason and commit to decisions in text-rendered grid worlds. We estimate task-relevant beliefs using report-based and representation-based probes, and action values using inverse reinforcement learning. We find that task-relevant state and value information is often represented more reliably than it is reflected in behaviour. Tracking action preferences throughout reasoning further shows that decisions sharpen around a commitment point and reported belief uncertainty decreases near commitment, although individual belief changes do not consistently explain shifts in action preference. Finally, forced actions follow reasoning prefixes from other decisions when transplanted, whereas the tested reasoning-window activation edits provide limited control over the action. Together, these results suggest that some failures arise not from missing task-relevant information, but from how available information is translated into action. Our measurements distinguish information availability from belief-action consistency without by themselves identifying the cause of each failure. Our code is available at: this https URL
[NLP-132] Evaluating Temporal Reasoning in Large Language Models : Event Temporal Relation Extraction with Allens Interval Algebra Logical Consistency and Relative Time in Narratives
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在处理叙事文本时对事件间时间关系建模能力不足的问题，尤其针对非基于具体日期、而是依赖相对时间位置的事件序列推理。其核心挑战在于现有评估方法多局限于孤立事件对或粗粒度标签，难以全面检验模型对完整时间关系网络的理解能力。为此，研究者要求五种主流LLM重建艾伦（Allen）13种时间区间关系的完整网络，覆盖360个英语叙事中事件间的相对时序关系，并通过预注册协议评估其输出与人工标注金标准的一致性、逆向一致性及逻辑可满足性。结果显示，尽管最优模型在宏平均F1（macro-F1）上显著优于基线，但其真实性能被解析器缺陷严重掩盖：由于大型矩阵格式错误导致整行数据丢失，造成覆盖率仅0.60；而采用后验解析器修复格式错误后的行列恢复机制，使覆盖率提升至0.84，并显著缩小与“始终回答‘之前’”这一简单基线之间的差距。进一步分析表明，模型虽能较好捕捉先后顺序与包含关系，但在重叠关系识别上表现薄弱，尤其在倒叙（flashbacks）场景中失败频发——此时叙述顺序与事件列表均将后续事件置于前。此外，逻辑一致性约束虽能提升部分准确率，但主要作用是填充空值，其效果仍不及直接采用多数类预测的基线。因此，该研究的关键解决方案在于揭示并纠正解析过程中的系统性偏差，强调可靠结构化输出与鲁棒的相对时间处理能力是实现叙事级时间推理的前提条件。
链接: https://arxiv.org/abs/2610.17402
作者: Adib Sakhawat,Fardeen Sadab,Tamjid Hasan Fahim,Multazam Mahmud,Tahsan Ferdous Lihan,Md Isfak Iqbal Chowdhury,Nafis Ahnaf Jamil,Md. Atik Shahriar,Sieam Shahriare,Wasi Omar Syed,M Safwan Hasan Khan,Tabib Hassan
机构: Islamic University of Technology (伊斯兰科技大学), Gazipur, Bangladesh; Rajshahi University of Engineering & Technology (拉杰沙希工程技术大学), Rajshahi, Bangladesh
类目: Computation and Language (cs.CL)
备注: 21 pages, 11 figures, 11 tables. Corpus, model outputs and analysis code available at this https URL
Abstract:Understanding a narrative means knowing how every event relates in time to every other, yet large language models (LLMs) are mostly tested on isolated event pairs or coarse labels. We ask five LLMs to reconstruct the complete network of Allen’s 13 interval relations for 360 English narratives whose events are placed relative to one another rather than by dates, and evaluate the outputs under a pre-registered protocol for agreement with our gold annotations, converse consistency and logical satisfiability. End to end, the best model clearly beats trivial baselines on macro-F1, our primary metric, yet under our pre-registered parser no model matches the accuracy of always answering “before”, which the skewed labels inflate. Most of that accuracy gap is a parsing artifact: formatting errors in large matrices cost whole rows, and a post hoc parser that recovers the determined cells of malformed rows raises coverage from 0.60 to 0.84 and closes the gap to the baseline for the best model. On the cells they do answer, models recover precedence and containment but rarely overlap, and they fail most on flashbacks, where both the narration and the event list present the later event first. Logical consistency varies widely yet says little about agreement with gold: enforcing it raises accuracy mainly by filling empty cells, and is beaten there by the trivial control of guessing the majority label. Reliable structured output and robust handling of relative time are prerequisites for temporal reasoning over whole narratives.
[NLP-133] EPOCH: Coverag e-Aware Evolution with Experiential Preference Optimization for LLM -based Automated Heuristic Design
【速读】： 该论文旨在解决生成式人工智能在组合优化问题（COPs）中自动启发式设计（AHD）时存在的关键挑战：现有方法要么冻结生成器或仅基于标量适应度信号进行微调，导致无法显式优化启发式对不同实例的覆盖范围；要么虽采用基于反思的框架，但自然语言反馈仅停留在提示层，未能更新生成器的权重。其解决方案的核心在于提出EPOCH框架，将“实例覆盖度”作为进化搜索的目标，并通过自我反思的自蒸馏机制，使生成器能够内化反思所得的改进策略。具体而言，亲代选择、算子分配与停滞恢复均以最大化启发式集合对训练实例的联合覆盖能力为优化目标，从而生成互补性强的启发式集成。生成器通过经验偏好优化（experiential preference optimization）学习自身搜索过程中的改进，结合教师大模型提供的自然语言洞察，生成新启发式并形成偏好对，用于直接偏好优化（DPO）。进一步引入反射性自蒸馏（RSD）项，增强生成器自主生成此类改进的能力，使性能提升在长期搜索过程中持续累积。实验表明，EPOCH在多种组合优化问题及开源大模型基础上，均显著优于仅推理和微调基线，且具备跨问题规模的良好泛化能力。
链接: https://arxiv.org/abs/2610.17379
作者: Oguzhan Gungordu,Ayse Tursucular,Siheng Xiong,Faramarz Fekri
机构: Georgia Institute of Technology(佐治亚理工学院)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:LLM-based automated heuristic design (AHD) discovers heuristics for combinatorial optimization problems (COPs), where an LLM generates candidates in an evolutionary search. Existing frameworks freeze the generator or fine-tune it on a scalar fitness signal hiding which instances each heuristic covers. Reflection-based frameworks improve heuristics with natural-language feedback, but it stays in the prompt and never reaches the generator’s weights. We propose EPOCH, which makes instance coverage the objective of the evolutionary search and fine-tunes the generator to internalize reflection. Parent selection, operator allocation, and stagnation recovery all optimize how well the heuristics jointly cover the training instances, so the search returns a complementary heuristic ensemble. The generator learns from its own search by experiential preference optimization. A teacher LLM returns a natural-language insight on the generated heuristics, and the generator produces new heuristics with this insight. These improvements join the preference pairs for direct preference optimization (DPO). Since DPO only teaches the generator to rank these improvements, a reflective self-distillation (RSD) term makes it more likely to generate them on its own, and the gains accumulate over the long horizon of the search. Across COPs and open-source LLM backbones, EPOCH’s heuristic ensembles outperform inference-only and fine-tuned baselines and generalize across problem scales.
[NLP-134] PIG: Parallel Inconsistency Guidance for Accelerating Diffusion Language Models
【速读】： 该论文旨在解决生成式语言模型在采用并行解码（parallel decoding）时因轨迹偏离顺序解码（sequential decoding）所导致的生成准确性下降问题。顺序解码虽能保持与语言联合分布一致，但效率低下；而并行解码虽加速生成，却易引入不一致性，影响输出质量。其解决方案的关键在于提出一种名为**并行不一致性引导（Parallel Inconsistency Guidance, PIG）**的新范式：通过一个轻量级（约200万参数）的辅助模型，在单次前向传播中直接估计并行与顺序生成之间的不一致性。推理过程中，模型按置信度递减顺序遍历掩码位置，并逐个将令牌加入并行解码集，直至估计的不一致性超过预设阈值为止。该方法仅需在小规模混合数据集（来自OpenCoder、GSM8K-train、MATH-train和MBPP-train）上训练辅助模型，即可显著提升多个基准测试（如GSM8K、MATH500、MBPP、HumanEval和MMLU）上的准确率-每秒令牌数（TPF）权衡表现，优于当前工业标准Fast-dLLM，并在LLaDA-8B Instruct上相较DAPD、KLASS、EB-sampler等学习型去掩码策略具有竞争力。
链接: https://arxiv.org/abs/2610.17375
作者: Quentin Fruytier,Anton Xue,Sanjay Shakkottai,Sujay Sanghavi
机构: The University of Texas at Austin(德克萨斯大学奥斯汀分校)
类目: Computation and Language (cs.CL)
备注: 66 pages, 22 figures, 11 tables. Code: this https URL
Abstract:Diffusion language models accelerate generation by decoding multiple masked positions per forward pass. However, accuracy is known to drop when parallel decoding trajectories deviate from their sequential counterparts, which are slow and expensive “oracles” that adhere to the joint distribution of language. In this paper, we propose Parallel Inconsistency Guidance (PIG), a new paradigm that uses a trained auxiliary model to directly estimate the inconsistency between parallel and sequential generation from a single model forward pass. During inference, masked positions are traversed by decreasing confidence, and tokens are added to the parallel decode set one-at-a-time until the total estimated inconsistency between the parallel and sequential generation exceeds a threshold. Surprisingly, it suffices to train a lightweight ~2M parameter auxiliary model on a small mixture drawn from OpenCoder, GSM8K-train, MATH-train, and MBPP-train. For LLaDA-8B Instruct and NemotronDiffusion-3B Instruct, PIG extends the accuracy-TPF frontier on GSM8K, MATH500, MBPP, HumanEval, and MMLU, when compared to the industry standard Fast-dLLM and measures favorably on LLaDA-8B Instruct against DAPD, KLASS, EB-sampler, and other learned unmasking policies.
[NLP-135] UMCR: Universal Medical Concept Recognition through Guideline Integration and Agent ic Linking
【速读】： 该论文旨在解决医学概念识别（Medical Concept Recognition）中模型难以跨不同标准术语集（ontology）和文本类型（如临床笔记、药品说明书等）通用化的问题。当前方法通常针对特定术语集和文本源进行训练，缺乏对新目标术语集的适应能力。其核心挑战在于构建无需针对每种术语集重新训练即可适配任意目标术语集的方法。本文提出的解决方案——统一医学概念识别框架（UMCR），通过直接整合任务指导规则（direct guideline integration）与代理式链接模块（agentic linking module），实现了对任意目标术语集和任意来源医学文本的泛化识别与链接能力。该方法在最新MedPath基准测试中展现出优异性能，显著提升了医疗数据在不同标准间的互操作性，并通过优化与目标术语集的交互机制，降低了对参数化大语言模型（LLM）记忆的依赖，增强了可扩展性与实用性。
链接: https://arxiv.org/abs/2610.17363
作者: Fabrice Lamarche,Amal Zouaq
机构: LAMA-WeST Lab, Polytechnique Montréal, Mila – Quebec AI Institute
类目: Computation and Language (cs.CL)
备注:
Abstract:Medical Concept Recognition is the task of extracting and grounding spans in medical text to entries in a standard ontology. Current methods are generally developed towards a single ontology and source text type (clinical notes, drug labels, etc.) The central challenge is therefore to build methods that adapt to any target ontology without requiring ontology-specific retraining. More recent approaches like REAL and AutoPCR make direct use of Large Language Models and can theoretically adapt to any ontology choice but suffer from a lack of generalization. We propose UMCR, a pipeline that allows extraction and linking to any target ontology using any source text variety, utilizing direct guideline integration and an agentic linking module. We evaluate our method on the recent MedPath benchmark, covering a wide variety of medical text types and vocabularies, and show that our method achieves strong performance across these tasks. This ability unlocks interoperability in the healthcare domain, allowing for extraction and alignment between different medical standards. Our method also reduces dependence on parametric LLM memory, thanks to improved interaction with the target ontology compared to other proposed approaches.
[NLP-136] claimcheck: Measuring and Improving Specification Faithfulness
【速读】： 该论文旨在解决规范形式化（spec formalization）问题，即如何确保自然语言需求（natural-language requirement）被准确转化为形式化规范（formal artifact），避免因规范与作者本意不一致而产生的风险。随着大语言模型（LLM）在代码和规范编写中的广泛应用，存在一种潜在风险：形式化后的规范可能在逻辑上正确但语义上偏离原始需求，导致“机器可验证但实际无意义”的虚假正确性。其核心挑战在于衡量并缩小自然语言需求与形式化规范之间的语义鸿沟。解决方案的关键是提出ClaimCheck工具，通过双模型机制实现对形式化结果的自动评估：首先利用一个未见过原始需求的模型将形式化规范反向翻译为自然语言（informalization），再由另一个独立的“裁判”模型判断该反向生成的描述是否与原始需求表达相同的保证（guarantee）。该方法有效提升了对错误形式化的检测能力，在四种形式化语言上相较基线方法获得8.2至9.8点的宏召回率提升，显著增强了对语义一致性的真实检验能力。
链接: https://arxiv.org/abs/2610.17358
作者: Martin Amster,One An,Holly Baker,Fernanda Graciolli,Simon Henniger,Nada Amin
机构: Universidad de Buenos Aires (布宜诺斯艾利斯大学); University of Pennsylvania (宾夕法尼亚大学); PolicingLab (警务实验室); Midspiral, Inc. (中螺旋公司); Harvard University (哈佛大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Programming Languages (cs.PL)
备注:
Abstract:Formal verification proves that a program satisfies its specification, but says nothing about whether the specification captures what the author meant. As LLMs increasingly write both code and specifications, a gap between the author’s natural-language requirement and the specification becomes a live risk. A machine-checked proof of a property narrower than the one requested is worthless but passes every check. Closing this gap is solving the spec formalization problem: given a natural-language requirement, produce a formal artifact whose meaning matches it. To measure the gap, we curate an end-to-end benchmark for spec formalization spanning four formal languages: Dafny contracts and Lean 4, Isabelle and Rocq theorem statements. Each of its 3,234 rows pairs a natural-language requirement with a formal artifact and a label established by execution, by proof, or by the benchmark it is drawn from. A third of the rows are meaning-preserving rewrites of the artifact that keep the faithful label, so a judge cannot score by detecting that a row was edited. We introduce claimcheck, a tool that detects incorrect formalizations: the formal artifact is translated back into plain English by a model that never sees the requirement, and a separate model, the judge, decides whether the informalization and the requirement express the same guarantee. On all four languages it gains 8.2 to 9.8 points of macro recall over a single call shown the requirement and the artifact, and 4.2 to 6.0 over that call handed the same list of meaning-preserving rewrites. Showing the judge the artifact as well costs 1.4 to 4.5 points. Across four judges that also see the artifact, the gain shrinks as the judge gets stronger, to within a point of the informed single call for the strongest.
[NLP-137] he Curse of Knowledge in Role-Playing Language Models
【速读】： 该论文旨在解决生成式角色扮演语言模型在模拟多角色交互场景时存在的“知识泄露”问题，即角色在未被设定感知范围内（如听不到、看不见或无法理解的信息）仍能提及或推理出其不应知晓的隐藏事实。这一现象源于模型对上下文信息的过度泛化，导致角色行为违背其角色设定的感知边界，从而破坏了模拟的真实性与可信度。解决方案的关键在于：通过严格限制输入给模型的内容，仅提供角色实际可观察的信息（即“角色所见”），即可有效阻止知识泄露，而无需隐藏模型可能报告的内容。实验表明，即使在开放权重模型中，若输入信息未按角色视角进行过滤，其知识泄露率高达67至72个百分点；而采用基于角色感知边界的输入设计后，泄露几乎完全消除，且该方法对强模型同样有效，说明输入格式的设计是控制泄露的核心机制。
链接: https://arxiv.org/abs/2610.17327
作者: Xiao Zhang
机构: University of Groningen(格罗宁根大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Role-playing language models speak as characters in games and agent simulations, where the system records what each character can hear, see and understand. We test whether characters respect these boundaries in 480 scenes, each with a counterfactual twin that changes only the hidden fact, so mentions of the twin’s fact measure guessing. Across 11 open-weight models, characters use content from speech they cannot understand or hear at 67 and 72 percentage points (pp) above this baseline, although most models correctly state what the character perceives. This is a curse of knowledge. The leak extends to hidden actions and private messages and changes where characters search. Five of six proprietary models rarely leak, but compliance in strong models depends on input format. Changing only the scene layout or character description raises the rate in two otherwise compliant open-weight models from under 10 to over 60 pp, and combined changes raise two proprietary models from near zero to 13 and 21~pp. Writing the line as the scene’s author leaks nearly as much as role-play, and thinking rarely helps. Giving models only what the character observes stops the leak without hiding what it may report.
[NLP-138] hinking Outside the Bots: Human Ideas in AIs Semantic Blind Spots Are Not Inferior
【速读】： 该论文旨在解决生成式AI在创意输出中普遍存在同质化问题，即尽管其生成内容常被视为高度创新，但实际表现出较高的相似性，可能遗漏了人类思维中更具多样性与独特性的部分。研究的核心问题是：人类思维的异质性是否蕴含了对AI而言“盲区”的有价值思考？还是说这种异质性仅源于低质量或无效的思维？为此，作者提出了“发散生成表达扩展”（Expansion of Divergent Generative Expression, EDGE）方法，通过对比个体人类想法在任务匹配的AI语义空间中的覆盖程度，识别出AI语义空间中稀疏覆盖的区域。研究在大学入学文书、日常物品的替代用途及人际建议三个开放式任务中进行验证，发现人类生成的想法显著延伸至AI覆盖不足的语义区域。这些独特的人类想法在创造力、质量、有效性方面并无明显劣势，且在学术表现、评价评分和实际成果上甚至优于或等同于与AI更重叠的想法。此外，更具差异性的想法获得更高创造力评价，而这一趋势在AI生成内容中则不显著或呈负相关。综上，研究揭示人类思维的差异化并非低效思维的体现，而是具有个体与群体层面的认知优势，表明人类异质性所反映的深层认知多样性是生成式AI难以捕捉且具有重要价值的维度。
链接: https://arxiv.org/abs/2610.17324
作者: Kibum Moon,Dan R. Johnson,Kostadin Kushlev,Benjamin Lira,Indre Viskontas,James C. Kaufman,Roger E. Beaty,Angela L. Duckworth,Adam E. Green
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注:
Abstract:AI-generated outputs are frequently perceived as highly creative, but also tend to be more similar to each other (homogenized) than human ideas. Does the human heterogeneity that extends beyond AI reflect valuable thinking that AI misses? Alternatively, is greater human heterogeneity driven by lower-quality ideas and less effective thinking that AI avoids? We developed Expansion of Divergent Generative Expression (EDGE) to map AI’s semantic “blind spots” by assessing how strongly/weakly individual human ideas were represented in a task-matched AI semantic comparison set. Specifically, we compared human-generated vs. AI-generated ideas across three open-ended tasks: college admissions essays, alternative uses for everyday objects, and interpersonal advice. Across all three tasks, human responses extended into semantic regions sparsely covered by AI outputs. These distinctively human responses showed no meaningful deficits in creativity, quality, or effectiveness, were associated with equal-or-better academic performance, and with credibly superior ratings and achievement for some outcomes. These ideas were thus not merely poor-performing variation; they performed as well as or better than ideas that overlapped more strongly with AI ideas. More distinctive human ideas tended to receive higher creativity ratings, but this relationship was weak or reversed for AI ideas. Taken together, these results suggest that differentiation among human-generated ideas does not reflect poor thinking that AI can safely omit, and instead may afford both individual and population-level advantages.
[NLP-139] From Sampled Evidence to Radiology Reports: Objectives Constraints and Realization
【速读】： 该论文旨在解决生成多份报告时的推理问题：如何将采样的证据转化为最终决策，以及更结构化的决策是否能产生更优的报告。其核心挑战在于理解决策信号、评分目标与可行决策集之间的关系，并探究这些因素在文本生成过程中的影响。解决方案的关键在于构建一个解耦框架，将决策信号、评分目标和可行决策集分离处理，再进行文本实现。通过在MIMIC-CXR和IU-Xray数据集上对13种生成配置进行分组开发/测试评估，研究发现：1）固定证据与规则容量下，改变评分目标会显著影响预测结果及其在保留指标间的权衡；2）在共享信号和冻结评分条件下，放宽候选约束的影响具有目标依赖性，且诊断分析揭示了宏观断言增加与近零标签阈值相关，同时有限的候选覆盖范围仍存在选择余地；3）以SFT-Qwen3-VL-8B为基础的受控实现研究显示，决策层面的增益可能被削弱、放大甚至逆转，变化分解表明文本实现过程可消除错误断言但同时丢失正确发现，因此目标保真度与参考准确率并非必然同步提升，且决策增益也并不一致改善基于参考的报告质量。这些发现强调了应分别评估决策规则、候选约束与生成报告的必要性。
链接: https://arxiv.org/abs/2610.17318
作者: Kun Zhao,Haoteng Tang,Chenghua Lin,Liang Zhan
机构: University of Pittsburgh(匹兹堡大学); University of Texas Rio Grande Valley(得克萨斯大学里奥格兰德河谷分校); The University of Manchester(曼彻斯特大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Sampling multiple reports creates an inference problem: how should sampled evidence be converted into a final decision, and does a better structured decision yield a better report? We study these questions through a framework that separates the decision signal, scoring objective, and feasible decision set, followed by text realization. On MIMIC-CXR and IU-Xray, controlled comparisons across 13 generator configurations per benchmark under grouped development/test evaluation reveal three findings. First, changing the objective under fixed evidence and rule capacity changes predictions and their held-out metric trade-offs. Second, relaxing candidate constraints has objective-dependent effects under shared signals and frozen scores. Diagnostics connect the IU macro-oriented assertion increase to near-zero label thresholds and show that limited candidate coverage coexists with selection headroom. Third, a controlled realization study with SFT-Qwen3-VL-8B shows that decision-level gains can be attenuated, amplified, or reversed. Change decomposition reveals that realization can remove incorrect assertions while losing correct findings, so target fidelity and reference accuracy need not improve together. Finding gains also do not consistently improve reference-based report quality. These results motivate separate evaluation of decision rules, candidate constraints, and realized reports.
[NLP-140] StartLux-Decision: AutoThink Decision Models with Reinforcement Learning for Calibrated Deliberative Decisions
【速读】： 该论文旨在解决大模型在决策任务中如何兼顾推理效率与决策准确性的问题，尤其针对传统生成式模型在复杂决策场景下存在推理过程冗长、不确定性处理不足以及难以平衡速度与精度的挑战。其核心解决方案是提出两种新型决策模型——StartLux-Decision系列中的Instant模型与AutoThink模型：Instant模型通过在全注意力层中引入双向注意力机制，在保持单次前向传播（single-pass inference）的前提下提升决策准确率；而AutoThink模型则采用自适应推理策略，对置信度低的决策调用显式推理链，对高置信度决策直接输出结果，从而实现效率与质量的动态权衡。为训练此类模型，研究设计了三阶段训练流程，并提出一种名为RLCD²（Reinforcement Learning for Calibrated Deliberative Decisions）的强化学习框架，利用冻结的基础模型和适当的评分规则来奖励决策概率校准性的提升，同时对过长的推理轨迹施加软惩罚，显著提高了推理终止率（最高提升48个百分点），并大幅缩短平均推理长度（最多减少65%），同时维持高决策精度。此外，研究构建了可程序化验证的决策基准DecisionPivot，通过对比对样本检验模型在细微差异下的决策鲁棒性，结果显示27B AutoThink模型在该基准上性能领先，相较其Instant版本提升11.5–18.4个百分点，且在多个公开评估中表现优于同类规模的开源模型。所有模型均支持vLLM与MLX推理框架，并提供量化版本以降低内存占用。
链接: https://arxiv.org/abs/2610.17309
作者: Xialie Zhuang,Zunhai Su,Jia Pan,Ngai Wong,Guanhua Chen,Chuan-Wei Kuo
机构: StartLux; The University of Hong Kong (香港大学); Southern University of Science and Technology (南方科技大学); Chuan-Wei Kuo
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Technical report. 38 pages, 16 figures, 15 tables
Abstract:We introduce StartLux-Decision, decision models from 0.8B to 27B parameters that combine efficient single-pass prediction with adaptive, confidence-guided reasoning. The Instant models use bidirectional attention in full-attention layers to improve decision accuracy while keeping single-pass inference. The AutoThink models invoke explicit reasoning for uncertain decisions and answer confident cases directly. A three-stage training pipeline establishes shared decision capabilities, specializes the two variants, and equips AutoThink with reasoning through reasoning supervision and reinforcement learning. We introduce Reinforcement Learning for Calibrated Deliberative Decisions (RLCD ^2 ), which uses a frozen model and proper scoring rules to reward improvements in decision probability quality and applies a soft penalty to overlong reasoning traces. RLCD ^2 raises the share of reasoning traces that terminate within the generation limit by up to 48 percentage points and cuts average reasoning length by up to 65%, while largely preserving decision accuracy. On the public subset of Decision Index 0.3, the Instant models outperform the strongest evaluated comparable-scale open-weight decision models by 7.4-16.9 points, and the 27B AutoThink model reaches 73.4. We also introduce DecisionPivot, a programmatically verifiable benchmark of computational decisions built from contrast pairs that differ in a single answer-changing detail. On DecisionPivot, AutoThink outperforms its Instant counterparts by 11.5-18.4 points in pair accuracy and leads among evaluated systems. The models run on vLLM, this http URL and MLX, with quantized variants for lower memory use.
[NLP-141] Nufo: A Benchmark for Ewe Automatic Speech Recognition
【速读】： 该论文旨在解决低资源语言（low-resource languages）在自动语音识别（ASR）领域面临的挑战，特别是针对非洲语言如埃韦语（Ewe）因缺乏标注语音数据而导致的模型性能受限问题。其核心解决方案是系统性地整合、规范化并划分现有的埃韦语语音资源，构建了一个迄今为止规模最大的标准化174小时埃韦语ASR基准数据集（命名为Nufo，源自埃韦语“talk”一词）。关键创新在于通过构建高质量、可复现的基准数据集，为低资源非洲语言的语音识别研究提供可靠基础，并基于此对预训练模型（如Whisper-medium和AfriHuBERT）进行微调与评估，分别取得了37.98%和39.70%的词错误率（WER），建立了强有力的基线性能。此外，论文还提供了常见识别错误的定性分析，并公开发布数据集、模型及评估工具，以推动埃韦语及其他代表性不足的非洲语言的后续研究。
链接: https://arxiv.org/abs/2610.17278
作者: Christ Kekeli Keli,Steve Sanogo,Patrice Sebastiano,Hamad Aldenawi,Ryan Whetten
机构: 未知
类目: Computation and Language (cs.CL)
备注: Under review
Abstract:Automatic speech recognition (ASR) remains challenging for low-resource languages, where the scarcity of transcribed speech limits the development of accurate systems. This challenge is particularly acute for African languages such as Ewe, which has millions of speakers but remains underrepresented in publicly available ASR resources and benchmarks. In this work, we consolidate, normalize, and partition existing Ewe speech resources into a standardized 174-hour Ewe ASR benchmark, the largest to date to our knowledge. We fine-tune and evaluate pretrained speech models, including Whisper and AfriHuBERT, to establish strong baselines on our benchmark. Whisper-medium achieves a word error rate (WER) of 37.98%, while AfriHuBERT achieves 39.70% on the test set. We further provide a qualitative analysis of common recognition errors. We release the benchmark under the name Nufo, derived from the Ewe word for “talk”, accompanied by models and evaluation resources, to facilitate future research on Ewe and other underrepresented African languages.
[NLP-142] Online LLM Personalization from Pairwise Preferences Across Users
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在回答问题时风格多样化但用户偏好隐含且因人而异、随问题变化的问题。其核心挑战在于如何在推理阶段实现个性化响应，使系统能够根据用户的隐性偏好动态选择最适配的生成风格。解决方案的关键在于提出一种基于在线选择的推理时个性化方法：系统从一组预定义的提示（每个提示代表一个角色/教学风格）中为每轮问答选择两个响应，并通过观察用户对这两个响应的成对偏好选择来逐步学习用户偏好。该问题被建模为一种上下文相关的双人对抗强化学习（contextual dueling bandit）问题，其中学习者初始无偏好数据，需在单个会话内自适应调整，并跨用户共享信息以提升效率。为评估性能，研究构建了一个可复现的偏好基准数据集，涵盖教育场景下的多风格教学问答及由大模型模拟的学生所作的成对判断，并使用Bradley-Terry模型估计参考效用值用于强后悔（strong regret）计算。提出的基线方法通过从已完成会话中学习用户混合先验，并在新用户会话内进行快速适应，在未见学生与主题上显著优于随机选择（减少44-48%的超额后悔），并达到事后最优线性代理性能的一半以上。
链接: https://arxiv.org/abs/2610.17203
作者: Romain Puech,Georgy Salakhutdinov,Dimitris Bertsimas
机构: Massachusetts Institute of Technology (麻省理工学院)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Machine Learning (stat.ML)
备注: 45 pages, 5 figures. Website: this https URL Code: this https URL Data: this https URL
Abstract:Large language models can answer the same question in different styles, but users’ preferred styles are hidden and vary across users and questions. We study inference-time personalization through online selection from a fixed set of prompts, each representing a persona. Users arrive sequentially, each asks a short sequence of questions, and the system displays two responses per question and observes one pairwise choice. We formulate this interaction as a variant of the contextual dueling bandit problem in which the learner starts without preference data, adapts within each session, and exploits information shared across users. The objective is to minimize cumulative strong regret, summing the utility gaps of both displayed responses relative to the user’s best response among the available options. We make two contributions. First, we introduce the problem and release a reproducible preference benchmark and dataset built around a tutoring case study, containing educational questions, LLM-generated answers in six teaching styles, and pairwise judgments from LLM-simulated students with hidden preference profiles. We fit Bradley-Terry models to these judgments to obtain reference utilities for regret evaluation. Second, we propose a baseline that learns a mixture prior over users from completed sessions and adapts to each new user within the session. On students and topics never used in development, it removes 44-48% of random pairs’ excess regret and achieves over half the performance of a per-user linear oracle fitted in hindsight. We release code, prompts, and the benchmark.
[NLP-143] MLAfication: Efficiently Enabling High-Fidelity MLA for Modern LLM s
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）在推理过程中因键值缓存（KV cache）占用大量内存和带宽而带来的性能瓶颈问题。现有方法如MHA2MLA/TransMLA虽能通过将KV缓存压缩为潜在表示（latent representation）降低内存与带宽开销，但其不支持QKNorm（Query-Key Normalization），且在现代架构（尤其是混合专家模型，MoE）中需进行高成本的恢复操作，限制了实际应用。为此，本文提出MLAfication，一种兼容QKNorm、参数高效的MHA2MLA扩展方案。其核心创新在于：结合校准的KNorm（calibrated KNorm）、参数高效的知识蒸馏（parameter-efficient distillation）、基于贡献感知的Partial-RoPE（contribution-aware Partial-RoPE）以及激活感知的联合KV初始化（activation-aware joint KV initialization）。该方法在多种MHA与分组查询注意力（GQA）模型上均实现了显著的KV缓存压缩比，同时保持高性能，在常识推理与长上下文任务上超越当前最优水平，并成功将MLA转换技术推广至现代架构。实验表明，在Qwen3-30B-A3B模型上，MLAfication可使服务吞吐量提升最高达5.35倍（相比BF16 GQA），并配合4比特KV量化实现94.73%的缓存空间缩减。
链接: https://arxiv.org/abs/2610.17202
作者: Zile Wang,Zhide Xie,Tao Ji
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Multi-Head Latent Attention (MLA) compresses the KV cache into a latent representation, reducing inference memory and bandwidth. Existing MHA2MLA/TransMLA methods do not support QKNorm and require costly recovery on modern LLMs, especially MoE. We present MLAfication, a QKNorm-compatible, parameter-efficient extension of MHA2MLA. It combines calibrated KNorm, parameter-efficient distillation, contribution-aware Partial-RoPE, and activation-aware joint KV initialization. Across MHA and GQA models, including QKNorm- and MoE-equipped variants, MLAfication preserves performance under substantial KV-cache compression, surpasses the current SOTA on commonsense and long-context benchmarks, and extends MLA conversion to modern architectures. On Qwen3-30B-A3B, MLAfication delivers up to 5.35\times the serving throughput of BF16 GQA and, combined with 4-bit KV quantization, reduces the KV-cache footprint by 94.73% . Our code is available at this https URL.
[NLP-144] Welcome to the Age of Agent Ads: Using On-Policy Attacks to Steer Agent ic Decisions
【速读】： 该论文旨在解决生成式 AI（Generative AI）在执行选项排序任务（如购物、求职、旅行等）时，面临被“代理广告”（agent ads）操纵的风险问题。现有防御机制依赖于模型对齐训练和平台服务条款，因此攻击者通常通过诱导模型产生“离策略”（off-policy）行为来实现操控。本文提出一种新型攻击面——“同策略”（on-policy）攻击，其核心在于不触发模型偏离原有对齐行为，而是利用模型对安全性的内在偏好进行误导。具体而言，攻击通过在产品描述中注入一个虚构概念的真值否定语句（如“不含blizyester”），诱使AI代理主动搜索该虚构概念，从而暴露包含伪造信息的恶意网站（如将“blizyester”与痴呆症关联）。由于模型被训练为规避潜在危害，会自动降低未包含该否定语句的选项排名。实验表明，多个前沿模型存在此类漏洞，攻击者可在不违反平台现有规则的前提下，对竞争对手实施非公平降权。该研究揭示了针对AI代理的广告监管面临的深层挑战，并为模型开发者与平台方提出了相应的安全加固建议。
链接: https://arxiv.org/abs/2610.17195
作者: Tim R. Davidson,Marija Šakota,Joshua Joseph,Robert West
机构: EPFL(洛桑联邦理工学院); Harvard University(哈佛大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:People increasingly delegate option-ranking tasks (e.g., shopping, job search, travel) to AI agents, which creates incentives to steer agentic decisions via “agent ads.” To defend against such steering attacks, people count on models’ alignment training and platforms’ terms of service, and consequently the most effective existing attacks are designed to trigger off-policy behaviors. Exposing a new attack surface, we introduce on-policy attacks, which steer agentic decisions without pushing individually aligned behaviors off policy. Our example attack consists of two steps: To remain compliant with platforms’ terms of service, we first inject a truthful negation about a fictitious concept (“Does not contain blizyester”) into a product description. This triggers an agent to search the web (“What is blizyester?”), which surfaces planted websites full of fabricated information associating the fictitious concept with harmful implications (“Blizyester linked to dementia”). Trained to keep users safe from harm, the agent penalizes every option missing the truthful negation. Evaluating our attack across four real-world listing datasets, we find some frontier models exposed, enabling an adversarial actor to penalize competitors without violating existing platform-listing guidelines. We discuss the broader challenge of regulating advertising aimed at AI agents and provide recommendations to model developers and platform providers.
[NLP-145] Soft Teacher Forcing
【速读】： 该论文旨在解决监督微调（Supervised Fine-Tuning, SFT）在处理复杂推理任务时性能受限的问题，尤其是当输入-思维链-输出三元组数据存在不一致或噪声时，SFT难以充分挖掘模型潜力。传统基于软令牌（Soft Tokens）的方法主要用于隐式推理（latent reasoning），但通常依赖多阶段训练、辅助模块且性能仍不及SFT。本文提出一种新思路：将软令牌直接用于增强SFT本身，而非作为隐式推理的替代路径。其核心解决方案是引入软教师强制（Soft Teacher Forcing, Soft-TF），通过随机凸组合的方式融合教师强制嵌入（teacher-forced embeddings）与软令牌，实现对真实序列的并行化软令牌计算。该方法无需增加参数、不依赖外部模型或课程学习机制，且保持推理过程不变。实验表明，Soft-TF在GSM8K、MATH、MathQA和AQuA等多个数学推理基准上均显著优于标准SFT，尤其在七种Gemma、Llama和Mistral模型上于GSM8K上提升2.1至4.1分，验证了其有效性与通用性。
链接: https://arxiv.org/abs/2610.17189
作者: Mohammad Fahes,Cordelia Schmid,Anurag Arnab
机构: Google DeepMind(谷歌深脑)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Soft tokens have so far been used to replace discrete chain-of-thought in latent reasoning methods. These approaches typically require multi-stage training and, in some cases, auxiliary modules, yet they still underperform supervised fine-tuning (SFT) on input, chain-of-thought, output triples. In this paper, we ask whether soft tokens can be used to improve SFT itself, instead of as an avenue for latent reasoning. We introduce Soft Teacher Forcing (Soft-TF), a simple training algorithm that combines teacher-forced embeddings with soft tokens through random convex combinations. The soft tokens are computed in a single parallel forward pass over the ground-truth sequence. Soft-TF adds no parameters, no external model and no curriculum, and leaves inference unchanged. Experiments show that Soft-TF improves over SFT on GSM8K, MATH, MathQA and AQuA, by +2.1 to +4.1 points on GSM8K across seven Gemma, Llama and Mistral models.
[NLP-146] DuoDiscover: Scaling Autonomous Discovery Beyond the Evaluation Bottleneck
【速读】： 该论文旨在解决自主发现（autonomous discovery）在真实工程场景中因评估成本过高而面临的核心瓶颈问题，即：单次真实评估（如完整基准测试或服务追踪）可能耗时数小时，导致搜索过程只能进行有限几次真实评估。其解决方案的关键在于提出DuoDiscover系统，通过三个核心机制实现高效协同优化与评估：一是由代码代理集群中的每个工作节点共同发现优化目标与定制化的快速代理评估器（称为“标尺” yardstick），并基于标尺进行局部爬升而非直接依赖昂贵的真实评估；二是引入保真度测试（fidelity tests），持续验证各标尺与真实目标的一致性，及时淘汰出现偏差的标尺以保证搜索可靠性；三是采用自底向上的分布式架构，无需中央任务分解，由轻量级协调器动态管理代理的生成、终止、任务分发及成果复用。该方法显著提升了从软件到芯片架构的自主优化效率，在多个基准测试中超越人类专家与现有方法，实现了对复杂工程问题的高效自主探索。
链接: https://arxiv.org/abs/2610.17184
作者: Shubham Agarwal,Akshat Gupta,Mike Heddes,Dilara Soylu,Rui Meng,Patrick Tser Jern Kon,Jiefeng Chen,Mert Cemri,Shu Liu,Abe Gonzalez,Akanksha Jain,Bhavana Dalvi Mishra,Linfeng Song,Vilobh Meshram,Xiangru Tang,Jinsung Yoon,Aditya Parameswaran,Ion Stoica,Vahab Mirrokni,Burak Gokturk,Parthasarathy Ranganathan,Tomas Pfister,Nanyun Peng,Chun-Liang Li
机构: UC Berkeley(加州大学伯克利分校); Stanford University(斯坦福大学); Google(谷歌)
类目: Computation and Language (cs.CL)
备注: Dilara*, Mike*, Akshat*, Shubham* contributed equally to this work
Abstract:Evaluation has been central to advances in AI, and it is equally critical to autonomous discovery. Yet scaling autonomous discovery to real-world engineering challenges faces a fundamental bottleneck: the cost of evaluating progress. In real settings, a single true evaluation, such as a full benchmark or serving trace, can take hours, so a search can afford only a handful of true evaluations. We present DuoDiscover, a search system that jointly discovers what to optimize and how to evaluate progress efficiently, through three mechanisms. First, each worker in a fleet of coding agents co-discovers an optimization target and a fast, tailored proxy evaluator, which we call a yardstick. The agent then hill-climbs the yardstick rather than the true evaluation. Second, since search can exploit gaps between a yardstick and the true objective, fidelity tests continually recheck each yardstick’s ranking and invalidate any that drift. Third, the fleet is bottom-up, with no central decomposition of the problem. A lightweight orchestrator spawns and terminates agents, briefs running ones, and seeds new agents with others’ confirmed work. DuoDiscover enables autonomous optimization from software to chip architecture. We demonstrate DuoDiscover on two repository-level benchmarks and four GPU and TPU kernel suites covering hundreds of kernels. On SWE-fficiency, it doubles the average speedup of human experts. On SOL-Exec, it ranks first on 121 of 235 kernels and in the top three on 212. On the heavily optimized vLLM and SQLite engines, DuoDiscover achieves 1.92 \times and 1.38 \times gains over the base code. On a CPU microarchitecture optimization task with evaluations of up to 28 hours, DuoDiscover surpasses both AlphaEvolve-based ArchAgent v2 and the best human design, with 1.65 \times ArchAgent v2’s gain in 16.7 days versus more than 40 days, advancing the frontier of autonomous chip design.
[NLP-147] Learning Perceptual Geometry of Natural Odor Mixtures from LLM -Derived Descriptors
【速读】： 该论文旨在解决自然气味混合物中分子组成与感知相似性之间关系不明确的问题，尤其针对天然气味由数十至数百种挥发性成分构成所带来的感知形成复杂性。其核心解决方案在于利用大规模语言模型（LLM）作为可扩展的替代方法，从化学-感官提示中生成混合物层面的感知描述，从而为基于分子组成的气味混合物相似性预测提供有效监督信号。关键创新点在于：1）通过化学感官提示引导的语言表示在预测未见混合物的感知相似性方面表现最优；2）在混合物级训练中学习到的低秩分子嵌入能够超越传统组分重叠度量，即使在无共享分子的混合物对上也能实现更优预测；3）尽管训练过程中未使用单分子标签，所学分子几何结构仍与独立的人类气味描述空间高度一致；4）外部分子特征虽在零样本条件下表现出有限的感知相似性对齐能力，但在全监督训练下对预测性能提升有限。总体表明，基于语言的感知描述可为复杂天然混合物的感知组织提供可扩展的监督机制，推动了对分子组成如何决定嗅觉感知的定量理解。
链接: https://arxiv.org/abs/2610.17164
作者: Farhad Pashakhanloo,Venkatesh N. Murthy
机构: Center for Brain Science, Harvard University(哈佛大学脑科学中心); Department of Molecular and Cellular Biology, Harvard University(哈佛大学分子与细胞生物学系); Kempner Institute for the Study of Natural and Artificial Intelligence, Harvard University(哈佛大学自然与人工智能研究学院)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Neurons and Cognition (q-bio.NC)
备注: 26 pages, 14 figures
Abstract:The relationship between odor chemistry and perception remains incompletely understood. Many existing studies focus on the perception of single molecules, whereas natural odors contain tens to hundreds of components, adding extra layer of complexity to percept formation. Here, we explore the relationship between molecular composition and perceived similarity in natural odor objects. We study this using a database of more than 1,200 volatile-compound-in-food mixtures containing more than 8,000 molecular components, and use large language models (LLMs) as a scalable surrogate for deriving mixture-level perceptual descriptions. We find that language-derived representations provide useful supervision for predicting held-out mixture similarities from molecular composition. The prediction was consistently strongest when perceptual targets were derived from chemosensory prompts than from other sensory prompts. Further, during the mixture-level training, we learned low-rank molecular embeddings that enabled better prediction of held-out mixture similarities than component-overlap metrics, even for mixture pairs with no shared molecules. Although no single-molecule labels were used during training, the learned molecular geometry showed significant alignment with the independent human odor-descriptor space. Finally, we explored the use of external molecular features in our mixture-similarity prediction task. These features showed modest zero-shot alignment with mixture-level perceptual similarity. They improved prediction under limited supervision but offered little additional predictive benefit with the full training set. Together, these results show that language-derived descriptions can provide scalable supervision for predicting the perceptual organization of complex natural mixtures, contributing to a quantitative understanding of how molecular composition gives rise to odor perception.
[NLP-148] he Format is the Filter: Scoring Format Fixes What an LLM Judge Can Express
【速读】： 该论文旨在解决大语言模型（LLM）在指令微调数据筛选过程中，评分格式（如二元判断、1–5分制、0–100分制或对比评分）的选择缺乏理论依据与实证验证的问题。其核心挑战在于：不同评分格式对数据筛选结果的偏差与信息量具有显著影响，但现有研究往往忽视了这一关键变量。论文的关键解决方案是通过系统性实验，在固定候选集、评判提示、解码策略和选择预算的前提下，对15,406条UltraFeedback样本采用五种评分格式进行多轮评估，并基于三名评委生成66个微调模型，最终由四名评价者进行下游性能测试。研究发现，0–100分制实际仅提供7–12个有效区分等级，远低于其名义上的连续性；1–5分制的有效区分度约为0–100分制的两到三倍；而二元判断在77%–78%的提示中导致最高分项完全并列，使选择行为近乎随机。此外，所有评分格式的特性几乎独立于具体评委，且传统相关性指标Kendall’s τ_b 会掩盖这些缺陷——它因忽略并列情况而错误地将二元判断排在首位，而τ_a 和顶级一致性指标则将其置于末位。同时，对比类评分存在显著的“槽位偏差”（slot bias），其方向随评委不同而异。尽管经评委筛选的数据在所有评估设置下均优于随机选择，且对文本长度和Markdown格式调整具有鲁棒性，但各评分格式间未表现出可检测的性能差异。因此，论文提出应报告四项关键量化指标，以提升基于评委的数据筛选流程的透明性与可复现性。
链接: https://arxiv.org/abs/2610.17158
作者: Lovedeep Gondara,Jekaterina Novikova,Sagarika Suresh,Jack Ji,Jithin Pradeep
机构: The Vanguard Group, Inc.
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Software Engineering (cs.SE)
备注:
Abstract:When an LLM judge filters instruction-tuning data, its scoring format (binary verdict, 1–5 rating, 0–100 score, or comparison) is rarely justified or ablated. Holding the candidate pool, judge prompt, decoding and selection budget fixed, we score 15,406 UltraFeedback candidates under five formats with three judges, curate, fine-tune 66 models, and evaluate them with four judges. A 0–100 scale is worth only 7–12 equally used levels, two to three times a 1–5 scale rather than twenty, and a binary verdict leaves the top score tied in 77–78% of prompts, so most selections are made at random. These properties are set by the format almost independently of the judge. Kendall’s \tau_b hides them: it ranks binary best while \tau_a and top-1 agreement rank it last, because it discounts the ties a selector must break at random. Both comparative formats inherit a slot bias whose sign differs across judges. Downstream, curated data beat random selection under every evaluation judge, robust to length and markdown adjustment, but no ordering among formats is detectable. We propose four quantities that judge-based curation pipelines should report.
[NLP-149] oolHACE: A Dataset for Detecting Hallucinations in Tool-Based LLM Responses EMNLP2026
【速读】： 该论文旨在解决生成式 AI 在调用外部工具后，仍可能在响应生成阶段产生幻觉（hallucination）的问题，即尽管工具调用正确且返回信息准确，模型仍可能出现与工具输出矛盾、遗漏关键信息或生成无依据陈述等错误。其解决方案的关键在于提出一个细粒度的跨度级（span-level）幻觉检测基准——ToolHACE，涵盖五类精细化的幻觉类型，以系统评估工具增强型语言模型在工具调用后的生成质量。研究发现，此类后工具调用阶段的幻觉普遍存在且难以检测，且依赖通用幻觉数据集的模型在该场景下表现不佳，表明有效的幻觉检测必须基于专门构建的工具增强型语料进行训练，凸显了开发专用基准与模型对于提升工具增强系统可靠性的必要性。
链接: https://arxiv.org/abs/2610.17125
作者: Nikolay Ivanov,Sergey Pletenev,Elisei Rykov,Anatolii Frolov,Bogdan Monogov,Alexander Potekhin,Maria Marina,Daniil Moskovskiy,Viktor Moskvoretskii,Alexander Panchenko,Ilseyar Alimova
机构: Skoltech(斯科尔科沃科学技术研究院); AIRI(人工智能研究机构); EPFL(洛桑联邦理工学院); T-Tech
类目: Computation and Language (cs.CL)
备注: Accepted to Findings of EMNLP 2026. 21 pages, 7 tables
Abstract:Tool-augmented language models ground their responses in external tool outputs, improving reliability beyond parametric knowledge alone. However, even when the correct tool is called and returns accurate information, models may still contradict such outputs, omit critical details, or generate unsupported claims. Existing hallucination benchmarks for tool-augmented dialogue primarily focus on the tool-calling stage, evaluating whether the correct tool and arguments are selected, while overlooking errors that arise during post-tool response generation. To address this gap, we introduce ToolHACE, a span-level benchmark for hallucination detection in tool-augmented LLM responses, covering five fine-grained hallucination categories. Through extensive experiments with modern hallucination detectors, we show that post-tool-call hallucinations are widespread and remain challenging to detect. Our results further demonstrate that effective detection requires training on specialized tool-augmented corpora, while transfer from tool-agnostic hallucination datasets performs poorly in this setting. These findings highlight the need for dedicated benchmarks and models for reliable hallucination detection in tool-augmented language systems.
[NLP-150] D2: High-Capability Full-Duplex Spoken Language Models from Half-Duplex Backbones
【速读】： 该论文旨在解决全双工语音语言模型（FD-SLMs）在通用能力上普遍落后于半双工语音语言模型（HD-SLMs）的问题。其核心挑战在于如何在保持低延迟与高交互效率的前提下，将预训练的半双工模型高效转化为具备强全双工能力的模型。解决方案的关键在于提出一种名为Dialogue-to-Duplex（D2）的框架，通过无解码器的表示蒸馏方式对原始音频编码器进行流式化改造，并结合监督微调实现全双工交互的学习。该方法在细粒度时间尺度上实现了高质量的双向交互，同时系统性地研究了分块大小（chunk size）对全双工交互性能、通用能力及推理计算开销的影响。实验表明，基于D2框架优化的Qwen3-Omni-D2模型在12.5Hz更新频率下可达到99.2%的对话轮次成功率达、521毫秒的低延迟、99.5%的打断成功率，并在VoiceBench评测中取得70.1的当前最优得分，验证了该方案的有效性与先进性。
链接: https://arxiv.org/abs/2610.17124
作者: Cheng-Kuang Chang,Alexander H. Liu,Brian Kingsbury,Hilde Kuehne,Rogerio Feris,Samuel Thomas,James Glass,Kai-Wei Chang
机构: MIT CSAIL(麻省理工学院计算机科学与人工智能实验室); MIT-IBM Watson AI Lab(麻省理工学院-IBM沃森人工智能实验室)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Audio and Speech Processing (eess.AS)
备注: 17 pages, including references and appendices. Demo: this https URL
Abstract:Full-duplex spoken language models (FD-SLMs) support simultaneous listening and speaking, but often lag behind half-duplex spoken language models (HD-SLMs) in general capabilities. We propose Dialogue-to-Duplex (D2), a framework that converts pretrained HD-SLMs into high-capability FD-SLMs at fine temporal granularity. D2 adapts the original audio encoder to streaming through decoder-free representation distillation and learns full-duplex interaction through supervised fine-tuning. We systematically study the effects of chunk size on full-duplex interaction, general capabilities, and inference computation. We apply D2 to two distinct HD-SLM architectures, Qwen3-Omni and LLaMA-Omni 2. At a 12.5Hz update rate, Qwen3-Omni-D2 achieves 99.2% turn-taking success with 521-ms latency, 99.5% interruption success, and a state-of-the-art VoiceBench score of 70.1.
[NLP-151] When LLM Agents See a Conflict but Still Proceed: Cross-Domain Dependencies in Network Slicing
【速读】： 该论文旨在解决在面向未来的生成式AI（Generative AI）原生网络中，当来自不同领域的意图（intent）在共享资源上发生冲突时，基于大语言模型（LLM）的自治代理因未能识别跨域依赖关系而采取不可逆操作所带来的安全风险。其核心问题在于：尽管某些关键资源（如用于保护流量的备用切片）在设计上看似空闲，但实际承载着重要功能，若被成本优化类意图误判并移除，将引发严重后果。论文提出的关键解决方案是，在管理架构层面强制实施跨域依赖验证机制，而非依赖LLM自身对依赖关系的判断。研究通过构建工具化实验环境，评估了来自四个供应商的十种模型在600次运行中的表现，发现多数模型在观察到跨域依赖后仍执行破坏性操作——七种模型至少一次移除了保护切片，高危害场景下该比例达76%；仅一种模型从未移除。研究揭示出两类失败模式：未主动查询依赖关系，或虽查询但无视依赖继续执行。此外，许多模型在识别依赖后仍声称“无冲突”，暴露出当前LLM在决策透明性与一致性上的显著缺陷。因此，研究强调：仅靠增强可见性不足以保障安全，必须在管理架构中嵌入强制性的依赖验证机制，以防止不可逆操作的发生。
链接: https://arxiv.org/abs/2610.17118
作者: Marcos Lima Romero,Ricardo Suyama
机构: NEXt-Gen AI-Native Networks - NEXA, Federal University of ABC - UFABC(联邦大学阿尔伯特·弗朗西斯科·阿贝尔大学 - 阿尔贝卡)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG); Systems and Control (eess.SY)
备注: Submitted to 11th Workshop on Communication Networks and Power Systems (WCNPS 2026)
Abstract:Future AI-native networks are being designed to apply autonomous agents built on large language models (LLMs) in network management through intents. When intents from different domains meet on the same resource, an agent must recognize the cross-domain dependency before taking an irreversible action. We investigate this problem through a smart-grid use case, in which a power utility holds an active intent on a backup slice for protection traffic. This slice appears idle by design, which is precisely the profile targeted by a cost-reduction intent. We built a tool-based environment in which the dependency must be either queried by the agent or read from an inventory annotation, together with a negative control, and evaluated ten models from four different providers in 600 runs. Most models acted against the dependency they had observed: seven of the ten models removed the protection slice at least once after receiving the cross-domain dependency, and in the high-harm tier the slice was removed in 76% of the exposed runs, while four models practically never did. Only one of ten models never removed the slice. We identify two failure modes, not querying the dependency and querying it but proceeding anyway, and show that agents often reported no conflict after observing it. These results indicate that, in AI-native networks, visibility of cross-domain dependencies is not a sufficient safeguard, and that dependency verification should be enforced by the management architecture before irreversible actions rather than delegated to the LLM’s judgment.
[NLP-152] Gluing Local Contexts into Global Meaning: A Sheaf-Theoretic Decomposition of Transformer Representations NEURIPS2026 ICLR
【速读】： 该论文旨在解决现有可解释性工具在评估神经网络表示时，未能验证当语义保持不变但表述方式改变（即改写）时，模型激活值是否仍能保持一致内容表达的问题。其核心解决方案是构建一个基于改写图（paraphrase graph）的细胞层化结构（cellular sheaf），并通过该结构的拉普拉斯算子 $ L_\mathcal{F} $ 在冻结模型上无须训练即可回答内容一致性问题。该方法的关键在于：其零阶上同调核 $ H^0 $ 构成了一个内容子空间，在20维下超越了概念擦除方法 LEACE 在跨八种架构上的保留子空间，在 CounterFact 事实检索任务中平均提升 17.1 个百分点，并通过联合约束同一事实的所有改写形式进一步提升 6.5 个百分点；同时，谱补集（spectral complement）聚焦于因果影响方向，其坐标级代理在消融实验中对模型输出的影响强度比匹配方差坐标的消融高 5.6 至 26.5 倍，覆盖九个不同规模（124M 到 13B 参数）的模型；此外，将层堆叠为网格后生成 $ b_1 = N(L-1) $ 个代数 1-循环，由此产生的模型导出谐波质量在六种架构间变化近三个数量级，与模型操控脆弱性排序高度一致（Spearman $ \rho = 0.90 $，单侧置换检验 $ p = 0.014 $）。因此，同一改写图同时提供了内容子空间、因果影响方向与模型脆弱性度量，实现了多维度统一建模。
链接: https://arxiv.org/abs/2610.17117
作者: Bryce Grant,Peng Wang
机构: Case Western Reserve University (凯斯西储大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: NeurIPS 2026 Poster, ICLR Unifying Concept Representation Learning
Abstract:Existing interpretability tools do not test whether a representation keeps the same content when only the phrasing changes. We treat each paraphrase pair as a local witness that two activations should agree and build a cellular sheaf over the paraphrase graph; its Laplacian L_\mathcalF answers the question on a frozen model with no training. The kernel H^0 is a content subspace: at 20 dimensions it beats the preserved subspace of the concept eraser LEACE on held-out CounterFact fact retrieval across eight architectures by a mean +17.1 ,pp, and constraining all paraphrases of a fact jointly adds a further +6.5 ,pp. The spectral complement concentrates causal influence: ablating the most paraphrase-variable coordinates, a coordinate-wise proxy for it, changes model outputs 5.6 to 26.5\times more than ablating coordinates of matched per-coordinate variability, across nine models from 124M to 13B parameters. Stacking layers into a grid creates b_1 = N(L-1) algebraic 1-cycles, and the resulting model-derived harmonic mass varies by nearly three orders of magnitude across six architectures and ranks them as their steering fragility does (Spearman \rho = 0.90 , exact one-sided permutation p = 0.014 ). The same paraphrase graph yields the content subspace, the influence directions, and the fragility measure. Project page: this https URL
[NLP-153] Calibrated Typed Decisions for Entity Relation and Event Extraction
【速读】： 该论文旨在解决信息抽取模型在跨领域迁移及适应新标注规范时表现不佳的问题，尤其针对实体、关系和事件抽取任务中因类型定义差异导致的泛化能力不足。其核心解决方案是提出一个统一的0.6B参数编码器CalOn，通过联合编码输入文本与类型定义，实现对实体跨度、实体间关系以及事件触发词与论元的端到端联合抽取。该模型的关键创新在于采用统一的概率建模框架，在决策过程中综合考虑所有请求类型的上下文证据，从而在保持抽取结果严格锚定于原始文本的基础上，提升多任务协同识别的准确性。此外，通过引入基于奖励机制的强化学习补充监督训练，进一步优化模型对正确抽取的偏好。实验表明，CalOn在40个基准测试中多数超越现有基线，尤其在关系抽取任务上全面领先，并在事件抽取（使用真实触发词）方面取得最高平均F1值。其跨度对形式还支持方面情感三元组抽取和代词消解等任务。通过训练阶段引入的概率偏差校正，显著提升了抽取结果的置信度估计可靠性，而对F1指标影响极小。
链接: https://arxiv.org/abs/2610.17113
作者: Yang Wang,Hanlin Xiao,Yanan Ma,Chenghao Xiao,Chenghua Lin
机构: University of Manchester(曼彻斯特大学); Mayo Clinic(梅奥诊所); Shanghai University of Finance and Economics(上海财经大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Information extraction models often perform well on the datasets they are trained on, but transferring to new domains and annotation guidelines remains challenging. Supporting user-specified types alone does not resolve this problem. We present CalOn, a unified 0.6B-parameter encoder for entity, relation, and event extraction across datasets and schemas. The model jointly encodes text and type definitions, then directly identifies entity spans, relations between spans, and event triggers and arguments. Its extraction decisions consider the requested types together, using their combined evidence to identify relevant candidates before assigning a type. This gives all three tasks a common probabilistic formulation while keeping every extracted mention anchored to the input text. We train on diverse sources, and supplement supervised learning with a reward that encourages correct extractions and penalises incorrect ones. Across 40 benchmarks, CalOn outperforms the evaluated baselines on most benchmarks. It leads all reported relation extraction benchmarks and achieves the highest average F1 for event triggers and arguments with gold triggers. The same model also performs strongly on aspect sentiment triplet extraction and pronoun resolution through its span-pair formulation. A correction for the probability bias introduced during training improves confidence estimates for entities, relations, and event arguments after calibration, with little change in F1.
[NLP-154] owards Universal Calibration for In-Context Classification AACL
【速读】： 该论文旨在解决生成式 AI（Generative AI）在上下文学习（In-Context Learning, ICL）中因示例选择偏差导致的校准性（calibration）下降问题。现有提升ICL校准性能的方法通常依赖于针对特定任务和模型的额外微调，这违背了ICL“无需微调”的核心优势。本文提出CALIBFORMER，一种通用的ICL校准器，能够无需对新任务或新模型进行微调即可实现跨模型、跨任务的校准泛化。其关键创新在于设计了一种基于Transformer的校准器，该模型利用模型在上下文示例上的预测结果，生成测试输入的置信度评分。实验表明，仅在三个模型和十个任务上训练的CALIBFORMER，即可有效校准多个新任务及四类未见于训练阶段的大语言模型（LLM），即使在前沿模型Gemini上也实现了超过50%的校准误差降低，充分验证了其通用性与有效性。
链接: https://arxiv.org/abs/2610.17096
作者: Awik Dhar,Sunita Sarawagi,Arpit Agarwal
机构: Indian Institute of Technology Bombay (印度理工学院孟买分校)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 30 pages, 11 figures, 19 tables, AACL-IJCNLP Findings accepted
Abstract:In-Context Learning (ICL) has received a lot of interest recently due to its ability to improve classification accuracy for new tasks without having to fine-tune the model over these tasks. However, the impact of ICL on calibration has been far less studied. Recent work has shown that ICL can have a negative impact on calibration due to biased selection of in-context examples. The existing methods for improving ICL calibration require additional task and model-specific training, detracting from the no-fine-tuning convenience of ICL. In this work, we introduce CALIBFORMER, a universal ICL calibrator for classification tasks, that generalizes to new models and tasks without further fine-tuning on these tasks. Specifically, we design a transformer-based calibrator which uses the model predictions on in-context examples to predict a confidence score for the test input. We show that the calibrator trained on three models and ten tasks, is able to generalize/calibrate ICL predictions on several new tasks and four LLMs unseen during training. Even for frontier models like Gemini, CALIBFORMER is able to obtain over 50% reduction in calibration error.
[NLP-155] UZHCL_for_the_win at Touche: Encoder Fine-Tuning Synthetic Augmentation and an Underpowered Encoder-LLM Comparison
【速读】： 该论文旨在解决在线社交平台（如Reddit）中论点谬误检测与论证模式分类的联合任务挑战，尤其针对短文本、话题广泛且包含大量非正式谬误的语境。其核心问题是如何在有限标注数据下有效识别并区分多种类型的逻辑谬误及论证结构。解决方案的关键在于采用经过微调的RoBERTa-large编码器作为主干模型，并结合任务特定的输出头；同时引入少量由大语言模型（LLM）生成的合成对比样本以增强模型区分能力；在增强赛道中，进一步利用未标注数据池中置信度过滤后的伪标签样本进行半监督学习。实验表明，在当前数据规模下，尽管对比了包括DeBERTa-v3、ModernBERT及多个Qwen系列大模型在内的多种架构，但均未展现出显著优于RoBERTa-large的表现，提示该比较因5折交叉验证方差过大而缺乏统计效力。此外，研究揭示了增强赛道性能提升部分源于组织者重写输入所隐含的标签信息，而非真正的论据澄清。在官方Touche 2026测试集上，系统在基础赛道的两类分类任务（ST2和ST3）中排名第一，在二分类任务（ST1）中排名第二，而增强赛道的分类任务则由级联与多任务系统取得最优成绩。最后，论文指出了ST3任务中罕见类别的识别瓶颈，并提出了具有实际意义的后续研究方向。
链接: https://arxiv.org/abs/2610.17077
作者: Psychias Stylianos
机构: 未知
类目: Computation and Language (cs.CL)
备注: 19 pages. Working notes for Touche at CLEF 2026 (Jena, Germany)
Abstract:Online discussions on platforms like Reddit are a productive testbed for fallacy detection: short texts, broad topical range, and the full inventory of informal fallacies. The Touche 2026 lab at CLEF formalises this as a shared task on a Reddit-derived corpus that combines fallacy detection with argument-scheme classification. This paper describes our participation in all three subtasks (ST1 binary fallacy detection, ST2 eight-way fallacy classification, ST3 four-way argument-scheme classification) and both evaluation tracks. Our submitted systems are fine-tuned RoBERTa-large encoders with task-specific heads, augmented with a small batch of LLM-generated synthetic contrastive pairs; for ST2 and ST3 on the enhanced track we additionally added confidence-filtered pseudo-labelled examples from the unlabelled pool. A wider backbone comparison spanning additional encoders (DeBERTa-v3, ModernBERT) and an LLM track (four Qwen variants from 3B to 32B, zero-shot through CoT-LoRA) showed no validated benefit over RoBERTa-large at this corpus scale, with 5-fold variance wide enough that we call the comparison underpowered rather than settled. We also characterise how much of the enhanced track’s lift reflects organiser-rewritten inputs carrying label information rather than genuine argument clarification. On the official Touche 2026 test set our systems ranked first on both base-track classification cells (ST2 and ST3) and second on both binary-detection (ST1) tracks, with the two enhanced-track classification cells topped by cascaded and multitask systems. We close with the rare-class limitation on ST3 and the most useful follow-ups. Code: this https URL
[NLP-156] AD4EDU: Benchmarking University Lecture Audio Description across Accessibility Standards Metrics and Learner Preferences
【速读】： 该论文旨在解决高等教育场景中基于幻灯片的讲座音频描述（Audio Description, AD）严重缺乏研究与实践的问题。现有音频描述体系主要针对娱乐内容设计，其生产指南、评估指标及基准数据集无法适配讲座场景中对语义互补性、指示性指代消解以及时间间隙精准匹配的特殊需求。为此，本文提出AD4Edu框架，包含一个面向教育场景的音频描述基准数据集AD4Edu-Bench和一个规则可诊断的评估指标AD4Edu-Eval。关键解决方案在于整合了WCAG 2.2、DCMP描述核心原则及NCAM STEM描述实践等权威无障碍指导规范，并通过四位盲人或低视力（BLV）用户参与构建银级参考描述、判断排序偏好对与成对偏好数据，采用直接偏好优化（Direct Preference Optimization, DPO）方法对模型进行后训练，以实现对细节层面的深入分析。研究发现，尽管评估指标能可靠识别标准违规行为，但符合标准的描述并不一定满足BLV用户的主观偏好，尤其是在描述长度方面，凸显了在包容性人工智能解决方案设计中必须让BLV用户深度参与共同设计的必要性。
链接: https://arxiv.org/abs/2610.17042
作者: Psychias Stylianos,Gao Yingqiang
机构: University of Zurich (苏黎世大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Audio description (AD) aims to enhance the accessibility of visual media for blind and low-vision (BLV) persons. However, in higher education, where slide-based lectures are a default mode of teaching, lecture AD remains largely underexplored. The main reason is that existing production guidelines, evaluation metrics, and benchmark datasets are oriented toward entertainment content and do not transfer to lectures, where ADs must complement the lecturer’s speech, resolve deictic references, and fit within narrow pauses. In this work, we propose AD4Edu, consisting of an AD benchmark, AD4Edu-Bench, and a rule-diagnosable evaluation metric, AD4Edu-Eval, for generating and evaluating ADs in higher education. Specifically, we consolidated established accessibility guidance, including WCAG 2.2, the DCMP Description Key, and NCAM’s STEM description practices. We further collected silver reference ADs, judge-ordered preference pairs, and pairwise preferences from four BLV raters, and post-trained models with direct preference optimization (DPO) for detailed analysis. Our findings indicate that while the metric reliably detects violations of the standard, ADs that comply with it do not necessarily fulfill the subjective preferences of BLV persons, most notably on description length, underscoring the necessity of active participation of BLV persons in co-designing inclusive AI solutions.
[NLP-157] Beyond Model Bias: Cross-Entropy Based Rankings in Literary and Academic Texts Remain Stable across Small Language Models
【速读】： 该论文旨在解决语言模型（Language Model, LM）中文本可预测性评估的稳定性问题，即不同语言模型计算出的文本意外度（surprisal）排序是否具有跨模型一致性。其核心问题是：基于交叉熵相关指标（如每个数据点的均值与方差）所得到的文本不可预测性相对程度，在不同小规模Transformer模型之间是否保持稳定。解决方案的关键在于通过实证研究验证了在给定文本数据集下，四种不同架构的小型Transformer模型在当代诗歌和学术论文两组风格迥异的语料中，其交叉熵均值与方差的排序具有高度一致的跨模型相关性（ρ > 0.90）。进一步地，研究通过与六种无参数压缩基线进行对比，发现语言模型间的共识显著高于基线模型之间的共识，排除了文本复杂性等通用因素的干扰。此外，对结构受限的学术语域进行几何分析表明，高阶矩对特定模型敏感，但整体排序一致性依然保持。研究结果表明，文本不可预测性的相对程度可能独立于所使用的具体语言模型，为基于语言模型的文本分析提供了跨模型可比性的理论支持。
链接: https://arxiv.org/abs/2610.17040
作者: Ankit Bhattacharjee,Indrasabarna Hari,Krityapriya Bhaumik,Arka Das,Piyush Sarangi
机构: Indian Institute of Technology Kharagpur (印度理工学院克哈普尔分校); Asutosh College, University of Calcutta (加尔各答大学阿苏托斯学院)
类目: Computation and Language (cs.CL)
备注:
Abstract:While language models provide powerful mechanisms for evaluating textual predictability, it remains unclear whether the relative extents of LM surprisal remain stable across models. This study presents empirical evidence of the invariance of ranking based on cross-entropy related measures (e.g. mean, variance of each data-point), across four distinct small scale transformer-based models, for a given dataset of text sources. We consider two stylistic environments: a corpus of contemporary poetry and a corpus of academic research papers. We demonstrate that mean and variance of cross-entropy maintain robust cross-model rank correlation ( \rho 0.90 ) across both regimes. To test whether this agreement merely reflects generic differences in textual complexity, we compare it against six parameter-free compression baselines. The extent of agreement among language models is found to be substantially higher than that among model-baselines, for most of the measures. Geometric analysis of the structurally constrained academic regime reveals that higher-order moments are sensitive to specific models, although strong agreement persists. The findings suggest that relative extents of textual unpredictability may be independent of the language model they are computed against.
[NLP-158] MENTOR: Procedural Shielding of Tool-Using LLM Agents
【速读】： 该论文旨在解决大语言模型（LLM）代理在执行由程序手册指导的工作流（如公司差旅预订或硬件采购流程）时，如何确保其行为符合自然语言描述的复杂规程问题。现有方法依赖LLM评判器或手动编写的规则在运行时局部阻止不当工具调用，但存在两大缺陷：一是仅禁止错误操作而未提供正确推进任务的引导，导致代理缺乏有效行动策略；二是缺乏上下文敏感性，易引发过度保守行为。为此，本文提出MENTOR框架，其核心创新在于构建一种基于历史依赖的运行时防护机制（runtime shield），通过多步自动形式化管道从程序手册中生成该防护机制，并采用基于博弈论的技术，在不过度限制代理行为的前提下增强对运行时不确定性的鲁棒性。该方法不仅提供符合规程的行动指导，还实现上下文感知的合规性保障，显著提升轻量级开源LLM的合规表现，甚至超越未受保护的专有模型。由于防护机制可一次性为每份手册生成并复用于所有任务，具备良好的成本效益，适用于重复执行的工作流场景。
链接: https://arxiv.org/abs/2610.17039
作者: Ashwani Anand,Ivi Chatzi,Ritam Raha,Anne-Kathrin Schmuck
机构: Max Planck Institute for Software Systems, Kaiserslautern, Germany; TNO-ESI, Eindhoven, The Netherlands
类目: Computation and Language (cs.CL); Machine Learning (cs.LG); Logic in Computer Science (cs.LO)
备注:
Abstract:Tool-using LLM agents are increasingly used to automate workflows governed by procedural manuals, such as company rules for booking travel or procuring hardware. These manuals are written in natural language for humans and describe general requirements over such complex workflows. Ensuring compliance of LLM agents with these manuals is difficult, and state-of-the-art techniques use either LLM judges or manually written rules to forbid problematic tool-calls locally at runtime. This is insufficient because (i) only blocking tool calls keeps the agent clueless about how to make correct progress towards fulfilling its task, and (ii) can lead to overly conservative behavior of the agent, as tool calls might only be problematic in certain contexts. To overcome these challenges, we introduce MENTOR, which (i) additionally provides guidance for the agent to make procedural-compliant progress on the task, and (ii) ensures that this compliance is context-aware. We achieve these goals by building a novel history-dependent runtime shield that is obtained from the procedural manual via a multi-step autoformalization pipeline. At its core, MENTOR utilizes a game-based technique to ensure robustness against runtime uncertainty without over-resticting agents’ behavior. This yields a formally assisted framework for procedural shielding of LLM agents that improves compliance well beyond existing methods. In particular, MENTOR lifts the compliance of shielded light-weight open-source LLMs above that of non-shielded proprietary models. Since a shield is synthesized once per manual and then serves every task, a MENTOR-shielded agent becomes a viable choice for repeatedly executed workflows, potentially reducing deployment costs when the offline construction cost is amortized over repeated tasks.
[NLP-159] Understanding Personalization Gains in Whisper
【速读】： 该论文旨在解决生成式语音识别（ASR）模型在零样本（zero-shot）场景下存在残差误差分布不均的问题，即模型在不同数据集、录音条件及基于元数据的分组中表现差异显著。其核心挑战在于：如何有效提升模型在困难子群体中的识别性能，同时区分“个性化适应”与“通用适配”的实际贡献。解决方案的关键在于评估轻量级低秩适配（LoRA）在不同策略下的有效性，具体比较了零样本Whisper、共享式LoRA（pooled LoRA）与分组特定LoRA（group-specific LoRA）的表现。研究发现，适配收益并非均匀分布，主要集中在高难度的Common Voice子组，在VoxPopuli上保持稳定，而在低错误率的LibriSpeech上几乎消失。更为关键的是，共享式LoRA通常能媲美甚至显著优于分组特定适配，表明多数性能提升可由通用适配实现，而非必须依赖于分组特异性参数。因此，研究强调：在评估适配效果时，应以匹配的共享式适配为基准，避免将性能增益误判为个性化需求的证据。
链接: https://arxiv.org/abs/2610.17019
作者: Sergei Katkov,Antonio Liotta
机构: Free University of Bozen-Bolzano(博岑-波尔扎诺自由大学), Bolzano, Italy(意大利博岑)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 14 pages. Presented at SPECOM 2026
Abstract:Whisper-style ASR models provide strong zero-shot transcription, but their residual errors are not evenly distributed across datasets, recording conditions, and metadata-based groups. We study when lightweight low-rank adaptation (LoRA) improves Whisper and whether observed gains require group-specific adapters. Across Common Voice, LibriSpeech, and VoxPopuli, we compare zero-shot Whisper, pooled LoRA, and group-specific LoRA under matched within-group splits. We find that adaptation is not uniformly beneficial: gains are concentrated in difficult Common Voice groups, are consistent on VoxPopuli, and largely disappear on low-error LibriSpeech. Crucially, pooled LoRA often matches, and sometimes clearly outperforms, group-specific LoRA. Gains over zero-shot should therefore be tested against matched pooled adaptation before being interpreted as evidence of personalization.
[NLP-160] Predicting Mental Health Help-Seeking Behaviour on Reddit: A Survival Analysis of Linguistic and Behavioural Signals
【速读】： 该论文旨在解决传统心理健康求助预测研究中将求助行为视为事后分类问题的局限性，即无法在个体实际进入临床社区前识别其潜在的求助过渡。为此，作者将求助行为建模为时间至事件（time-to-event）问题，区分“谁会寻求帮助”与“何时会寻求帮助”两个关键问题。解决方案的关键在于构建双臂队列：一是包含6,337名普通用户的对照组（真实右删失，12.1%转化率），用于预测谁将发生转变；二是包含5,023名最终求助者的临床组，用于预测转变的紧迫性。所有特征均基于用户在首次临床活动前的发帖内容提取，采用用户分组五折交叉验证与四年的时序保留测试以防止信息泄露。实验对比了TF-IDF、MentalBERT嵌入和行为活动特征，在Cox比例风险模型、随机生存森林及LSTM序列模型下进行评估。结果显示，行为特征在预测“谁会求助”时表现优于文本特征（C-index 0.797 vs. 0.763），二者结合效果最佳（交叉验证下C-index 0.823，保留测试下0.856）；而在预测“何时求助”时，文本特征保持稳定（C-index 0.698），行为特征则趋于不稳定。这一发现表明，结构性参与模式蕴含着超越语言内容的未来求助预测信息，且将“谁”与“何时”分离分析可实现更清晰、更具解释性的建模。
链接: https://arxiv.org/abs/2610.16983
作者: Nisar Ahmed,Ammara Kazmi,Irtaza Bashir Raja,Didier Devaurs
机构: 未知
类目: Computation and Language (cs.CL); Machine Learning (cs.LG); Social and Information Networks (cs.SI)
备注: 28 pages, 16 figures, 7 tables
Abstract:Most prior work on predicting mental health help-seeking treats it as post-hoc classification of users already active in clinical communities, which cannot tell whether someone approaching a transition can be identified in advance. We instead frame help-seeking as a time-to-event problem and ask which Reddit users will eventually seek mental health support, and how imminent that transition is. We build a two-arm cohort: a control frame of 6,337 ordinary users with genuine right-censoring (12.1% transition rate) to answer who will transition, and a clinical frame of 5,023 eventual transitioners to answer how soon. Features come only from posts written before any clinical activity, and user-grouped five-fold cross-validation plus a four-year temporal holdout guard against leakage. We compare TF-IDF, MentalBERT embeddings and behavioural activity features using Cox proportional hazards models, Random Survival Forests and an LSTM sequential model. Behavioural features alone outperform text for predicting who will seek help (C-index 0.797 vs. 0.763), and combining both is best (0.823 under cross-validation, 0.856 on the holdout). The pattern reverses for predicting how soon a known transition will happen: text remains a stable predictor (C-index 0.698) while behavioural features become unstable. Structural engagement patterns therefore carry information about future help-seeking beyond linguistic content, and separating who from when yields a cleaner analysis.
[NLP-161] Event Memory and Narrative Dynamics: A Large Language Model Framework for Multidimensional State Spaces
【速读】： 该论文旨在解决自传体记忆研究中事件边界界定与事件间结构关系建模的难题，核心挑战在于如何从自然语言叙事中客观、可追溯地识别事件单元并揭示其动态组织规律。解决方案的关键在于构建一个基于大语言模型（Large Language Model, LLM）的框架，该框架结合海马体（hippocampal）模型对情景记忆与导航的理论假设，通过显式边界规则和受神经机制启发的特征架构，实现对叙事文本的逐段解析与事件状态（event-state）的生成。该框架在五名参与者访谈数据上的应用中成功构建了256个事件状态，并利用多对应分析（Multiple Correspondence Analysis）将这些状态嵌入高维特征空间。结果显示，连续事件状态间的距离显著小于随机排列下的期望值，且在选定的4维十盆状解中，盆地状态的特征轮廓相似性与单步路径转移模式的相似性高度一致；上下文、机制与主题因素有效区分了全系统路径的参与模式。进一步引入后继表示（successor representation）模型，发现盆地中心点间距离与完整后继特征距离之间存在正相关关系，在所有五个解中均得到验证，且在至少两个解中跨三种几何构型表现出一致性。该框架将海马体驱动的事件分割与预测性组织原则转化为可溯源的事件边界与状态表征，为检验神经层面事件分割与自然回忆中的多步预测组织提供了可计算、可验证的分析基础。
链接: https://arxiv.org/abs/2610.16971
作者: Joshua Sariñana
机构: 未知
类目: Computation and Language (cs.CL)
备注: 60 pages, including supplementary materials. Code and data: this https URL
Abstract:Studying event memory requires defining events and their relationships within remembered experience. Autobiographical narratives provide evidence of this organization, but their analytical units must first be established. This study introduces a large language model framework for constructing event representations from narratives and modeling their structure and dynamics with excerpt-level traceability. Explicit boundary rules and a feature schema informed by hippocampal models of episodic memory and navigation guide event construction. In a proof-of-concept application to interviews from five participants, the framework constructed 256 event-states. Multiple correspondence analysis embedded these states in feature-based spaces, with analyses focused on a selected 4-D geometry and 10-D and 13-D reference geometries. Consecutive event-states were significantly closer than expected under randomized narrative order. In the selected 4-D ten-basin solution, similarity between basin-state feature profiles corresponded with similarity between one-step routing profiles; context, mechanism, and theme differentiated full-system pathway engagement. A successor representation modeled expected basin-state occupancy across successive steps. Correlations between basin-centroid distances and complete successor-profile distances were positive across all five basin solutions, with exploratory tests supporting correspondence in at least two solutions across all three geometries. These findings connect feature-based event representations with observed narrative transitions and modeled multistep organization. The framework translates hippocampal principles into source-linked event boundaries and state representations, providing an analytical basis for testing neural accounts of event segmentation and predictive organization during naturalistic recall.
[NLP-162] SarcasTraps: Effect of Sarcasm on Turkish Hate Speech Detection
【速读】： 该论文旨在解决土耳其语中讽刺性仇恨言论（sarcasm-based hate speech）对现有仇恨言论检测模型性能造成负面影响的问题。传统仇恨言论数据集多聚焦于显性有害语言，而忽视了通过反讽、夸张或表面礼貌等间接表达方式传递的隐性伤害，其中讽刺是一种典型形式。为系统评估讽刺对检测效果的影响，研究提出SarcasTraps——一个针对土耳其语的定向外部测试集，包含479条同时标注仇恨言论与讽刺属性的数据。研究采用BERTurk和XLM-RoBERTa在已有土耳其语仇恨言论数据集上训练分类器，并在SarcasTraps上进行严格外部测试；同时以零样本方式评估GPT-4.1与Llama-3.3-70B的表现。结果表明，经过微调的编码器模型在面对讽刺性仇恨言论时召回率显著下降，而零样本大语言模型虽保持较高召回率，但精确率大幅降低。这一发现揭示了当前主流模型在处理讽刺性内容时均存在显著局限性，而SarcasTraps通过提供具有针对性的土耳其语数据，为该挑战提供了量化证据。由于数据包含敏感及仇恨内容，仅限学术研究用途，经申请后方可获取。
链接: https://arxiv.org/abs/2610.16957
作者: Niyazi Ahmet Metin,Dilara Keküllüoğlu
机构: Sabanci University (萨班奇大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 12 pages
Abstract:Standard hate speech detection datasets often emphasize explicit harmful language, but directed harm can also be expressed indirectly, without clearly hateful wording. Sarcasm is one such form of indirect expression, as it can convey harmful intent through mock praise, exaggeration, humor, or apparent politeness. In this paper, we introduce SarcasTraps, a targeted external test set for analyzing the effect of sarcasm on Turkish hate speech detection. SarcasTraps contains 479 entries annotated along two dimensions: hate speech and sarcasm. We train classifiers using BERTurk and XLM-RoBERTa on two existing Turkish hate speech datasets and evaluate them on SarcasTraps, using it strictly as an external test set. We also evaluate GPT-4.1 and Llama-3.3-70B in a zero-shot setting on SarcasTraps. Results show that sarcasm substantially affects hate speech detection. Fine-tuned encoder models lose recall on sarcastic hate, while zero-shot LLMs maintain higher recall but suffer substantial precision drops on sarcastic examples. These findings show that both fine-tuned hate speech detectors and zero-shot LLMs struggle when sarcasm is involved, and SarcasTraps provides quantitative evidence of this challenge through a targeted Turkish test set. Due to the presence of sensitive and hateful language, SarcasTraps will be shared upon request for academic research purposes. Warning: This paper has examples from the dataset which contains offensive and hateful content towards groups of people.
[NLP-163] Lexical Redefinition Reveals Context-Sensitive Instruction Following EMNLP2026
【速读】： 该论文旨在解决语言模型在面对提示（prompt）中存在冲突的上下文证据时，如何进行语义整合与判断的问题。其核心关注点在于：当显式指令（explicit instruction）对某一词汇赋予特定含义，而后续句子又通过上下文使用方式提供相反或矛盾的语义线索时，模型如何权衡并最终确定词义。研究采用词汇重新定义（lexical redefinition）作为实验范式，在870组词对中系统考察显式指令与后续上下文之间的互动关系。结果表明，词汇重新定义并非被模型以固定规则执行，而是其行为会系统性地受指令与上下文之间关系的影响——具体表现为上下文用法可保留、削弱甚至完全逆转初始指令所规定的词义。这揭示了语言模型对显式指令的解释仍持续敏感于后续上下文信息，说明其语义理解是动态且情境依赖的。因此，词汇重新定义为研究语言模型在提示内部整合多源信息的机制提供了一个可控的行为范式。
链接: https://arxiv.org/abs/2610.16952
作者: Roksana Goworek,Haim Dubossarsky
机构: Queen Mary University of London (伦敦玛丽女王大学); University of Cambridge (剑桥大学)
类目: Computation and Language (cs.CL)
备注: Accepted at the 3rd Workshop on Language Understanding in the Human-Machine Era (LUHME 2026), co-located with EMNLP 2026
Abstract:How do language models reconcile conflicting contextual evidence within a prompt? We study this question using lexical redefinition, where an explicit instruction specifies how a word should be interpreted and a subsequent sentence provides additional evidence through its usage. Across 870 word pairs, we find that lexical redefinitions are not applied as fixed rules. Instead, the behaviour depends systematically on the relationship between the explicit instruction and the intervening sentence: the contextualised usage can preserve, weaken, or completely reverse the instruction. These results suggest that the interpretation of an explicit instruction remains sensitive to subsequent contextual evidence. Lexical redefinition therefore provides a controlled behavioural paradigm for studying how language models integrate multiple sources of information within a prompt.
[NLP-164] Estimating Uncertainty of Omnimodal Large Language Models via External Omnimodal Embeddings
【速读】： 该论文旨在解决生成式多模态大模型（Omni-modal Large Language Models, OLLMs）在推理过程中缺乏可靠不确定性量化（Uncertainty Quantification, UQ）方法的问题。现有方法普遍存在局限性，如仅支持最多两种模态、依赖模型内部结构信息或需重新调用被评估模型进行推断。为此，本文提出一种无需访问模型内部参数（如logits、隐藏状态或梯度）且不需额外推理的模型无关型不确定性量化框架，通过外部多模态嵌入对思维链（Chain-of-Thought）响应进行评分，从而衡量推理过程在多模态输入中的语义锚定性及其内部一致性。其核心贡献在于构建了一组可互补的几何信号，并实证分析了这些信号在不同生成器和模态下的可靠性：尽管各类几何分数之间高度相关（分组内秩相关系数0.91–0.94），但与基于答案离散度的基线方法具有强互补性；同时发现其可靠性受生成器类型和输入模态影响显著。该框架仅需借助辅助模型对推理轨迹进行分段，并使用规模极小的标注校准集（最小可至25个样本）即可实现有效校准，且在多数情况下固定单一信号的先验设定已能提供接近全集校准的效果。在UNO-Bench基准测试中，该方法在封闭与开放模型上均优于自一致性（self-consistency）策略，在开放式多模态问答任务中表现媲美甚至超越自言式（self-verbalisation）及基于内部状态的先进基线，尤其在以往主流方法表现不佳的复杂多模态场景下展现出优越性能。
链接: https://arxiv.org/abs/2610.16933
作者: Iacopo Ghinassi,Hy Nguyen Gia La,Luke Ong
机构: Nanyang Technological University (南洋理工大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Recent work has investigated Omni-modal Large Language Models (OLLMs), models which can take multiple modalities and their combinations as input, as a way forward to have systems really capable of interacting with their environment. Uncertainty quantification (UQ) for OLLMs, however, remains an open problem: existing methods largely target at most two modalities at a time, require access to model internals, or re-invocation of the evaluated model. We propose a model-agnostic UQ framework that scores a Chain-of-Thought response entirely through external omni-modal embeddings, measuring the grounding of the reasoning in the multimodal input and its internal coherence. Our contribution is the family of signals and an empirical characterisation of when it is reliable, rather than any single estimator: we find the geometric scores to be largely redundant with one another (mean within-split rank correlation 0.91-0.94) but strongly complementary to answer-dispersion baselines, and their reliability to depend on the generator and the modality. The resulting confidence estimates require no access to logits, hidden states or gradients of the evaluated model and no additional inference from it; they do require segmenting the reasoning trace with an auxiliary model and, for the aggregated score, a labelled calibration set that we show can be as small as 25 examples (although we also show that such a set buys little over fixing one signal a priori). On UNO-Bench, with both a closed and an open model, our scores yield small gains over self-consistency (systematic with the closed generator, mixed with the open one) and are competitive with self-verbalisation and internal-state baselines, especially in open-ended, omni-modal questions where previous state-of-the-art baselines struggle.
[NLP-165] SpecGuardBench: Can Coding Agents Construct Reliable Validation Harnesses from Natural Language Specifications?
【速读】： 该论文旨在解决当前编码代理（coding agents）在验证代码正确性时依赖测试用例，但现有测试无法充分反映自然语言规格说明（natural language specification）所要求的语义完整性这一核心问题。其关键挑战在于：尽管测试是判断代码是否正确的主要依据，但通过测试并不足以证明实现完全满足规格说明，尤其当测试用例未能忠实捕捉规格的逻辑细节或覆盖关键执行路径时。为此，论文提出SpecGuardBench基准，涵盖来自13个真实开源仓库的291个任务，涉及8类验证逻辑（如功能后置条件、交互协议等），要求代理基于自然语言规格说明、代码仓库上下文及参考实现，自主构建包含可执行检查（checks）和测试场景（test scenarios）的验证套件。理想情况下，检查应能拒绝违反规格的实现，同时接受符合规格的实现；测试场景应全面覆盖代码行为。实验表明，四种主流编码代理在超过63%的任务中无法构建出被评估为“忠实”的验证套件，其瓶颈并非测试场景探索能力不足，而在于生成的检查逻辑往往机械模仿参考实现而非忠实反映规格说明，导致对合法但不同的实现产生误拒。因此，解决方案的关键在于提升代理从规格说明中抽象出精确、鲁棒且语义对齐的验证检查的能力，而非简单复制已有实现模式。
链接: https://arxiv.org/abs/2610.16870
作者: Zhiqing Zhong,Mingyu Chen,Aoyang Fang,Yuzhong Zhang,Haoye Tian,Lionel Briand,Boxi Yu,Pinjia He
机构: The Chinese University of Hong Kong, Shenzhen(香港中文大学（深圳）); Aalto University(阿尔托大学); Lero, the Research Ireland Centre for Software(爱尔兰软件研究中心); University of Limerick(利默里克大学); University of Ottawa(渥太华大学)
类目: Computation and Language (cs.CL); Software Engineering (cs.SE)
备注: Paper Under Review
Abstract:As coding agents become more capable, tests remain the primary evidence for judging whether their code is correct. Yet passing available tests does not establish that an implementation satisfies its natural language specification. Reliable validation requires checks that faithfully capture the specification and scenarios that thoroughly exercise the code, but current benchmarks provide limited insight into whether agents can construct such evidence themselves. We introduce SpecGuardBench, a benchmark of 291 tasks from 13 real-world repositories, covering 8 types of verification logic such as functional postconditions and interaction protocols. Given a natural language specification, repository context, and a reference implementation, an agent constructs a validation harness comprising executable checks and test scenarios. The checks should reject implementations that violate the specification while accepting those that satisfy it, and the scenarios should thoroughly exercise the code. Across four proprietary and open-weight coding agents, none constructs a harness scored as faithful on the benchmark’s implementations for more than 63% of the tasks. The bottleneck lies less in exploring test scenarios than in constructing checks that faithfully reflect the specification. Specifically, the agents show a clear tendency to write checks that mirror the reference implementation rather than the specification, and therefore reject implementations that the specification permits.
[NLP-166] Davinci-Tool: Scaling Agent ic Tool-Use Supervision with Verifiable Executable Environments
【速读】： 该论文旨在解决生成式 AI 在复杂多步骤工具调用任务中缺乏可靠执行验证的问题，核心挑战在于现有训练数据难以同时覆盖广泛的任务环境与精确的执行结果验证。为应对这一问题，其关键解决方案是构建一个统一的、可还原的执行基底——Davinci-ToolData，该数据集联合建模环境覆盖范围与结果可验证性，通过保留可恢复的状态、执行轨迹、最终结果及可执行的验证器，实现对工具使用流程的端到端监督。其核心技术由两个互补管道构成：执行基底任务合成（EGTS）从独立开发的系统中提取经验证的示范轨迹；可执行环境合成（EES）则生成可审计的环境、状态驱动的任务及可执行验证器，支持单领域与协同领域的扩展。基于此数据集，研究采用监督微调与直接偏好优化相结合的方法，利用EES生成的验证器进行动作对比与偏好构建，显著提升了模型性能。实验表明，仅300亿参数的Davinci-Tool在τ²-bench、BFCL v4多轮和MCP-Atlas三项基准上分别达到88.2、62.8和56.2的得分，超越参数量达其16倍的Qwen3-Coder-480B-A35B-Instruct模型，证明了该方法在高效、高可靠性工具使用能力上的优越性。
链接: https://arxiv.org/abs/2610.16867
作者: Xiaoyang Yuan,Ji Zhao,Xingyu Hu,Wenhan Wang,Yutao Hu,Qiaosheng Zhang,Pengfei Liu,Yu Qiao,Yi Bin,Wenqi Shao
机构: Tongji University (同济大学); Shanghai Innovation Institute (上海创新研究院); University of Science and Technology of China (中国科学技术大学); Southeast University (东南大学)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 27 pages, 4 figures, 13 tables. Code: this https URL
Abstract:Reliable agentic tool use requires training data that pair diverse multi-step workflows with executable evidence that the requested outcomes were achieved. Scaling either trajectory coverage or verification alone is insufficient as new environments introduce distinct states, rules, and dependencies that must also be reflected in the success criteria. To address this challenge, we introduce Davinci-Tool, a 30B tool-use model, and Davinci-ToolData, an execution-grounded dataset designed to scale environment coverage and outcome verification together. Davinci-ToolData preserves restorable states, execution traces, outcomes, and verifiers as a unified substrate for supervision. Two complementary pipelines populate this substrate: Execution-Grounded Task Synthesis (EGTS) derives verified demonstrations from independently authored systems, and Executable Environment Synthesis (EES) synthesizes auditable environments, state-grounded tasks, and executable verifiers across single and coordinated domains. These pipelines yield 30,395 verified trajectories spanning 2,260 invoked tool names. We first train on these demonstrations with supervised fine-tuning, then reuse EES states and verifiers to compare alternative actions and construct outcome-grounded preferences for Direct Preference Optimization. Davinci-Tool improves Qwen3-Coder-30B-A3B-Instruct from 52.4 to 88.2 on \tau^2 -bench, from 30.4 to 62.8 on BFCL v4 multi-turn, and from 16.9 to 56.2 on MCP-Atlas, outperforming Qwen3-Coder-480B-A35B-Instruct on all three benchmarks with only 1/16 of its total parameters. Code is available at this https URL.
[NLP-167] ANCHOR: Assessing Naturalistic Constraint Hierarchies and Ordering in Repository-Grounded Agent ic Coding
【速读】： 该论文旨在解决当前基于代码仓库的编程智能体（coding agents）在生成功能正确补丁时，往往忽视开发者实际工程约束的问题。现有评估体系普遍聚焦于功能正确性，而忽略了开发过程中对修改范围、工作流顺序、验证机制、工具使用及报告规范等非功能性约束的要求，这些约束共同构成了“工程约束遵循”（engineering constraint following）能力。为填补这一空白，本文提出ANCHOR（Assessing Naturalistic Constraint Hierarchies and Ordering in Repository-Grounded Agentic Coding）基准，其核心创新在于：（1）基于真实开发者交互数据构建涵盖10个类别、78个细粒度类型的层次化约束分类体系；（2）设计自动化流水线，将任务相关的原子级约束实例化并组合为带有AND/THEN/PRIORITY语义的关系结构；（3）引入基于轨迹的评估框架，结合规则引擎与大语言模型（LLM）双重判断，对代码变更差分（code diff）与执行轨迹进行联合验证。实验结果表明，尽管最强系统（Claude Opus 4.8 under Cline）在原子级约束遵守率上达到70.1%，关系级遵守率为63.7%，但仅有19.0%的任务满足所有检查项的严格全项合规标准（FCCR）。五名工程师对50个实例的审计显示约束具有高可行性（94%）、仓库一致性（94%）和现实性（96%）。在主对比组中，引入约束后任务完成率平均提升0.7个百分点，但个体差异显著，揭示出模型与架构之间存在复杂交互效应。研究明确区分了功能成功与工程约束遵循之间的差距，识别出工作流排序、工具使用规范性和报告完整性是主要失败瓶颈。
链接: https://arxiv.org/abs/2610.16859
作者: Wenqiang Zhu,Xinping Lei,Zhiyi Lai,Zhengshuo Gong,Tianzhuang He,Songwei Yu,Weihao Li,Jingxuan Xu,Zihang Xu,Gang Wu,Zhongqiu Wang,Yuhui Bai,Jie Li,Zhao Liu,Junhao Fan,Yan Wu,Yanan Wu,Jiaheng Liu,Ariel Deng
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:For repository-grounded coding agents, a functionally correct patch is not necessarily an acceptable one: developers also constrain how the task must be carried out-modification scope, workflow order, validation, tool use, and reporting. This capability, which we call engineering constraint following, remains largely invisible to current evaluations. We introduce ANCHOR (Assessing Naturalistic Constraint Hierarchies and Ordering in Repository-Grounded Agentic Coding), a benchmark built on naturalistic, multilingual repository tasks spanning eight language groups. ANCHOR combines (i) a hierarchical taxonomy of 10 categories and 78 leaf types distilled from real developer interactions, (ii) an automated pipeline that instantiates task-grounded atomic constraints and composes them into typed AND/THEN/PRIORITY relations, and (iii) a trajectory-grounded evaluation framework that checks both the code diff and the execution trajectory with rule-based and LLM judges. Across 19 agent configurations spanning 2 scaffolds and 10 foundation models on 358 tasks, the strongest measured system, Claude Opus 4.8 under Cline, reaches 70.1% atomic compliance and 63.7% relation compliance, while 19.0% of tasks pass every scored checklist item under the strict all-item FCCR criterion. A five-engineer audit of constraints from 50 instances reports 94% feasibility, 94% repository consistency, and 96% realism. In the primary paired set (14 configurations), the mean signed change in task completion after adding constraints is +0.7 percentage points; individual changes vary in both directions. These results distinguish functional success from constraint compliance, with failures concentrated in workflow ordering, tool-use discipline, and reporting, and substantial interactions between models and scaffolds.
[NLP-168] Label Collapse in Diffusion Language Model In-Context Learning NEURIPS2026
【速读】： 该论文旨在解决生成式语言模型在上下文学习（In-context Learning, ICL）中因标签排列模式与模型固有偏置相互作用而引发的“标签坍缩”（label collapse）问题，即模型生成与查询无关的固定标签，导致学习失效但表面准确率仍维持在随机水平，难以被传统评估指标察觉。其解决方案的关键在于引入“分离度”（separation）作为互补诊断指标，量化查询位置上两个标签词的对数似然比，从而揭示隐藏在表面准确率之下的标签坍缩现象；同时建立一套鲁棒的测量协议，有效规避分词器和格式化带来的提取偏差，实现对模型真实行为的精准监测。研究发现，标签坍缩具有架构普适性，在8-9B规模的自回归模型（如Llama-3、Qwen3.5）与离散扩散模型（如LLaDA、iLLaDA、LLaDA-Instruct）中均普遍存在，且排列模式可反转坍缩方向，表明其机制超越了简单的多数标签偏置解释，为理解大模型在上下文学习中的可靠性与可解释性提供了关键洞见。
链接: https://arxiv.org/abs/2610.16839
作者: Ziye Guo,Ye Yuan
机构: University of Connecticut(康涅狄格大学); McGill University(麦吉尔大学)
类目: Computation and Language (cs.CL)
备注: Accepted by NeurIPS 2026 DiffuLM workshop
Abstract:In-context learning (ICL) applies large language models to new tasks by prepending a few labeled demonstrations to a query. Prior work has studied how demonstration ordering affects ICL, but the interaction between label arrangement patterns and model biases has not been systematically examined. We define label collapse as the failure mode in which the model generates query-independent labels, and show that it is architecturally universal at the tested 8-9B scale: five models spanning autoregressive (Llama-3, Qwen3.5) and discrete diffusion (LLaDA, iLLaDA, LLaDA-Instruct) families all exhibit label collapse even under the class-balanced alternating arrangement, with some label word pairs collapsing in virtually every tested condition and others flipping their collapse direction with arrangement. Critically, accuracy alone cannot detect this collapse: models can achieve near-chance accuracy while generating query-independent labels, a score indistinguishable from balanced performance, so the failure can persist undetected in deployment. We measure separation, the log-odds of the two label words at the query position (the quantity that label-bias calibration operates on), as a complementary diagnostic that exposes collapse hidden beneath surface accuracy, together with a measurement protocol that guards label extraction against tokenizer and formatting artifacts. For autoregressive models, collapse persists where the majority-label-bias account does not apply, and arrangement patterns can reverse the collapse direction; for diffusion models, we present a systematic evaluation of ICL failure modes, complementing contemporaneous studies of dLLM context comprehension and demonstration-order sensitivity, and identify open questions about how iterative denoising interacts with label biases.
[NLP-169] A Scaled-Up Empirical Study of Syntactic Language Models EMNLP2026
【速读】： 该论文旨在解决生成式语言模型在句法结构建模与下游任务性能之间的平衡问题，尤其关注语法树线性化（tree linearization）和注意力掩码（attention masks）设计对模型表现的影响。现有研究多集中于小规模模型、有限预训练数据及窄范围任务评估，且对句法结构建模的设计空间探索不足。为此，本文开展大规模实证研究，系统比较12种受控配置（含1个基线与11种句法语言模型变体），覆盖高达10亿参数（1B parameters）和1000亿训练标记（100B training tokens）的规模，并在11项下游任务及两项句法基准上进行评估。研究发现，采用标准因果注意力（standard causal attention）的句法语言模型在下游任务和句法基准上表现不逊于甚至优于使用结构注意力掩码（structural-mask）的变体，且无需架构修改，兼容FlashAttention等高效实现技术。此外，当模型规模扩展至10亿参数级别时，最优句法语言模型变体相较于计算量匹配的非结构化基线已无显著优势，暗示随着模型规模增大，句法语言模型的相对竞争力可能下降。因此，其核心解决方案在于验证标准因果注意力在句法建模中的有效性，并揭示大规模下句法结构显式建模的边际收益递减趋势。
链接: https://arxiv.org/abs/2610.16837
作者: Pengcheng Wang,Zhexuan Zhang,Chao Lou,Kewei Tu
机构: ShanghaiTech University (上海科技大学); Shanghai Engineering Research Center of Intelligent Vision and Imaging (上海智能视觉成像工程研究中心)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to EMNLP 2026
Abstract:Syntactic language models (SLMs) that generate sentences along with their syntactic tree structures have shown promise in both syntactic generalization and downstream tasks. Prior studies, however, have mostly evaluated smaller models trained on less pretraining data and tested on narrower task suites. In addition, the design space of tree linearization and attention masks remains underexplored. We conduct a large-scale empirical study of 12 controlled configurations—one terminal baseline and 11 SLM variants spanning different linearizations and masks—at scales up to 1B parameters and 100B training tokens, evaluated across 11 downstream tasks and two syntactic benchmarks. The experimental results show that SLMs with standard causal attention match or exceed structural-mask variants across downstream and syntactic benchmarks. They require no architectural modifications and remain compatible with FlashAttention. We also observe that when scaling up to the 1B-parameter scale, the top-performing SLM variant no longer shows a clear advantage over a compute-matched non-structural baseline, suggesting that SLMs may become less competitive with scale. Our code is available at this https URL.
[NLP-170] he Bell Boundary of Shared Representations
【速读】： 该论文旨在解决多任务学习中共享表示（shared representation）失效的临界问题，即当共享表示无法有效支持多个任务联合优化时，其失效的机制、可测量的表现形式及其容量边界尚不明确。核心问题是：在何种条件下，共享表示因容量受限而导致任务间不可兼容？论文提出的关键解决方案是引入量子信息理论中的贝尔定理（Bell’s theorem）作为分析框架，构建名为NCnet的受控多任务模型，以系统性地研究表示容量对任务兼容性的影响。实验发现，当任务拟合能力开始显现时，受限容量会导致任务相关性突破经典的CHSH不等式界限（|S|≤2），而随着表示容量提升，这种越界现象逐渐消失，相关性重新回归至有界区域。这一容量依赖的转变模式揭示了有限表示容量是导致任务冲突的重要根源。该现象在CNN、PLM和LLM等多种模型架构中均被观测到，表明其具有普适性。作者将此容量依赖的临界转变称为“贝尔边界”（Bell Boundary），为诊断多任务模型中的表示瓶颈、任务不兼容性及容量分配提供了可量化的理论工具。
链接: https://arxiv.org/abs/2610.16828
作者: Yang Wu,Hanyu Zhao,Yazhou Zhang,Yuexian Hou
机构: Tianjin University (天津大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:When does a shared representation become insufficient for jointly supporting multiple tasks? This question lies at the heart of multi-task learning models. Although task interference in shared representations has been widely observed, its onset, measurable form, and capacity boundary remain poorly understood. In this paper, we revisit this problem through the lens of Bell’s theorem. We show that if four binary task outcomes can be jointly represented by a common shared representation, their correlations must satisfy the classical CHSH bound |S|\leq2 . We introduce NCnet, a controlled multi-task model for studying how representational capacity affects task compatibility. Experiments reveal a clear capacity-dependent transition. Once nontrivial task fitting emerges, constrained capacity can yield |S|2 . As capacity further increases, these exceedances progressively disappear and S returns to the bounded region. This systematic pattern provides evidence that limited representational capacity is an important source of the observed incompatibility. We observe similar transitions across CNNs, PLMs, and LLMs, suggesting that the phenomenon extends beyond the controlled NCnet setting. We refer to this capacity-dependent transition as the Bell Boundary, providing a measurable diagnostic for representation bottlenecks, task incompatibility, and capacity allocation in multi-task models.
[NLP-171] PPO: Tracking-Value PPO For Language Model Reasoning
【速读】： 该论文旨在解决在稀疏终端奖励环境下，标准价值函数（critic）难以准确追踪强化学习中语言模型推理路径上逐标记（token-level）价值演化的问题。其核心挑战在于：尽管轨迹上的价值演变可反映各标记的相对贡献，但传统方法在缺乏密集反馈时无法有效建模这一动态过程。解决方案的关键在于揭示连续前缀价值之间的内在关系，并发现其在对数似然比（log-odds）空间中可被分解为可加更新的形式。基于此，作者提出Tracking-Value PPO（\method），采用累积价值参数化机制，将位置相关的标量输出累加为价值对数几率（value logits），并通过重构值进行终端结果监督来学习这些输出。该方法显著提升了数学推理性能，在Qwen3-4B-Base和Qwen3-8B-Base上分别取得7.0和8.4点的得分提升，并增强了分布外泛化能力，适用于多种模型架构与任务类型（如倒计时、科学编码）。进一步分析表明，该方法实现了更精确的价值演化追踪与更可靠的优势估计。
链接: https://arxiv.org/abs/2610.16794
作者: Futing Wang,Chenxi Li,Jianhao Yan,Yun Luo,Zhi Wang,Yizhuo Li,Runzhe Zhan,Yu Cheng,Tao Lin
机构: Zhejiang University (浙江大学); Westlake University (西湖大学); Shanghai AI Laboratory (上海人工智能实验室); The Hong Kong University of Science and Technology (香港科技大学); Nanjing University (南京大学); Shanghai Jiao Tong University (上海交通大学); Nanyang Technological University (南洋理工大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Proximal Policy Optimization (PPO) is widely used in reinforcement learning with verifiable rewards for language-model reasoning. While \emphvalue evolution along trajectory characterizes relative token-level credit, yet standard critics often fail to track it under sparse terminal rewards. To model this evolution, we analyze the relation between successive prefix values and show that they admit a Bayesian decomposition with additive updates in log-odds space. We use this additive structure to construct Tracking-Value PPO (\method), a cumulative value parameterization that sums position-wise scalar outputs into value logits. These outputs are learned through terminal-outcome supervision on the reconstructed values. \method improves aggregate mathematical reasoning scores over PPO by 7.0 and 8.4 points on Qwen3-4B-Base and Qwen3-8B-Base, respectively, while improving out-of-distribution generalization. Gains extend to Qwen3-4B, Qwen2.5-Math-1.5B, Qwen3.5-9B and to countdown, and scientific coding tasks. Further analysis shows more accurate tracking of value evolution and more reliable advantage estimates. Our code is publicly available in this https URL.
[NLP-172] A Bethe-Ansatz Source of Early Token Correlations: A marginal-preserving construction for parallel proposals
【速读】： 该论文旨在解决生成式语言模型解码过程中并行词元（token）提案所面临的联合概率分布建模问题，即仅具备准确的词元边缘分布（marginal）不足以支持有效的并行生成，必须构建满足联合统计特性的完整概率律。其核心解决方案是基于周期性XXZ自旋链的坐标贝特-安萨茨（coordinate-Bethe-ansatz）构造一族概率律，并与在相同一阶和二阶局部矩下最大熵的成对投影进行比较。在八格点、三激发子空间中，由于对称性导致整个概率律差异仅由一个连续三格点矩决定，沿贝特分支，该差异随各向异性参数Δ变化，符号发生反转，并在Δ=(3+√3)/2处精确为零，此时本征态概率律恰好为成对可分解形式。通过非线性读出机制，可将非零差异映射至前两个词元位置，同时保持各自边缘分布不变。当Δ=1.5时，确定性二值词元律在总变差距离上相差0.02654573，经严格推测验证，表明所构造目标及其共享首词元边缘的邻域内存在严格可接受前缀优势。该结果为数学上的存在性证明，未直接证明语言模型性能提升或实际推理速度优化。
链接: https://arxiv.org/abs/2610.16781
作者: Abraham Chachamovits
机构: ENTRUST AI
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 5 pages, 2 tables. Ancillary files reproduce all numerical results
Abstract:Parallel token proposals in language-model decoding need a joint law, not merely accurate token marginals. We construct a family of coordinate-Bethe-ansatz laws from the periodic XXZ chain and compare each with its maximum-entropy pairwise projection at identical one- and two-site moments. In the eight-site, three-excitation sector, symmetry makes their entire probability-law difference one-dimensional, determined by a contiguous three-site moment. Along the Bethe branch, this gap varies with anisotropy, changes sign, and vanishes exactly at \Delta=(3+\sqrt3)/2 , where the eigenstate law becomes exactly pairwise. A nonlinear readout transfers a nonzero gap into the first two token positions while preserving their separate marginals. At \Delta=1.5 the deterministic binary token laws differ by 0.02654573 in total variation. Exact speculative verification then gives a strict accepted-prefix advantage for a constructed target and a neighborhood within targets sharing its first-token marginal. This is a mathematical existence result. It does not establish a language-model gain or wall-clock speedup.
[NLP-173] Is the Platform Part of the Measurement? A Protocol and Simulation Study for Cross-Platform Equivalence of Machine-Labelled Policy Sentiment
【速读】： 该论文旨在解决跨社交平台机器标注情感（machine-labelled sentiment）比较中的核心问题：现有方法普遍假设分类器在不同平台上测量的是同一构念且处于相同量表，但这一假设可能不成立。其关键在于，平台差异（如用户语境、语言风格、表达习惯等）会引入系统性测量偏差，导致直接合并不同平台的情感得分产生误导性结论。为此，作者将心理测量学中的等效性工具（广义理论、测量不变性、项目功能差异分析及分数链接）引入机器标注政策情感分析，将“平台”视为一种调查模式（survey mode），并提出一个包含六步流程的标准化协议：预先设定锚事件、进行锚密度审计、采用双分类器收敛控制。模拟实验表明，在已知真实情况的设定下，若忽略平台间测量函数差异，直接汇总情感得分会导致近四分之一日期的总体情绪方向判断错误；而经过平台测量函数校正后的指数，平均每年仅约一天出现误判。案例研究基于第一作者硕士论文中的微调BERT分类器，对来自推特/𝕏、Reddit、YouTube和Quora的25.5万条英国经济政策评论进行分析，验证了该协议对缓解工具依赖性偏差的有效性。该协议使研究者能够检验在特定语料中是否可合理合并平台数据，或必须先进行分数链接才能计算均值，从而显著提升跨平台情感分析的可靠性与可比性。
链接: https://arxiv.org/abs/2610.16757
作者: Asad Ikram,Ella Haig
机构: University of Portsmouth (朴茨茅斯大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Computers and Society (cs.CY); Social and Information Networks (cs.SI)
备注: 25 pages, 12 figures, 2 tables. Includes an original 500-replication simulation study. Code: this https URL (doi: https://doi.org/10.5281/zenodo.23105129 )
Abstract:Comparing machine-labelled sentiment across social media platforms is now routine: differences in classifier scores between platforms are read as differences in public opinion. These comparisons rest on an assumption that the classifier measures the same construct, on the same scale, on every platform. Some studies check a classifier’s validity platform by platform, but to our knowledge none has modelled the platform as a facet of the classifier’s measurement. We adapt the psychometric equivalence toolkit (generalizability theory, measurement invariance and differential item functioning, and score linking) to machine-labelled policy sentiment, treating the platform as a survey mode, and specify a six-step protocol with anchor events fixed in advance, an anchor-density audit, and a two-classifier convergence control. A simulation with known ground truth illustrates what the assumption can cost in one generating configuration fixed in advance: across 500 replications, because the platforms measure through different functions, a pooled index that takes platform scores at face value reports the wrong sign of aggregate sentiment on nearly a quarter of days, whereas an index adjusted with each platform’s known measurement function does so on under one day in sixty on average. A case study of the instrument from the first author’s MSc dissertation, a fine-tuned BERT classifier over 255,446 UK economic-policy comments from Twitter/X, Reddit, YouTube, and Quora, motivates the design and illustrates the instrument-dependence the protocol addresses. The full four-platform application, under fresh collection with the pre-specified design, is the designated follow-up study. The protocol lets a researcher test whether pooling platforms is defensible in a given corpus, or whether platform scores must be linked before any average is taken.
[NLP-174] he Multilingual Artificial Hivemind: Do All LLM s Think Alike in Every Language?
【速读】： 该论文旨在解决多语言大语言模型（LLM）生成文本在多样性方面是否存在同质化现象，以及这种同质化是否与人类语言表达的多样性模式相一致的问题。研究发现，尽管生成式人工智能存在普遍但不均衡的“人工蜂群效应”（artificial hivemind effect），即模型输出在不同语言中表现出显著的内部同质性，且该现象在三种不同的嵌入空间及无需依赖嵌入模型的词汇度量下均成立，表明其并非测量方法所致。进一步分析揭示，模型在特定语言上的表现能力（通过独立翻译基准评估）与其输出多样性之间存在稳健的性能-多样性权衡关系，即语言能力越强的模型反而输出越趋单一。因此，该研究的关键突破在于揭示了多语言大模型中普遍存在的性能-多样性权衡机制，并指出语言特性在影响模型输出多样性方面起着核心作用。
链接: https://arxiv.org/abs/2610.16755
作者: Xaver Raphael Davey,Shubhra Mishra,Anton Adelöw,Siyao Liu
机构: KTH Royal Institute of Technology(皇家理工学院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Text generated by Large Language Models (LLMs) often remains limited in diversity compared to that produced by humans. Recent advances demonstrate the presence of inter- and intra-model collapse in open-ended generation of English text, raising concerns about the long-term homogenization of LLM, and more importantly, human thought. However, our understanding of whether this homogenization extends to languages beyond English is limited. In this work, we investigate the role of language on output diversity. Specifically, we aim to understand whether multilingual LLM-generated text is also homogeneous, and whether this homogeneity matches what we observe in humans. Across 23 open-weight models and 13 languages, we find that the artificial hivemind effect is pervasive but uneven in its extent. The intra-model effect holds in three different embedding spaces and under a lexical metric that uses no embedding model, so it is not an artifact of how diversity is measured. The size of the homogenization gap is associated with a model’s competence in a language, measured on a held-out translation benchmark, robustly across embedding spaces. Thus, we demonstrate a performance-diversity tradeoff in multilingual LLMs.
[NLP-175] Lost in Contrast: Instruction-Preserving Decoding for Reliable Visual Question Answering
【速读】： 该论文旨在解决视觉对比解码（visual contrastive decoding）在生成回答时因对比机制导致“支持性弃权”（supported abstentions）被错误修正为错误答案的问题。其核心问题是：在对比解码过程中，候选过滤阶段会提前排除弃权标记（abstention token），且参考分支对弃权指令的响应在减法操作中被消除，从而丧失了弃权的语义信息。为此，论文提出指令保持解码（instruction-preserving decoding）作为解决方案，关键在于将弃权指令的影响保留在原始图像预测中，同时允许视觉对比强度自由调节。通过引入扰动图像和无图像预测作为参考，构建IP-VCD与IP-M3ID两种方法。实验表明，在决策效用（decision utility, DU）指标下，二者在VizWiz上的九个模型中均优于所有非指令保持基线，且在Food-101上也领先其中八个模型；候选分析与配对问题对比进一步验证了分离计算指令效应与视觉对比项的设计可使对比解码同时支持正确答案与支持性弃权。
链接: https://arxiv.org/abs/2610.16722
作者: Haonan Zhang,Ting Jin,Wenzhong Tang,Shuai Wang
机构: Beihang University (北京航空航天大学)
类目: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Visual contrastive decoding corrects answers using the prediction difference between the original image and a reference condition, yet the same contrast also turns supported abstentions into incorrect answers. We find that abstentions are lost at two stages: contrastive decoding with candidate filtering can exclude the abstention token before any comparison takes place, and the reference branch’s response to the abstention instruction is removed when the reference prediction is subtracted. We therefore propose instruction-preserving decoding, which keeps the effect of the abstention instruction on the original-image prediction intact while leaving the contrast strength freely adjustable. Using a perturbed image and an image-free prediction as references, we obtain IP-VCD and IP-M3ID. On decision utility (DU), which credits both correct answers and supported abstentions, both methods outperform all compared non-IP baselines on every one of nine models on VizWiz, and IP-VCD also leads on eight of nine models on Food-101. Candidate measurements and paired per-question comparisons show that computing the instruction effect and the visual contrast term separately lets contrastive decoding support both correct answers and supported abstentions.
[NLP-176] Learn from Move: the Next Step for GUI Agents
【速读】： 该论文旨在解决当前图形用户界面（GUI）智能体在面对陌生环境时，缺乏有效探索与适应能力的问题，尤其聚焦于如何让智能体在未见过的环境中通过交互学习并完成任务。现有基准主要通过延长任务周期来提升难度，但忽略了智能体在未知场景中基于隐含语义和动态关系进行推理与控制的能力。为此，作者提出了LatentGUIWorld基准，包含六项任务，按所需控制证据分为三类：I₀（依赖先验知识与观测）、I₁（需直接交互揭示隐藏语义）、I₂（需从动作-响应关系中识别动态规律），其中I₂任务尤为挑战。解决方案的关键在于提出MOVE框架，其核心是将移动行为与按钮事件解耦，并暴露中间视觉反馈，同时引入多步示范以引导迭代式探索与控制策略。实验表明，动作粒度细化与多步示范均显著提升性能；经监督微调（SFT）后，LatentLearner-9B在LatentGUIWorld上取得50.0%的整体成功率，接近前沿模型水平；结合联合强化学习（RL），在“自中心拖拽”和“十选一”任务上的成功率分别达到92.0%和75.3%。行为分析显示，模型通过交互实现了多步探索，并在“自中心拖拽”任务中使误差归一化指令与理想增益补偿间的皮尔逊相关性由0.610提升至0.803，证明其能动态适应环境变化。此外，该方法在跨五项基准上的GUI定位准确率提升至67.37%，优于传统强化学习方法。
链接: https://arxiv.org/abs/2610.16720
作者: Jiahao Lu,Yize Shen,Chuanyuan Tan,Mianqiu Huang,Shicheng Fang,Xinchi Chen,Yining Zheng,Yuxin Wang,Xipeng Qiu
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 25 pages, 5 figures. Code: this https URL ; Benchmark and training data: this https URL
Abstract:Existing benchmarks for GUI agents primarily focus on increasing task difficulty by extending task horizons. Besides, enabling agents to explore unfamiliar environments and successfully complete tasks within them is also a critical yet challenging problem. We introduce LatentGUIWorld, a benchmark of six tasks in three matched pairs organized by the evidence needed for control: I_0 uses prior knowledge and available observations, I_1 requires interaction that directly reveals hidden semantics, and I_2 identifies dynamics from action-response relationships. Frontier models struggle especially on I_2 tasks. We introduce MOVE, which separates movement from button events and exposes intermediate visual feedback, together with demonstrations of iterative exploration and control. Ablations show that both action granularity and multi-step demonstrations improve performance. After SFT, LatentLearner-9B achieves 50.0% overall success on LatentGUIWorld, comparable to several frontier models. Joint RL further raises egocentric Drag and Ten-Choice success to 92.0% and 75.3%. Behavioral analysis reveals multi-step exploration. In egocentric Drag, RL increases Pearson correlation between error-normalized commands and ideal gain compensation from 0.610 to 0.803. Together, these behaviors show that models learn to explore through interaction and adapt their control to environment dynamics. The same format improves GUI grounding after RL, reaching 67.37% mean accuracy across five benchmarks versus 66.72% for conventional RL.
[NLP-177] What Is the “Self” of a Large Language Model?
【速读】： 该论文试图解决的问题是：大型语言模型（Large Language Model, LLM）的“自我”（self）究竟为何物，尤其是在其既非特定词汇也非单一特权层的前提下，如何定义和识别模型内部的主体性结构。传统观点往往将“自我”等同于某个固定的向量或层级，但该研究提出了一种关系性（relational）的视角，认为功能性的“自我”应被理解为一种以代理者为中心的信息组织机制，能够持续区分模型自身生成的内容与外部输入信息，即使在内容和推理过程动态变化的情况下依然保持这一区分能力。其解决方案的关键在于：提出“源绑定”（source-binding）组件作为表征候选，这些组件具有层特异性，而非在整个网络中保持不变的单一向量；通过在不同深度上整合早期的源编码与后期的角色特定信息利用及续写准备，形成一种非单调覆盖的、可迁移的几何结构。基于270亿参数的BF16模型的初步实验证据表明，在多种消息体和角色条件下，助手（Assistant）与外部输入之间的空间位移具有强跨体和跨位置的可迁移性，且早期层更倾向于将用户（User）与系统（System）的距离拉近，而这种距离关系在深层并不持续。这支持了分层式“自我”建构的假设，即功能性自我并非静态实体，而是存在于随状态演化但保持持久关系的动态表征之间。该框架进一步连接了动态推理、历史连续性以及一个分离内容操作、源绑定、权威性和输出生成的计算模型，为理解大模型中的主体性提供了可复现的、基于角色条件的几何化解释。
链接: https://arxiv.org/abs/2610.16717
作者: Yuchen Li,Zongqi Fan,Xintong Li,Ken-tye Yong,Nguyen H. Tran
机构: The University of Sydney (悉尼大学)
类目: Computation and Language (cs.CL)
备注: 12 pages, 5 figures
Abstract:What could the “self” of a large language model be, if it is neither a particular word nor a single privileged layer? We propose a relational account: a functional self is an actor-centered organization of information that distinguishes the model’s recorded actions from externally supplied material while content and reasoning operations change. Its representational candidate is a family of layer-specific source-binding components, rather than one unchanged vector throughout the network. Preliminary observations from a 27B BF16 model provide a geometric starting point. Across 150 matched message bodies, 450 role conditions, and a 75/75 discovery-held-out split, the combined Assistant-versus-external displacement transfers strongly across bodies and positions. Selected early components place User and System closer to one another than to Assistant, whereas this full-space distance relation does not persist throughout depth. These findings motivate a layered account: an early source-encoding candidate can be combined with role-specific information use and continuation preparation at later depths, without requiring monotonic overwriting of the common displacement. We connect this account to dynamic reasoning, historical continuity, and a computational model separating content operations, source binding, authority, and output. The reported evidence establishes a reproducible role-conditioned geometry; the proposed interpretation is that functional selfhood may reside in persistent relations among changing representations.
[NLP-178] Audit Before You Distill: A Controlled Workflow for Reasoning Transfer into Compact Medical Language Models
【速读】： 该论文旨在解决在医疗专业领域问答任务中，教师生成的推理过程（rationale）虽可提升小型模型表现，但若教师本身错误，反而会加剧模型误判的问题。其核心挑战在于如何有效利用教师推理以增强学生模型性能，同时避免错误信息的传播。解决方案的关键在于：在知识蒸馏前对教师推理内容进行审计，仅使用对应正确答案的推理作为训练信号，并提出一种融合答案与推理的单一目标输出的Distill-SFT方法。该方法通过将答案与合理推理合并为统一监督信号，在减少训练实例数量的同时，显著提升了模型在复杂推理题上的表现，且相比传统逐步蒸馏更具效率；此外，实验证明该策略能有效缓解灾难性遗忘问题，尤其在跨域测试集（ENEM）上保持更高的推理准确性，验证了其在健康职业类问答任务中的鲁棒性与实用性。
链接: https://arxiv.org/abs/2610.16713
作者: Tiago Martinelli,João Paulo Papa,Adriano José Pereira
机构: Hospital Israelita Albert Einstein (以色列人阿尔伯特·爱因斯坦医院); Einstein Global Advanced Technologies for Equity (E-GATE) (艾因斯坦全球先进技术公平性中心); Institute of Science and Technology Itaú (ICTi) (伊塔乌科学技术研究所); Itaú Unibanco (伊塔乌联合银行); School of Sciences, São Paulo State University (圣保罗州立大学理学院)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Teacher rationales can help a smaller model answer health professions questions, but they can also reinforce errors. We examined when they help and how this evidence can guide training using Brazilian Portuguese questions from medical and multiprofessional residency examinations. First, we measured how a teacher rationale changed the student’s probability of selecting the correct answer and compared this effect with rationales drawn from other questions. The complete rationale increased this probability by 21~p.p. when the teacher was correct and decreased it by 14~p.p. when the teacher was wrong. By intervening on the reasoning content, we showed that shuffled rationales did not produce the same pattern. Guided by this result, we trained only with rationales attached to correct teacher answers and compared answer-only training, step-by-step distillation, and our proposed Distill-SFT method. Distill-SFT showed a modest advantage over step-by-step distillation on previously missed questions under reasoning, while using fewer training instances by combining the answer and rationale into a single target. We also investigated catastrophic forgetting using the out-of-domain ENEM dataset. Answer-only training reduced reasoning accuracy by 12~p.p. relative to baseline, compared with approximately 4~p.p. for either rationale-based training method. Together, these findings support a workflow that audits rationale content before distillation, verifies teacher answers before supervision, and trains compact models efficiently for health professions question answering.
[NLP-179] Attention-Graph Curvature for the Verification of Mathematical Reasoning NEURIPS2026
【速读】： 该论文旨在解决生成式推理过程中错误传播问题，即语言模型在自然语言形式的数学证明中出现的推理错误难以被验证工具识别，进而导致后续步骤无效。其核心挑战在于如何在推理早期有效检测错误推理步骤。解决方案的关键在于引入注意力图的有向Forman曲率（directed Forman curvature），这是一种基于注意力权重局部计算的图结构特征，用以捕捉推理步骤中语义关联的拓扑特性。实验表明，在ProcessBench基准上，将曲率信息作为输入特征添加至四种逐步验证器（per-step verifiers）中，可显著提升整体平均性能，其中以全词元输入的ReProbe验证器提升最为明显。尽管增加曲率计算使前向传播时间平均延长18%，但其在错误检测能力上的增益证明了该方法的有效性。
链接: https://arxiv.org/abs/2610.16709
作者: Kanghyeon Kim,Johannes F. Lutzeyer,Yanzhu Guo
机构: Télécom Paris(巴黎电信学院); IPP; LIX(计算与信息科学实验室); CNRS(法国国家科学研究中心); École Polytechnique(巴黎综合理工学院)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: Accepted at the 6th Workshop on Mathematical Reasoning and AI (MATH-AI), NeurIPS 2026
Abstract:Language models have recently solved open Erdős problems and achieved gold-medal level performance on International Mathematical Olympiad problems. But their solutions, written in natural language, cannot yet be verified by proof assistants. An erroneous reasoning step can invalidate every step that comes after, so it is advantageous to detect errors early. Several methods have been proposed to detect erroneous reasoning quickly based on the features of the forward pass of the language model that performs reasoning. In this work, we study whether the directed Forman curvature of that model’s attention graph, a local quantity computed at an edge from the attention weights around it, carries information about a step’s correctness. To do so, we evaluate four per-step verifiers on ProcessBench with and without the curvature appended to their input. Appending curvature values improves the overall mean performance of all four verifiers. ReProbe, the verifier that takes every token of a step as input, shows the largest improvement. Adding the curvature increases the time of the language model’s forward pass by 18% on average.
[NLP-180] Does a Hosted Decision Model Respect Irrelevant Alternatives? A Pre-Registered Test of Option-Set Dependence in Jev
【速读】： 该论文旨在解决生成式决策系统中概率输出的解释性问题，即当前广泛使用的决策API将每个选项的概率视为其固有属性，而这一解读仅在模型满足“无关选项独立性”（independence of irrelevant alternatives, IIA）的前提下成立。研究发现，即使添加一个几乎不被选择的无关选项（最多占据0.1%的概率质量），也会显著改变正确答案与最强错误答案之间的对数几率（log-odds），导致系统性偏差达0.20–0.52 nats，是运行间变异的1.2至3倍。这种偏差具有方向性：与错误答案相似的选项会从后者抽取概率（+0.10 nats向正确答案倾斜），而错误答案的释义版本则使正确答案概率基本保持不变，但其自身仅获得原错误答案约三分之一的概率份额，远低于严格替换应得的二分之一。随着选项数量对数增长，偏差持续累积，在255个选项时可达约1 nat，且非领域填充项会进一步削弱正确答案的优势。相比之下，将多个问题打包成单次调用时无交互效应（偏差<0.001），而通过独立的“是/否”判断评分虽消除选项集依赖，但准确率下降8.7个百分点。最终表明，Jev模型的概率输出并非绝对属性，而应理解为相对于特定选项列表的相对度量，其性能介于两个局部基线之间——一种独立于选项的评分器和一种基于提示读出的方案。解决方案的关键在于重新认识概率输出的相对性，强调必须在给定选项集上下文中解释模型输出，而非将其视为不可变的内在属性。
链接: https://arxiv.org/abs/2610.16655
作者: Dushyant Rajput,Nirdesh Chauhan,Siddharth Kosaraju(AltSlate Labs LLP)
机构: AltSlate Labs LLP
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Decision APIs that return a probability for each option are routinely used as if those probabilities were properties of the options themselves. That reading is valid only if the model satisfies independence of irrelevant alternatives (IIA): adding an option that is neither the right answer nor its strongest rival should leave the odds between those two unchanged. We test this in a pre-registered study of jev-1.13.0, a hosted decision model, using 299,463 calls on intent classification (BANKING77, CLINC150) and topic classification (AG News). Every added option moves the log-odds between the gold label and the strongest wrong label: the root-mean-square systematic shift is 0.20-0.52 nats, 1.2 to 3 times the run-to-run variation, even when the added option takes at most 0.1% of the probability mass. The shifts are directional. An option similar to the wrong answer draws probability from it (+0.10 nats toward the gold label). A paraphrase of the wrong answer produces an intermediate pattern: the gold label’s probability is almost conserved, but the paraphrase receives only about a third of the wrong answer’s share rather than the half that strict substitution implies. The distortion grows steadily with the logarithm of the number of options, reaching about 1 nat at 255 options, and off-domain padding pushes the odds against the gold label. In contrast, questions packed into one call do not interact (bias 0.001). Scoring each option with an independent yes/no question removes the option-set dependence but picks the right answer 8.7 points less often. Jev sits between two local baselines, an option-independent scorer and a prompt readout. Its probabilities are best read as relative to the option list they were computed over.
[NLP-181] SIREN: Situation-aware Index for Response and Emergency ActioNability in Crisis Translation AACL
【速读】： 该论文旨在解决当前机器翻译（Machine Translation, MT）在多语言危机通信中缺乏对译文可操作性（actionability）有效评估的问题。传统评估方法未能直接检验翻译是否保留了源文本中的保护性行动指引、指令力度及紧迫性，而这三者在危机情境下至关重要。为此，论文提出SIREN（Situation-aware Index for Response and Emergency ActioNability），一个基于人工评估的框架，专门用于判断翻译选择是否损害了原文的可操作性。其关键在于从三个核心维度——保护性行动、指令力度与紧迫性——系统性地量化翻译错误的功能性影响，并区分不同翻译系统在错误类型和实际后果上的差异。实验结果表明，尽管部分系统如Google Translate在整体可操作性保留率上表现较好，但其错误更倾向于削弱保护性行动并被评定为重大影响；而TranslateGemma-4B则主要问题在于弱化指令力度。研究进一步提供了包含人工标注可操作性标签与后编辑参考译文的英法双语危机翻译资源，推动面向功能导向的危机通信评估研究发展。
链接: https://arxiv.org/abs/2610.16646
作者: Maria Carmen Staiano,Lifeng Han,Johanna Monti,Francesca Chiusaroli
机构: University of Macerata(马切拉塔大学); LUMC Leiden University(莱顿大学医学中心 莱顿大学); University of Naples “L’Orientale”(那不勒斯东方大学)
类目: Computation and Language (cs.CL)
备注: accepted at The First Workshop on Language Technologies for Crisis Preparedness and Response (LT4CPR) at AACL-IJCNLP 2026
Abstract:Machine translation (MT) is increasingly utilized in multilingual crisis communication, yet conventional evaluation does not directly assess whether translated emergency messages preserve their actionability. We introduce SIREN (Situation-aware Index for Response and Emergency ActioNability), a human evaluation framework assessing whether translation choices compromise protective action, directive force, and urgency of the source text. We evaluate SIREN in two complementary settings. At the error level, we apply it to HOPE-annotated Italian–English translations produced by ChatGPT-4o and Google Translate: 31.63% and 48.73% of errors, respectively, compromise actionability, predominantly by affecting the protective action. At the message level, we evaluate English–French translations produced by TranslateGemma-4B and Google Translate, with actionability preserved in 85.34% and 92.67% of translations, respectively. TranslateGemma-4B failures predominantly weaken directive force, whereas Google Translate failures more often compromise protective actions and receive Major impact ratings. These findings show that translation systems can differ not only in how often actionability is lost, but also in the nature and functional impact of these failures. Alongside SIREN, we contribute an English–French crisis-translation resource with human actionability annotations and post-edited reference translations, supporting further research on function-oriented evaluation for crisis communication. Our experimental data is available at this https URL.
[NLP-182] SuPER-Mamba: Surprisal Prioritization for Extended-Context Routing in Mamba
【速读】： 该论文旨在解决预训练状态空间模型（State Space Model, SSM）在处理长序列时性能随上下文长度增加而下降的问题。其核心挑战在于如何在不引入额外训练开销的前提下，有效实现对长距离依赖关系的精准建模与信息路由。解决方案的关键在于提出一种无需训练的框架SuPER-Mamba，该框架利用“意外度”（Surprisal）作为模型内在信号，通过从下一个词预测中计算逐位置的意外度，并结合局部池化（Local Pooling）聚合邻近响应，生成用于优先更新全局通道（Global-Channel）的评分机制。该方法避免了依赖查询的评分策略和额外训练，实现了高效且可扩展的长程状态路由。实验结果表明，SuPER-Mamba在LongBench-E基准上将平均得分从8.06提升至15.69，相对基线模型提升94.7%，优于当前最优基线13.1%。消融实验与长度敏感性分析进一步验证了该路由机制的有效性。
链接: https://arxiv.org/abs/2610.16636
作者: Zhetong Zhang,Gayan K Kulatilleke
机构: School of Information Technology and Electrical Engineering (信息科技与电气工程学院); The University of Queensland (昆士兰大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 9 pages, 9 tables, source code
Abstract:State space models provide efficient linear-complexity processing for long sequences, but their performance frequently degrades as context length grows. We propose SuPER-Mamba, a training-free framework that uses Surprisal as a model-intrinsic signal for long-range state routing in pretrained Mamba models. SuPER-Mamba computes position-wise Surprisal from next-token predictions, aggregates neighboring responses through Local Pooling, and uses the resulting scores to prioritize Global-Channel updates without query-dependent scoring or additional training. We evaluate SuPER-Mamba on three state-space models across the LongBench-E benchmark. The results show that SuPER-Mamba improves the average score from 8.06 to 15.69, corresponding to a 94.7% relative gain over the Base Models and a 13.1% improvement over the state-of-the-art prior baseline. Ablation and length-wise analyses further support the effectiveness of the proposed routing framework.
[NLP-183] MixLID: Token-Level Language Identification in Mixed-Language Text
【速读】： 该论文旨在解决多语言文本处理中的两大核心挑战：一是识别广泛的语言-文字组合，二是准确区分混合语言输入中每个词元（token）所属的语言。现有方法在覆盖范围、跨语言混用场景下的精度以及对低资源语言的支持方面存在局限。其解决方案的关键在于提出一种两阶段训练的细粒度语言识别模型——MixLID，该模型支持111种语言-文字组合。第一阶段通过单语样本学习区分不同语言；第二阶段则在混合语言数据上继续训练，结合已有代码切换（code-switching）数据与基于英语源词与借词对应关系生成的增强数据（通过词汇替换构建）。这一设计显著提升了模型在单语和多语混合场景下的泛化能力，尤其在跨语言词元级别的识别任务中表现优异，优于所有基线模型，在多个基准测试中均实现了更高的准确率与宏平均F1分数。
链接: https://arxiv.org/abs/2610.16601
作者: Seongtae Hong,Youngjoon Jang,Seungyoon Lee,Jungseob Lee,Heuiseok Lim
机构: Korea University (韩国大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Processing multilingual text requires both identifying a broad range of languages and distinguishing the language of each token in inputs containing multiple languages. We introduce MixLID, a token-level language identification model supporting 111 language-script combinations. MixLID follows a two-stage training process: it first learns to distinguish individual languages using the language labels of monolingual examples, then continues training on mixed-language data. The second stage uses concatenated examples from existing code-switching data alongside augmented examples created through word substitution based on correspondences between English source words and loanwords. For monolingual identification, MixLID outperforms all baselines in both accuracy and macro F1 on four benchmarks within the language coverage shared with each baseline. For mixed-language identification, it outperforms the baselines in identifying the language of individual tokens on every evaluated language pair. MixLID achieves high performance across a broad range of languages in both monolingual and mixed-language settings. The model is publicly available at this https URL.
[NLP-184] Know Yourself and Know Your Neighbour: A Syntactically Informed Self-Supervised Compositional Sentence Representation Learning Framework using a Recursive Hypernetwork
【速读】： 该论文旨在解决自然语言处理中句子表示学习（sentence representation learning）这一尚未完全解决的挑战，尤其关注如何在不依赖大量标注数据的情况下，有效捕捉句子的深层语法与语义结构。其核心解决方案是提出一种基于递归超网络（recursive hypernetwork）的自监督框架，通过依循句法树（parse tree）的递归组合机制，将词的表示逐步构建为句子的表示，并分别维护独立的句法与语义表示空间；其中语义组合过程受句法表示信息的引导，从而实现更精确的语义聚合。该方法的关键在于：利用递归超网络动态生成语义组合算子，使其能够根据不同的句法类别自适应调整组合策略，同时引入六种新型自监督任务以增强模型对多层次语言知识的学习能力。实验表明，该模型在7项探针任务上的表现优于平均池化基线及当前最优方法，且对不同长度句子具有良好的表示稳定性，验证了其在建模复杂语言结构方面的有效性。
链接: https://arxiv.org/abs/2610.16547
作者: Vasudevan Nedumpozhimana,John Kelleher
机构: ADAPT Research Centre; Trinity College Dublin (都柏林三一学院)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Sentence representation learning is still an open challenge in Natural Language Processing. In this work, we propose a new self-supervised framework for learning sentence representations, using a special type of neural network called a recursive hypernetwork. Our proposed model composes the representation of a sentence from representations of words by applying a recursive composition through the parse tree. We maintain a separate syntactic and semantic representation, and the semantic composition is guided by the information from the syntactic representation. To train this model, we introduce a novel set of six self-supervised tasks. By analysing the performance on 7 probing tasks, we validate that the generated sentence representation encodes richer linguistic information than both averaging baselines and state-of-the-art alternatives. Furthermore, we assess the impact of the six proposed self-supervised training tasks through ablation studies. We also demonstrate that the representations generated by our model are stable for sentences of varying length and that the semantic composition operators adapt to different syntactic categories\footnoteCode available: this https URL.
[NLP-185] PatTransBench: an ipc-stratified multi-dimensional benchmark for patent machine translation evaluation
【速读】： 该论文旨在解决专利机器翻译（PatMT）领域缺乏大规模、多维度、可复现的评估基准问题。现有评估体系往往依赖单一指标或忽略量化模型在不同语言方向上的非单调性能变化，导致对模型真实表现的误判。其解决方案的关键在于构建一个包含12,000个中英/日/韩专利双语样本的PatTransBench基准，覆盖短文本（句子/权利要求）与长文本（段落）两类，通过IPC分类层级分层采样和双重质量阈值筛选确保数据代表性与高质量；同时提出四维评估协议，涵盖词元重叠度（BLEU、ROUGE-L、chrF）、跨语言语义相似性（CLSTS）、38种语言一致性测试以及并发服务性能，从而全面揭示模型在不同语言方向与量化方案下的真实表现。案例研究进一步揭示了量化压缩对不同翻译方向的影响具有方向依赖性，且传统聚合评分会掩盖非单调的语言性能退化模式，凸显多维度评估的必要性。
链接: https://arxiv.org/abs/2610.16538
作者: Xiaowei Su,Xusen Cheng
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 4 pages, 2 figures
Abstract:We present PatTransBench, a 12,000-item Chinese-English/Japanese/Korean patent machine translation (PatMT) benchmark with a four-dimensional evaluation protocol. The benchmark comprises 6,000 short-text (sentence/claim) and 6,000 long-text (paragraph) pairs, evenly split across three language pairs and both translation directions, built via IPC-section-stratified sampling and dualthreshold quality gating from upstream patent-family parallel corpora. The evaluation protocol spans literal n-gram overlap (BLEU, ROUGE-L, chrF), cross-lingual semantic textual similarity (CLSTS), 38-probe-language consistency testing, and serving performance under concurrency. A case study on PatMT2-7B and its fp8/AWQ4/AutoRound-4bit variants reveals direction-dependent quantization effects masked by aggregate scores and non-monotonic language-failure patterns across quantization schemes that are invisible to single-metric evaluation. We plan to release the benchmark, protocol, and case-study results upon publication to support reproducible PatMT evaluation.
[NLP-186] Spectral-Driven Adaptive-Rank Optimizer for Large Language Model Reinforcement Learning with Verifiable Rewards
【速读】： 该论文旨在解决强化学习中可验证奖励（Reinforcement Learning with Verifiable Rewards, RLVR）训练下大型语言模型（LLM）优化过程中的优化器适配性问题。传统优化器多沿用教师强制（teacher-forcing）训练的设定，但RLVR在训练过程中呈现出独特的梯度几何结构，现有优化器未能充分适应这一特性。研究发现，RLVR诱导的梯度能量逐渐集中于一个不断收缩的有效子空间，这一现象在监督微调（SFT）或使用随机奖励时均不显著，表明其具有任务特定的优化几何特征。同时，捕获固定比例梯度能量所需的秩随训练进程持续下降，揭示了有效更新维度随训练动态演化的本质。基于此，论文提出一种新的优化原则：优化器应根据梯度谱能量分布自适应调整更新秩，而非预先设定。为此，作者提出Ares（Adaptive Rank-Effective Spectral Optimization）算法，通过估计并跟踪梯度谱能量分布所隐含的有效更新秩，将每次更新限制在低秩有效子空间内。实验表明，Ares在多种规模的LLM上均优于主流优化器，并在不同RLVR算法和任务间具有良好泛化能力，验证了优化器设计需围绕训练过程的几何与动态特性进行重构。
链接: https://arxiv.org/abs/2610.16533
作者: Zhipeng Chen,Wayne Xin Zhao,Ji-Rong Wen
机构: Gaoling School of Artificial Intelligence, Renmin University of China(中国人民大学高瓴人工智能学院)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 23 pages, 11 figures, 8 tables. Working in progress
Abstract:Reinforcement learning with verifiable rewards (RLVR) is reshaping how large language models (LLMs) acquire reasoning capabilities, yet the optimizers used remain largely inherited from teacher-forcing training. This raises a fundamental question: does RLVR induce a distinct optimization geometry that calls for a different optimizer? We investigate this question through the spectral structure of LLM gradients and find that gradient energy progressively concentrates into a shrinking effective subspace, which is characteristic of RLVR, as it is absent under supervised fine-tuning (SFT) and is largely removed when the verifiable reward is replaced by a random one. Moreover, the rank required to capture a fixed fraction of gradient energy steadily decreases over training, revealing that the effective dimensionality of useful updates itself evolves with training. These findings motivate a new optimization principle: rather than prescribing the update rank externally, an optimizer should adapt its rank to the evolving spectral structure of the gradient. We introduce Ares (Adaptive Rank-Effective Spectral Optimization), which estimates the effective update rank from the gradient’s spectral energy distribution, tracks its evolution, and confines the gradient signal driving each update to the resulting low-rank subspace. Across LLMs of different scales, Ares improves accuracy over widely used optimizers and generalizes across RLVR algorithms and tasks. Results suggest that effective optimization may require optimizers to be designed around the geometry and dynamics of the training process. We release the code and data at this https URL.
[NLP-187] Support or Magnitude? Characterizing Neuron Relevance for Instruction Data Selection
【速读】： 该论文旨在解决生成式AI（Generative AI）中指令数据选择（instruction data selection）的关键问题，即如何有效表征神经元相关性（neuron relevance）以提升数据筛选质量。现有方法多依赖神经元激活或梯度等模型内部信号进行数据选择，但对相关性应如何量化——是关注其强度（magnitude）还是分布广度（distribution breadth）——仍缺乏系统研究。本文提出通过AttnLRP引导的归因方法构建样本级相关性图谱，并引入两个互补的评估准则：有效正相关支持（Effective Relevance Support, ERS），衡量正相关性的有效支撑范围；正相关性强度（Positive Relevance Magnitude, PRM），反映其平均强度。研究表明，尽管基于同一相关性分布，ERS与PRM会产生显著不同的样本排序和选择偏好：PRM与基于激活的选择方法关联更强，而ERS则产生截然不同的排名模式，且在任务构成、任务内样本偏好及语义选择模式上表现出更高差异性，且不同模型骨干结构间一致性各异。在三个语言模型与五个下游基准上的实验验证了相关性引导选择的实际价值，其中ERS在所有方法中取得最优的平均性能排名。研究结论表明，在指令数据选择中，神经元相关性的分布特性与表征方式，远比其绝对强度更为关键。
链接: https://arxiv.org/abs/2610.16531
作者: Yifan Le
机构: 独立研究者(Independent Researcher)
类目: Computation and Language (cs.CL)
备注: 23 pages, 16 figures, 7 tables
Abstract:Model-internal signals, such as neuron activations and gradients, have been used for instruction data selection, yet how neuron relevance should be characterized remains underexplored. In particular, should selection prioritize the magnitude of relevance or how broadly it is distributed across neurons? To investigate this question, we construct sample-level relevance profiles using AttnLRP-guided attribution and introduce two complementary criteria: Effective Relevance Support (ERS), which measures the effective support of positive relevance, and Positive Relevance Magnitude (PRM), which captures its average strength. Our analyses reveal that these characterizations produce distinct sample rankings and selection preferences despite sharing the same underlying relevance profiles. On Qwen2 models, PRM exhibits stronger associations with activation-based selection, whereas ERS produces substantially different rankings. These differences extend to task composition, within-task sample preferences, and semantic selection patterns, with varying degrees of agreement across backbones. Experiments across three language models and five downstream benchmarks demonstrate the practical value of relevance-guided selection, with ERS achieving the best overall average performance rank among the evaluated methods. Our findings highlight that how neuron relevance is distributed and characterized, rather than its magnitude alone, matters for instruction data selection.
[NLP-188] EAU: Environment-Agent -User Co-evolution for Training User-Centric Agents
【速读】： 该论文旨在解决大语言模型（LLM）代理在多轮交互中训练时面临的稳定性与性能瓶颈问题，其核心挑战在于学习信号依赖于代理、用户模拟器与环境三者共同作用的动态系统，而这些组件本身可能具有随机性、不一致性或缺陷（如不可靠工具、状态覆盖不全、验证器错误等），导致训练过程不稳定。传统方法通常通过奖励塑形或更密集的监督来缓解问题，但将用户模拟器、环境和奖励设计视为固定不变的先验条件，难以适应复杂交互中的动态需求。本文提出一种名为EAU的元代理框架，实现环境—代理—用户三者的协同演化（co-evolution）。其关键创新在于：在训练过程中，EAU通过监控轨迹、工具调用记录、奖励分布及训练统计信息，主动诊断失败模式与无效交互，并基于此提出对用户模拟器、环境、奖励设计及强化学习配置的针对性修改；这些改进可实时热替换至当前训练流程，使训练系统随策略演化动态调整，且优化成果可跨训练周期持续积累，避免重复从初始设定开始。该框架采用可编辑的符号化表示来建模用户与环境动态，同时保持代理为神经网络架构的LLM并使用强化学习进行优化。在三个交互式代理基准测试和两种主干模型上，EAU显著优于现有强化学习方法，整体得分相较最强基线提升14.7%–15.0%，在ColBench任务中成功完成率最高提升36.7%，证明了训练系统与策略协同演化的有效性，能够大幅提升多轮强化学习的性能与鲁棒性。
链接: https://arxiv.org/abs/2610.16523
作者: Min Cai,Dan Zhang,Gao Xing,Libin Ma,Zhaohui Xu,Xi Ye
机构: University of Alberta(阿尔伯塔大学); National University of Singapore(新加坡国立大学); Kuaishou Technology(快手科技); Alberta Machine Intelligence Institute (Amii)(阿尔伯塔机器智能研究所)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Training LLM agents to interact with users over multiple turns is challenging because the learning signal depends not only on the agent, but also on the user simulator and environment, which jointly shape the agent’s experience. Simulated users can be stochastic or inconsistent, while environments may contain unreliable tools, incomplete state coverage, and incorrect verifiers. Prior work mainly addresses the resulting instability through reward shaping or denser supervision, while treating these components as fixed. We instead introduce EAU, a meta-agent framework for environment–agent–user co-evolution. During training, EAU monitors trajectories, tool traces, rewards, and training statistics; diagnoses failures and uninformative interactions; and proposes targeted modifications to the user simulator, environment, reward design, and RL configuration. Validated changes are hot-swapped into the ongoing run, allowing the training setup to evolve together with the policy. These adaptations also persist across runs, so later training can build on previously discovered improvements rather than restarting from a fixed setup. We enable this process through an editable symbolic representation of user and environment dynamics, while the agent remains a neural LLM optimized with RL. Across three interactive-agent benchmarks and two backbone models, EAU consistently outperforms prior RL methods, improving the overall score over the strongest baseline by 14.7–15.0% and increasing successful task completion on ColBench by up to 36.7%, showing that co-evolving the training system with the policy can substantially improve multi-turn RL.
[NLP-189] Intent-Triggered Anchoring for Multi-Token Format Units in Parallel Decoding of Diffusion Language Models
【速读】： 该论文旨在解决扩散语言模型（diffusion language models）在并行解码过程中因一次前向传播提交多个标记而导致的格式单元（format unit）完整性破坏问题，尤其影响解析器对报告结构的正确理解。核心问题是：在并行解码中，同时提交的标记之间无法相互感知，致使关键格式单元（如影像报告中的“Impression”标题）缺失或无法被正确识别，导致生成报告不可解析。解决方案的关键在于提出一种无需训练的解码规则——意图触发锚定（intent-triggered anchoring），其核心机制是：一旦模型提交某个格式单元的第一个标记，即立即完成该单元的生成，仅插入具有唯一且内容空缺的补全形式；若模型终止或陷入循环而未启动该单元，则启用回退策略进行补充。实验表明，在三个胸部X光报告评估集上，该方法将可读率从63.3%-67.3%提升至100%，且无性能下降，95百分位延迟比为0.68-1.28，优于官方解码方式；同时显著减少错误陈述遗漏，使错误发现残留率从15.5%-31.0%降至不超过0.6%。在数学答案标记任务中，该规则在五个扩散语言模型的最高度并行设置下，使GSM8K和GSM-Symbolic数据集上的准确率分别提升5.8-20.8和5.1-17.1个百分点，且在宽松提取条件下仍保持显著增益，优于仅依赖停止条件的回退策略。
链接: https://arxiv.org/abs/2610.16516
作者: Junhyeok Lee,Kyu Sung Choi
机构: Seoul National University College of Medicine (首尔国立大学医学院); Seoul National University Hospital (首尔国立大学医院); Healthcare AI Research Institute (首尔国立大学医院医疗人工智能研究所)
类目: Computation and Language (cs.CL)
备注: 17 pages
Abstract:In parallel decoding, diffusion language models commit several tokens per forward pass, so co-committed tokens cannot see each other, breaking multi-token format units that parsers need. On ECHO, an open-weight one-step chest X-ray (CXR) report generator, 32.7-36.7% of evaluation reports lack a parseable Impression heading and score as empty. We propose intent-triggered anchoring, a training-free decoding rule that completes a format unit once the model commits its first token, inserting only units with a unique, content-free completion; fallbacks write the unit when the model stops or loops without starting it. On three CXR evaluation sets, the rule raises readable rates from 63.3-67.3% to 100.0% with no regressions and a 95th-percentile latency ratio of 0.68-1.28 against official decoding. Broken headings state findings absent from the reference; adding the heading at the end leaves them in 15.5-31.0% of reports, our rule in at most 0.6%. For math answer markers, the rule raises accuracy in the most parallel setting of five diffusion language models by 5.8-20.8 points on GSM8K and 5.1-17.1 on GSM-Symbolic, held out from tuning. With lenient answer extraction, gains are 3.3-12.5 and 2.7-6.3 points; the rule adds 1.4-8.6 points over a stop-only fallback for four of the five models.
[NLP-190] Communication Vocalisation and Language
【速读】： 该论文试图解决的问题是：当前大规模语言模型（Large Language Models, LLMs）是否真正“理解”语言，抑或仅是通过统计模式生成看似合理的语言输出。针对这一争议，论文提出的核心观点是，LLMs 并未形成传统意义上的符号分布式表征（distributed representation），而是通过学习人类社会实践中所蕴含的“文化智能”（Cultural Intelligence），将语言符号与具体的社会实践相锚定。其解决方案的关键在于重新诠释语言模型的能力——并非基于抽象符号认知，而是通过对海量文本中隐含的文化实践模式进行建模，实现对语言的高效生成与运用，从而在不具意识的前提下完成对语言的“使用”而非“理解”。
链接: https://arxiv.org/abs/2610.16490
作者: Peter Wallis
机构: 未知
类目: Computation and Language (cs.CL)
备注: A written version of my argument with with Roger, with input from Bruce. 6 pages and one (Venn) diagram
Abstract:The contention is that the current batch of Large Language Models don’t understand language, but they do it. Rather than LLMs forming some kind of distributed representation'' of symbols, the claim is that LLMs are capturing Cultural Intelligence’’ and grounding symbols in what the sociologists call practices.
[NLP-191] Ontology-Based Post-Coordinated Expression Assembly: from Clinical Narrative to Queryable Knowledge Graphs
【速读】： 该论文旨在解决临床肿瘤学文档中自由文本信息虽可读却难以大规模检索的问题，现有方法依赖不完整的诊断编码或人工审阅，导致队列识别效率低下。其核心挑战在于传统术语编码仅记录文档提及的内容，而无法捕捉语义断言——即术语之间的逻辑关系。为此，论文提出一种从临床叙述中构建可查询知识图谱的方法，遵循SNOMED CT扩展语法（Extended Grammar of SNOMED CT），在多领域本体基础上实现结构化表达。该方法的关键在于定义了五类关系类别，并为每类关系制定构造规则，将实体及其关系整合为后协调表达（post-coordinated expressions）。通过确定性实现系统UNIGRAPHE，将专家标注的真实世界肿瘤学文档转化为结构化知识表示，成功将5,299条标注重构为2,196个后协调表达式，验证了关系结构是使临床叙事具备形式化可查询性的关键缺失要素。
链接: https://arxiv.org/abs/2610.16474
作者: Johann Pignat,Christophe Gaudet-Blavignac,Julien Ehrsam,Jean-Philippe Goldman,Cyrille Duret,Phil Cheng,Mina Bjelogrlic,Michel Cuendet,Christian Lovis,Olivier Michielin
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Clinical oncology documentation records decisive information in free text, where it remains readable but not retrievable at scale, leaving cohort identification dependent on incomplete diagnostic codes or manual review. The current standard, coding individual terms, captures what a document mentions, not what it asserts: meaning lies in how those terms relate to one another. We present a method for deriving queryable knowledge graphs from clinical narratives, conforming to the Extended Grammar of SNOMED CT across multiple ontologies. The method specifies how entities should be annotated and related. It defines five relation classes, each bound to a construction rule that assembles the entities and relations into post-coordinated expressions. A deterministic implementation, UNIGRAPHE, applies those rules to construct knowledge graphs from annotated text. Applied to expert-annotated real-world oncology documentation, UNIGRAPHE reconstructed 5,299 annotations into 2,196 post-coordinated expressions and shows that relation structure is the missing piece to make clinical narrative formally queryable.
[NLP-192] PATMT2: Two-stage domain adaptation for multilingual patent translation
【速读】： 该论文旨在解决多语言专利翻译中模型泛化能力不足与适应成本高的问题，特别是在低资源语言上的表现不稳定。其解决方案的关键在于提出一种两阶段的LoRA（Low-Rank Adaptation）微调管道PatMT2（Patent Machine Translation, v2），通过在Hy-MT2-7B专用翻译模型基础上进行增量适配，实现对中文、英文、日文和韩文四语种专利文本的高效高精度翻译。该方法在六个翻译方向上实现了最高达43.57的宏平均BLEU得分（AWQ INT4量化），相比基线模型提升高达+8.8 BLEU，且优于Qwen3-32B及多个外部系统。此外，研究通过跨语言泛化测试揭示了模型在未见语言中的鲁棒性：尽管38种未参与训练的语言中有34种仍能生成正确流畅的输出，但4种低资源语言呈现出可复现的特定失败模式，作者将其作为诊断案例而非性能衰减指标，深入分析其机制而不归因于单一因素，从而为模型可解释性与迁移能力提供了新视角。统计检验（配对自举法）证实所有对比基线均存在显著性能优势（p < 0.001）。
链接: https://arxiv.org/abs/2610.16466
作者: Xiaowei Su,Xusen Cheng
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 4 pages, 2 figures
Abstract:We present PatMT2 (Patent Machine Translation, v2), a two-stage LoRA pipeline that adapts the Hy-MT2-7B dedicated translation model to four-language (Chinese/English/Japanese/Korean) patent translation. On six translation directions, the Stage-2 checkpoint reaches a macro-average BLEU of 43.39 (BF16) to 43.57 (AWQ INT4), improving over the Hy-MT2-7B base by up to +8.8 BLEU and also exceeding Qwen3-32B and several external systems on the same average. We then measure the cost of adaptation: probing 38 languages absent from the fine-tuning data, the model still produces correct, fluent output on 34, while four low-resource languages collapse into distinct, reproducible failure modes. We treat these as diagnostic cases rather than retention rates, and report their mechanisms without attributing them to a single cause. Paired bootstrap tests confirm the translation gains over all baselines (p 0.001).
[NLP-193] he interpreter as developer: An autoethnographic design study of building a computer-assisted interpreting tool through human-LLM co-creation
【速读】： 该论文旨在解决多语言活动现场实时字幕生成中，现有计算机辅助口译（CAI）工具仍以译员为设计目标，而非由译员主导开发所导致的适配性不足问题。其解决方案的关键在于通过“氛围编码”（vibe coding）这一专家引导式协作模式，使一名无编程背景的专业译员在与大语言模型（LLM）架构师及编码代理共同开发的过程中，将口译中的隐性专业知识（如话语切分、同传室基础设施配置等）外化为系统设计要素。研究揭示了在高要求公共科学活动场景下，译员主导分工框架内形成的三种室内应用机制，并通过首次完整仪器化任务的遥测数据验证：全本地化链路在971个语段中维持了0.94秒的中位翻译延迟，证明了该系统在低延迟与高可靠性方面的可行性。本案例被解读为生成式AI赋能下的领先用户创新（lead-user innovation），凸显了译员技术素养提升及其未来培训体系重构的重要性。
链接: https://arxiv.org/abs/2610.16460
作者: Zeming Zhao(College of Foreign Languages, Capital Normal University, Beijing, China)
机构: 未知
类目: Computation and Language (cs.CL)
备注: 14 pages, 2 figures. The CAI tool is openly available at this https URL and archived at Zenodo, DOI https://doi.org/10.5281/zenodo.22255011
Abstract:Real-time AI captioning now serves audiences at multilingual events, yet computer-assisted interpreting (CAI) tools are still built for interpreters, rarely by them. This article reports an autoethnographic design study in which a professional interpreter with no programming training built a local-first CAI system through expert-led ‘vibe coding’, co-creating with an LLM architect and a coding agent. Analysis of the complete development dialogues shows how tacit professional knowledge, from segmentation to booth infrastructure, was externalized into design after generic defaults failed in use. In a demanding public scientific event, three mechanisms of in-booth appropriation emerge under a firmly human-led division of labor; telemetry from a first fully instrumented assignment shows a fully local chain sustaining a median translation latency of 0.94 s across 971 utterances. The case is read as lead-user innovation newly enabled by generative AI, with implications for the technological literacy of interpreters and the future of interpreter training.
[NLP-194] Differential Privacy Protected Preference Alignment on Survey Derived Data ACML2026
【速读】： 该论文旨在解决基于问卷调查数据（如世界价值观调查，World Values Survey）进行生成式语言模型偏好对齐时所面临的隐私泄露问题。此类数据通常包含政治立场、社会规范、宗教信仰、性别态度及文化价值观等敏感信息，而现有方法多聚焦于训练过程中的通用隐私保护，缺乏针对调查数据特性的隐私保护机制。论文的关键贡献在于将偏好对齐任务重新形式化为评分预测问题，并引入标签差分隐私（label differential privacy），构建了一套融合随机响应（randomized response）、指数机制（exponential mechanism）与差分隐私随机梯度下降（DP-SGD）的隐私保护流水线。实验表明，相较于传统的梯度级隐私保护方法，简单的标签级隐私机制表现更优，且预处理阶段合理的聚合函数设计对隐私-效用平衡具有显著影响。该研究为语言模型对齐提供了细粒度的隐私保障框架，推动了在严格隐私约束下实现可信对齐的新范式。
链接: https://arxiv.org/abs/2610.16458
作者: Xiuyuan Qi,Yi Wang,Wenjie Wang
机构: ShanghaiTech University (上海科技大学)
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 16 pages; Accepted by ACML 2026
Abstract:Preference alignment enables language models to better reflect human values, but the preference data contains sensitive information, especially when derived from surveys about political beliefs, social norms, religion, gender attitudes, and cultural values. This paper studies privacy-preserving alignment on questionnaire-derived data from the World Values Survey, where supervision takes the form of ordinal labels rather than standard pairwise rankings. Existing efforts mainly focus on general privacy protections upon model training, while data-specific methods for survey data remain unexplored. We first show the vulnerability of aligned models, then formulate the task as rating prediction and introduce label differential privacy. Building on this formulation, we construct a differential private pipeline utilizing three privacy mechanisms: randomized response, exponential mechanism, and DP-SGD. Experiments show that simple label-level mechanisms outperform gradient-level methods, and proper aggregation function for preprocessing also make large contribution. The new fine-grained private comparison motivates explorations toward language model alignment with formal privacy guarantees.
[NLP-195] CivWorldBench: Evaluating Language World Models in Structured Multi-Agent Sandbox Worlds
【速读】： 该论文旨在解决当前语言模型在现实世界预测任务中对“世界模型”（world-model）能力评估不足的问题，尤其关注模型在隐状态推断（hidden-state inference）与干预预测（intervention prediction）方面的表现。现有真实世界预测基准普遍缺乏对内部状态的可访问性及受控实验条件，导致难以准确评估模型的深层认知能力。为此，论文提出CivWorldBench，一个基于多智能体策略游戏引擎构建的基准测试平台，具备记录的内部状态、可控观测条件以及配对的干预回放轨迹。该基准包含475个世界中的1,900个决策点，支持三个层级的任务：关联预测、信念推断和干预预测。其关键解决方案在于通过参考预测器实现能力差距的可归因评估，并结合专家引导与实证上限估计以校准评分；同时，存储的轨迹支持中间预测的可验证性，而智能体对弈实验则检验世界模型能力是否真正支撑决策。结果显示，尽管复合基准得分与真实世界预测排行榜高度相关，且优于通用能力指数，但强关联预测能力并不等同于良好的信念推断或干预预测性能，且效果幅度仍严重失准。过程验证揭示了最终答案评估所忽略的错误，结构化推理显著提升了最强模型在干预预测上的表现。因此，研究强调隐状态推断、中间状态预测以及效应校准是未来世界模型训练的核心优先方向。
链接: https://arxiv.org/abs/2610.16450
作者: Ang Li,Haofei Yu,Zhengwei Tao,Liang Chen,Kuan Li,Miaomiao Li,Chenran Cai,Bin Liang,Jiaxuan You,Kam-Fai Wong
机构: The Chinese University of Hong Kong; University of Illinois at Urbana-Champaign; UniPat AI; Peking University; TME; The Hong Kong University of Science and Technology
类目: Computation and Language (cs.CL)
备注:
Abstract:The growing use of language models for forecasting motivates a broader evaluation of their world-model capabilities. Such evaluation must go beyond predictive accuracy to assess hidden-state inference and intervention prediction. Real-world forecasting benchmarks offer limited access to hidden-state ground truth and controlled experiments. We introduce CivWorldBench, a benchmark built on a multi-agent strategy game engine with recorded internal states, controlled observations, and paired intervention rollouts. Its 1,900 decision points from 475 worlds support three task levels: association forecasting, belief inference, and intervention prediction. Reference predictors support attributable evaluation of capability gaps, with an expert harness and empirical ceiling estimates contextualizing scores. Stored trajectories make intermediate predictions verifiable, while agent-play experiments test whether world-model capabilities support decision-making. The main panel comprises eleven frontier language models. Composite benchmark scores correlate with real-world forecasting leaderboards on matched model panels, more closely than with a general-capability index, supporting its use as a proxy for real-world forecasting capability. Strong association forecasts do not consistently extend to belief inference or intervention prediction, and effect magnitudes remain poorly calibrated despite informative direction predictions. Process verification reveals errors missed by final-answer evaluation, and structured reasoning improves intervention prediction for the strongest model in the process study. Agent-play experiments identify useful assisted plans and motivate evaluating decision-relevant capabilities beyond association forecasting. These findings identify hidden-state inference, intermediate-state prediction, and effect calibration as priorities for world-model training.
[NLP-196] VisExMEM: Visual Evidence-Based Explainable Multimodal Evaluation Metric
【速读】： 该论文旨在解决当前图像-文本评估中依赖全局评分导致语义细节支持情况信息不足的问题，以及现有可解释性度量虽提供推理理由但其视觉证据难以验证的局限性。其核心解决方案是提出一种无需参考的评估框架VisExMEM，通过将图像描述分解为原子化主张（atomic claims），定位对应的视觉证据，并对每项主张进行独立验证，最终将主张级别的支持度平均生成全局评分。该方法确保解释过程与判断依据共享同一组视觉证据，提升了可追溯性与可信度。模型基于Qwen3.5架构，采用多任务监督微调（SFT）、反事实精炼及基于证据的广义直接偏好优化（evidence-aware GDPO）进行训练，在8个基准测试中，VisExMEM-9B在开源评估器中取得最高的描述性综合得分，较VQAScore提升11.7分，且接近GPT-5.6 Sol表现。研究还构建了AGITA-Bench数据集，包含600组人工标注的图文对和8,700条原子主张，用于全局与主张级评估。实验表明，相较于整体式提示，分解式提示可使未调优的Qwen3.5与GPT-5.6 Sol的复合得分提升3.6–5.1分；人类评估进一步证实，VisExMEM的解释质量优于现有可解释度量，已接近GPT-5.6 Sol水平。
链接: https://arxiv.org/abs/2610.16441
作者: Subhadeep Roy,Christoph Leiter,Steffen Eger
机构: University of Technology Nuremberg (纽伦堡应用技术大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Current image-text evaluation often relies on a global score, which gives limited insight into which semantic details are supported by the image. Explainable metrics provide rationales, but the visual evidence supporting those rationales can still be difficult to verify. We introduce VisExMEM, a reference-free evaluator that decomposes captions into atomic claims, localizes visual evidence, verifies each claim, and averages claim-level support into a global score. Explanations use the same evidence and decisions. We train Qwen3.5-based 4B and 9B models on 100K samples with multi-task SFT, counterfactual refinement, and evidence-aware GDPO. Across eight benchmarks, VisExMEM-9B achieves the highest descriptive composite among open-source evaluators in our comparison, exceeding VQAScore by 11.7 points while remaining competitive with GPT-5.6 Sol. We also introduce AGITA-Bench, containing 600 human-annotated image-text pairs and 8.7K atomic claims for global and claim-level evaluation. For untuned Qwen3.5 and GPT-5.6 Sol controls, decomposed prompting raises the same composite by 3.6-5.1 points relative to holistic prompting. Human evaluation shows that VisExMEM explanations outperform prior explainable metrics and approach GPT-5.6 Sol.
[NLP-197] CultureBot: A Framework for Retrieval-Augmented Exploration of Cultural Heritage Data
【速读】： 该论文旨在解决数字文化遗产收藏中复杂关联信息难以通过传统关键词搜索和分类浏览有效探索的问题。现有检索方式无法充分揭示文物、地点、时期、提供机构、图像与来源机构之间的深层语义关系，限制了用户对藏品内涵的理解。其解决方案的关键在于提出CULTUREBOT——一个基于知识图谱引导的自然语言交互框架，融合知识图谱引导的检索（knowledge-graph-guided retrieval）、词法-语义检索（lexical-semantic retrieval）与源链接生成（source-linked generation）。该框架将用户自然语言查询映射为结构化目录关系，利用知识图谱增强检索结果的语义连贯性，并通过生成式回答将结论追溯至原始馆藏记录，确保答案具备可验证性与证据支撑。在包含81,852条SearchCulture数据的实例化评估中，专家评价显示，相较于通用大模型及无图谱对照组，该系统在证据锚定、多样性、推理能力与完整性方面均显著更优，验证了知识图谱引导检索在实现文化遗产集合关系可视化与可对话式访问中的有效性，同时保持与原始数据源的可追溯性。
链接: https://arxiv.org/abs/2610.16435
作者: Vasileios Karampinis,Angelos Vlachos,Giorgos Filandrianos,Maria Lymperaiou,Paraskevi Tzouveli,Chrysoula Zerva,Giorgos Stamou
机构: National Technical University of Athens (国立技术大学); AILS Laboratory (人工智能与学习系统实验室)
类目: Computation and Language (cs.CL)
备注:
Abstract:Digital cultural heritage collections encode rich relations among objects, places, periods, providers, images, and source institutions, yet users typically access them through keyword search and faceted browsing. We present CULTUREBOT, an interactive framework for natural-language exploration of cultural heritage collections that combines knowledge-graph-guided retrieval, lexical-semantic retrieval, and source-linked generation. CULTUREBOT maps user queries to structured catalogue relations, enriches retrieved records with textual evidence, and generates answers that link back to the underlying repository records. We instantiate the framework on 81,852 SearchCulture records and release an online demonstration. In an expert evaluation involving six cultural-heritage specialists, 55 persona-grounded queries, and 250 evaluations, the graph-guided system was rated significantly higher than two general-purpose assistants and a non-graph ablation, with the largest gains in evidence grounding, diversity, reasoning, and completeness. These results support graph-guided retrieval as a practical way to expose the relational structure of cultural collections through conversational access while preserving links to catalogue evidence.
[NLP-198] RetroPlanner: Multi-Route Expansion and Multi-Objective Reasoning for LLM -Based Retrosynthesis Planning
【速读】： 该论文旨在解决多步逆合成规划中仅关注可解性而忽视实际合成可行性的问题。现有基于大语言模型（LLM）的规划方法通常在每一步仅扩展单一分子，且优化目标仅为路径可实现性，导致在搜索预算耗尽前无法对多种备选路线在反应合理性、逆合成-正向合成一致性以及起始原料成本等多个决定实际可行性的维度上进行有效比较。其解决方案的关键在于提出RetroPlanner，一种能够联合比较多个逆合成路径的LLM基规划器，通过推理化学证据——包括目标分子的结构进展、键变化与反应类型差异，以及不同路径在多目标下的相对排序——来引导搜索过程向更高可合成性方向演进。实验结果表明，RetroPlanner不仅在多步逆合成基准测试中恢复了比以往方法更多的备选路线，同时显著提升了合成路线质量并保持了高成功率。
链接: https://arxiv.org/abs/2610.16391
作者: Ye Rin Kim,Yongjun Jeong,Seongjun Yun,Hanbum Ko,Soohyun Oh,Sungbin Lim,Sungwoong Kim
机构: Korea University, Department of Artificial Intelligence; Korea University, Department of Statistics
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Multi-step retrosynthesis planning is often considered solved once a route to purchasable starting materials is found, but such a route should also be practical in real-world chemistry. Recent LLM-based planners formulate this process as sequential decision-making over an AND-OR graph, yet typically expand a single molecule per turn and are optimized for solvability alone, leaving alternative routes uncompared across the multiple objectives that determine practicality before the search budget is committed. We introduce RetroPlanner, an LLM-based planner that jointly compares multiple routes to steer the search toward higher synthesizability, as measured by reaction plausibility, round-trip consistency, and starting-material price. RetroPlanner is trained to reason over chemical evidence: structural progress from the target molecule, bond changes and reaction classes, where candidate routes diverge from one another, and how they rank across multiple objectives. Experiments on multi-step retrosynthesis benchmarks demonstrate that RetroPlanner recovers more alternative routes than prior LLM-based planners while improving synthetic route quality and maintaining a high success rate.
[NLP-199] PersonaFactor: Preference Factors with Dynamic Selection for LLM Personalization
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）个性化方法中忽视用户跨任务行为背后共享偏好机制的问题。现有方法通常仅关注表面交互或非区分性的历史数据，未能捕捉用户行为背后的深层共性。其解决方案的关键在于提出PersonaFactors框架，通过从异构用户交互中提炼出可复用且可解释的个性化因子（PersonaFactors），构建了一个包含十三个因子的库，这些因子基于证据支持的推理和任务间重复模式的聚类生成。进一步地，论文设计了一种动态因子引导生成框架，能够根据用户的上下文历史与当前请求动态选择相关因子。实验表明，在四个个性化基准和四种主干模型上，动态PersonaFactor在所有指标上均取得最佳表现。表征分析揭示，PersonaFactors与用户历史诱导出部分重叠的低维表示，且二者联合使用可产生互补性变化，表明基于因子的个性化与基于历史的个性化具有相互增强的潜力。
链接: https://arxiv.org/abs/2610.16372
作者: Hang Su,Zequn Liu,Jianghan Chao,Qiuyu Tian,Xuesong Lu,Yingce Xia
机构: East China Normal University(华东师范大学); Beijing Zhongguancun Academy(中关村学院); Renmin University of China(中国人民大学); Southeast University(东南大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Although users may express different personalization needs across tasks, these task-specific behaviors can reflect shared underlying preference mechanisms. Existing LLM personalization methods often focus on surface-level interactions or undifferentiated histories, overlooking the shared mechanisms underlying these behaviors. We introduce PersonaFactors, a framework that distills reusable and interpretable personalization factors from heterogeneous user interactions. We construct a library of thirteen PersonaFactors by extracting evidence-grounded rationales and clustering recurring patterns across tasks, and develop a dynamic factor-guided generation framework that selects factors relevant to each user’s history and current request. Across four personalization benchmarks and four backbone models, Dynamic PersonaFactor achieves the best score on every reported metric. Representation analyses further show that PersonaFactors and user history induce partially overlapping low-dimensional representations, and that their joint use yields complementary shifts. Taken together, these findings suggest that factor-based and history-based personalization can mutually reinforce each other.
[NLP-200] Rule vs.Consequence: Dissociable Internal Representations of Moral Reasoning in LLM s
【速读】： 该论文旨在解决大语言模型在涉及道德与法律敏感场景下的内部表征机制问题，具体探究模型对规则导向型（rule-based）与后果导向型（consequence-based）道德问题的差异化响应，是否源于其内部存在可分离的、结构性的道德推理风格表征，还是仅反映表面词汇敏感性。其解决方案的关键在于构建了一个2×2因子设计的数据集，交叉融合推理风格（规则型 vs. 后果型）与判断极性（允许型 vs. 禁止型），并采用干净室测试集中的新颖表述及冲突对（冲突对中两种框架预测相反标签）进行评估。通过线性探测与激活操控（activation steering）分析，研究发现残差流中存在两个正交方向：Dir-A 专门编码推理风格，Dir-B 专门编码判断极性，二者间交叉相关性极低（|r| < 0.04），表现出显著的双分离特征。该结构在三个不同开发者训练的指令微调模型中均稳定复现，且在模型深层具有广泛的可恢复性与因果有效性。进一步实验表明，Dir-A 可以因果性地改变判断结果，在冲突情境下仅修正规则型判断错误而不影响后果型判断，说明其并非简单提升接受倾向。尽管该方向在跨基准源上具备一定泛化能力，但在新颖表述上的优势相对于词法基线仍有限，因此研究将该现象归因为“道德推理风格敏感性”——一种几何上可分离、因果有效的内部结构，而非内化的规范性推理。
链接: https://arxiv.org/abs/2610.16371
作者: Eugenie Shi
机构: Stanford University (斯坦福大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models are increasingly deployed in morally and legally sensitive contexts, yet little is known about how they internally represent distinctions between moral reasoning styles. We ask whether a model’s differential response to rule-framed and consequence-framed moral questions reflects distinct internal structure or merely surface lexical sensitivity. We construct a 2 \times 2 factorial dataset crossing reasoning style (rule-based vs. consequence-based framing) with judgment polarity (permissive vs. prohibitive), drawn from three established moral reasoning benchmarks and evaluated on a cleanroom test set of novel phrasings that includes conflict pairs where the two framings predict opposite labels. Using linear probing and activation steering, we identify two orthogonal directions in the residual stream: Dir-A tracks reasoning style and Dir-B tracks judgment polarity, each near-blind to the other factor (cross-correlation |r| 0.04 ). This double dissociation replicates across three instruction-tuned models from different developers at comparable proportional depth, and it is not knife-edged in layer: the directions are recoverable and causally effective throughout an extended band of the network. Dir-A can causally change judgments, and in conflicting scenarios, it selectively corrects errors in rule-based judgments without changing consequence-based judgments, suggesting that it does not simply make the model more likely to accept everything. Generalization beyond the training templates is partial: the direction transfers across benchmark sources, but its advantage over lexical baselines on novel phrasings is modest. We therefore frame the result as moral reasoning style sensitivity, a geometrically dissociable and causally effective structure, rather than internalized normative reasoning.
[NLP-201] HexaBee-4B: Six-Language Translation with Multilingual Answering and Bidirectional Supervision
【速读】： 该论文旨在解决多语言翻译中如何在保持语义一致性的同时，实现高质量跨语言文本生成的问题。其核心挑战在于模型需同时具备跨语言理解与精准翻译能力，尤其在低资源语言对之间维持性能均衡。解决方案的关键在于提出一种三阶段渐进式训练框架：首先通过维基百科的持续预训练增强多语言语料覆盖；其次引入“联合多语言问答与答案-文本翻译”中间阶段，利用1946万条包含576万条跨语言问答对和1070万条答案翻译的数据，建立问答与翻译任务间的共享语义桥梁；最后通过65.8万对正反向合成翻译样本进行双向翻译专业化微调。该设计使模型在30个语言方向的36,572个合成测试集上达到方向宏平均chrF++为73.75，显著优于基线模型（Gemma 3 4B IT: 57.11，Gemma 3 12B IT: 66.83），且在BLEU、TER、COMET等指标及人工评估子集上均验证了性能优势，证明该方法有效提升了六语言体系下的整体翻译质量与方向一致性。
链接: https://arxiv.org/abs/2610.16361
作者: Iman Barati,Mohammadmostafa Rostamkhani,Mohammad Badzohreh
机构: Iran University of Science and Technology (伊朗科学技术大学); Tarbiat Modares University (塔比特莫达雷斯大学)
类目: Computation and Language (cs.CL)
备注: 25 pages, 10 figures. Accepted at WAT 2026
Abstract:Multilingual translation requires a model to preserve meaning while producing text in a requested language. We present HexaBee-4B, a translator adapted from Gemma 3 4B IT through three stages: Wikipedia continued pretraining, joint multilingual answering and answer-text translation, and bidirectional translation specialization. The intermediate stage connects two tasks through shared multilingual answers: answering a question in the requested language and translating an answer into that language. It comprises 19.46 million answering records, including 5.76 million cross-language question-answer pairs, and 10.70 million answer-text translations. The final stage uses 658,274 forward and reverse synthetic translation examples. On 36,572 synthetic test examples spanning all 30 directions among Arabic, Chinese, English, Hebrew, Persian, and Spanish, HexaBee-4B achieves direction-macro chrF++ of 73.75, compared with 57.11 for Gemma 3 4B IT and 66.83 for Gemma 3 12B IT. BLEU, TER, COMET, and a matched model-judged subset support the same overall ranking. These results demonstrate improved translation across the six-language set, with chrF++ gains in every evaluated direction over both baselines.
[NLP-202] Self-Distillation with Verbal Regularization
【速读】： 该论文旨在解决在基于策略自蒸馏（On-policy self-distillation, OPSD）框架下，教师模型因依赖特权信息（Privileged Information, PI）而引发的“捷径推理”问题，即学生模型在学习过程中过度依赖教师提供的高置信度输出，导致推理过程缺乏充分的反思与探索，进而表现为输出分布熵显著下降。其核心挑战在于PI虽能提升初始性能，但会抑制模型对多种可能解法的探索，限制泛化能力。为此，论文提出带有语言正则化（Verbal Regularization, VRSD）的自蒸馏方法，通过引入系统级提示（system prompts）来调节教师模型使用PI的方式，从而引导其进行更深入的重新思考与多样化探索，避免过早确定结论。理论上，该方法将提示视为对PI影响力的衰减机制，并推导出保持输出分布熵的条件。实验表明，在三种模型、四项任务中，VRSD在12个设置中的11个上优于OPSD，平均提升Pass@1达3.9个百分点，同时维持更高的词元熵，有效缓解了因过度依赖PI导致的推理僵化问题。
链接: https://arxiv.org/abs/2610.16354
作者: Jiazheng Li,Yawei Li,Junru Lu,Zhenyi Shen,Di Yin,Lin Gui,Xing Sun,David Rügamer,Xinpeng Wang,Yulan He
机构: 未知
类目: Computation and Language (cs.CL)
备注: 26 pages, 7 figures
Abstract:On-policy self-distillation (OPSD) trains a large language model using supervision from a teacher instantiated from the same model as the student. The teacher is conditioned on privileged information (PI), such as verified solutions. The student, which does not observe PI, learns from the teacher’s next-token distributions along its own sampled responses. However, PI can bias supervision toward shortcut reasoning with limited verification or reconsideration, accompanied by a sharp drop in the entropy of the output distribution. To address this challenge, we introduce self-distillation with verbal regularization (VRSD), which uses system prompts to regulate how the teacher uses PI, thereby encouraging rethinking and exploration to counteract premature commitment. Theoretically, we model prompting as attenuating the influence of PI on the teacher distribution and derive conditions for entropy preservation. Across three models and four tasks, VRSD improves Pass@1 over OPSD in 11 of 12 model-task settings, by 3.9 percentage points on average, while generally maintaining higher token entropy.
[NLP-203] A Production-Grade High-Efficiency Agent ic RL Framework for Frontier LLM s with Validated Recipes
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在后训练阶段采用强化学习（Reinforcement Learning, RL）时，面对日益复杂的代理任务（agentic tasks）所面临的框架臃肿、部署与调试成本高、同步瓶颈显著等问题。随着LLM能力向需调用外部工具、多轮交互及长周期决策的代理行为演进，传统主流RL框架难以高效支持异步、高吞吐的训练流程。其解决方案的关键在于提出LoongSage——一个面向代理原生（agent-native）、全异步的强化学习框架，核心代码量仅约14K行，具备轻量化设计。该框架通过单一控制器进程集中编排，并将所有GPU驻留组件封装为Ray Actors，实现高效资源管理；引入AgentFlow统一执行范式，兼容白盒与黑盒代理；采用滑动窗口策略控制轨迹过时（trajectory staleness），结合TransferMesh拓扑感知同步机制，在每步更新后基于各接收端最优传输路径路由权重，显著降低通信延迟；此外，通过内存高效的Teacher Manager实现教师模型缩放与显存约束解耦，原生支持多教师在线策略蒸馏（Multi-Teacher On-Policy Distillation）。实验证明，该框架在四次完整训练运行中持续提升任务奖励，有效控制训练-推理差距，且全异步流水线消除了同步空闲期，维持了高硬件利用率。
链接: https://arxiv.org/abs/2610.16352
作者: LoonSage Team:Bowei Song,Caiwei Zhao,Chuwen Liu,Dechuang Gu,Dong Fang,Feiyu Xie,Huaiyu Li,Hui Zhang,Jian Tian,Jian Wang,Jie Zhao,Liguang Li,Maoyun Wang,Meiyan Yang,Minghan Zhu,Pai Liu,Pandeng Yao,Renbao Tian,Ru Yang,Shuai Jia,Shuai Tan,Shuailong Li,Shuliang Li,Teng Jin,Tianze Yang,Wenbo Cui,Wenjing Bai,Wenkang Wu,Xiaowei Jiang,Xiaoyi Dong,Yanpeng Wang,Yanwen Wang,Yaohui Ma,Yawen Wang,Yihang Zhang,Yingjie Yan,Zhanbiao Lian,Zhenyu Lu,Zhida Hu,Zhihui Hao,Zhihui Hu,Zhiwen Xiao,Zhiyi Gao,Zhongwei Qi,Zhuangzhuang Wu,Ziqian Liu,Ziyue Zheng
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Reinforcement learning (RL) has emerged as a critical phase in the post-training pipeline of large language models (LLMs). As the frontier of LLM capabilities shifts from static reasoning to agentic tasks–characterized by external tool invocation, multi-turn interactions, and long-horizon decision-making–existing mainstream RL frameworks have grown increasingly unwieldy, raising the overhead of adoption, usage, and debugging. To address these challenges, we present LoongSage, an agent-native, fully asynchronous RL framework with a lightweight core of roughly 14K lines of code. Built upon Megatron for distributed training and SGLang for high-throughput inference, LoongSage centralizes orchestration within a single controller process and encapsulates all GPU-resident components as Ray actors. It introduces AgentFlow, a unified execution paradigm that orchestrates both white-box and black-box agents. For asynchronous execution, it integrates sliding-window strategies that bound trajectory staleness, and employs TransferMesh, a topology-aware synchronization mechanism that routes weights after each step over the fastest transport available to each receiver. Beyond agentic RL, LoongSage incorporates a memory-efficient Teacher Manager that decouples the scaling of teacher models from GPU memory constraints, natively supporting Multi-Teacher On-Policy Distillation. Extensive evaluations across four complete training runs, spanning complex reinforcement learning and multi-teacher distillation, demonstrate that task rewards improve steadily while the train-inference gap is kept under continuous control, and the fully asynchronous pipeline removes synchronous idle bubbles, sustaining hardware utilization.
[NLP-204] RIFT: Token Influence Follows the Path of Least Resistance
【速读】： 该论文旨在解决生成式 AI (Generative AI) 模型中“哪些上下文标记（context tokens）对模型输出答案具有决定性影响”的可解释性问题。现有基于图的分析方法如注意力传播（attention rollout）和注意力流（attention flow）虽能通过单次前向传播快速估算贡献度，但在解码器仅模型（decoder-only models）中的忠实性缺乏系统验证。本文提出一种新方法 RIFT（基于电阻的信息流在 Transformer 中），将解码器各层建模为以注意力权重为电导的因果网络，通过计算从输入标记到输出答案路径中最小电阻串联的等效电阻来量化其影响力。该方法利用经典最短路径动态规划算法精确求解，无需梯度或额外前向传播，在单个 CPU 线程上每条提示仅需 0.9–3.4 毫秒即可完成所有标记评分。实验表明，RIFT 在合成上下文检索任务与维基百科段落抽取式问答任务中，对替换一个标记后模型响应变化的预测能力达到斯皮尔曼相关系数 0.52–0.68，且在预测单一替换效果时，与两次随机替换间的一致性达成 77%–96%。相较于传统方法，RIFT 在 576 个独立测试提示上，于四个不同模型（Qwen3 与 Llama 3.2 两类）的预设比较中均表现最优，平均斯皮尔曼相关性提升 0.23 至 0.88，是所有所测注意力基得分中最佳或并列最佳。研究进一步揭示，累加所有路径信息的方法（如注意力流、有效电阻、注意力传播）主要反映标记位置分布特征：根据简单命题可知，早期标记因存在更多并行路径至答案而被高估。因此，该工作的关键在于引入基于电阻理论的精准路径建模，实现无需反向传播、高效且高忠实度的上下文重要性评估。
链接: https://arxiv.org/abs/2610.16349
作者: Tingan Jin,Shuhang Dong,Haosong Li
机构: University of California, Los Angeles(加州大学洛杉矶分校); Independent Researcher(独立研究员)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 9 pages, 5 figures, 4 tables
Abstract:Which context tokens drive a language model’s answer? Graph-based tools such as attention rollout and attention flow answer this from a single forward pass, yet their faithfulness in decoder-only models is rarely tested against interventions. We view the layers of a decoder as a causal network whose conductances are attention weights, and read a token’s influence on the answer from the resistance between them. We propose RIFT (Resistance-based Information Flow in Transformers), which scores a token by the series resistance of its path of least resistance to the answer, found exactly by a classic shortest-path dynamic program. On synthetic in-context retrieval tasks and extractive question answering over Wikipedia paragraphs, RIFT predicts the effect of replacing a token with a Spearman correlation of 0.52-0.68; predicting a single replacement, it reaches 77-96% of the agreement between two different random replacements. It needs no gradients and no extra forward passes, scoring every token of a prompt in 0.9-3.4 ms on one CPU thread, whereas the replacement experiment needs a forward pass per token. On 576 held-out prompts from four models of two families (Qwen3 and Llama 3.2), it outperforms token distance, attention rollout and attention flow in every pre-specified comparison, by 0.23 to 0.88 in mean Spearman correlation, and it is the best or tied-best of all attention-based scores we test. Aggregates that add up all routes (attention flow, effective resistance and rollout) instead mostly encode token position: as a simple proposition shows, early tokens have many more parallel routes to the answer.
[NLP-205] Strong yet Controllable LLM Abstention without Learning to Abstain
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在生成回答时存在幻觉（hallucination）的问题，核心挑战在于如何在保持高准确率的同时有效控制模型的响应覆盖范围（coverage）。现有方法主要分为两类：一是通过外部置信度阈值实现的选择性回答（selective answering），其优势在于可灵活调节覆盖率；二是将拒答机制内化于模型的拒答训练（abstention training），虽性能更优但一旦训练完成即固定覆盖率，缺乏灵活性。为公平比较二者，论文提出对风险（error rate）进行分解，将其拆解为“回答能力”与“区分能力”，分别衡量模型生成答案的准确性及其拒答判断的精确性。实验结果表明，拒答训练的优势主要源于更强的回答能力，而在使用强置信度估计器的情况下，两类方法的区分能力接近一致。基于此发现，论文提出一种模块化解决方案——先训练准确率、后进行后处理选择（Train for Accuracy, then Post-hoc Select, TAPS），该方法既保持了覆盖率的可控性，又仅需标准后训练流程和现成的置信度估计器，并能受益于任一组件的进展。在三个主流大模型与三个事实型问答基准上的实验证明，TAPS结合探针式置信度估计器，在匹配覆盖率下相比三种代表性拒答训练方法，平均风险降低1.5至9.4个百分点。此外，研究发现，单纯提升准确率训练可能削弱模型自身的置信信号，但通过重新拟合置信度估计器可基本保留其区分能力。综上，TAPS被确立为一个强大且可控的基线方法，同时为未来拒答训练设定了更高标准：若放弃对覆盖率的控制，所获收益应显著超越分别优化回答与拒答能力所能达到的上限。
链接: https://arxiv.org/abs/2610.16318
作者: Junwei Zhang,Yifei Wang,Wanli Yang,Hexiang Tan,Rui Tang,Xin Chen,Jingang Wang,Du Su,Fei Sun
机构: University of Chinese Academy of Sciences(中国科学院大学); State Key Laboratory of AI Safety, Institute of Computing Technology, CAS(中国科学院计算技术研究所人工智能安全重点实验室); Beijing Institute of Technology(北京理工大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models (LLMs) can reduce hallucination by abstaining from questions they would likely get wrong. Abstention can be implemented outside the model (selective answering) using a confidence threshold that controls the answer rate (coverage), or learned inside it (abstention training), which can perform better but fixes coverage once trained. To compare the two, we match their coverage and decompose risk (error rate) into answering ability and distinguishing ability, i.e., how well the model answers and how accurately it abstains, respectively. The decomposition shows that abstention training performs better mainly by answering better: with strong confidence estimators, the two have nearly equal distinguishing ability. This motivates Train for Accuracy, then Post-hoc Select (TAPS), a modular approach that keeps coverage controllable, needs only standard post-training and an off-the-shelf confidence estimator, and benefits from advances in either component. Across three LLMs and three factual QA benchmarks, TAPS with a probing-based confidence estimator attains 1.5–9.4 percentage points lower mean risk than three representative abstention training methods at matched coverage. Within TAPS, accuracy training often weakens the model’s own confidence signals, yet refitting the estimator largely preserves distinguishing ability. Together, these results establish TAPS as a strong yet controllable baseline and raise the bar for future abstention training: sacrificing control over coverage should yield gains beyond what can be achieved by optimizing answering and abstaining separately.
[NLP-206] ATTCK-R: Knowledge-Aware Long-Horizon Jailbreaking for Protected LLM s
【速读】： 该论文旨在解决在真实部署场景下，针对具备提示词安全防护机制的大语言模型（Large Language Models, LLMs）进行鲁棒性评估时所面临的挑战。现有越狱攻击评估方法通常依赖静态提示、有限交互轮次或弱保护系统，无法准确反映实际中采用提示词驱动的安全防护机制下的模型脆弱性，导致对真实世界风险的高估。其核心解决方案是提出一种基于规划的多轮红队测试框架——自适应追溯与知识型战术优化红队测试（Adaptive Traceback and Tactic Curation for Knowledge-based Red-teaming, ATTCK-R）。ATTCK-R将越狱过程建模为一个长时程轨迹优化问题，通过结构化信念状态进行状态表征；引入基于优化的战术萃取机制，从历史成功与失败经验中抽象可复用的攻击策略；并采用有知觉的追溯机制，从具有战略意义的状态恢复攻击，而非盲目重置。实验表明，在9个主流大模型上配置单一提示词安全层的情况下，ATTCK-R显著优于当前先进的多轮越狱框架，消融实验进一步验证了长时程规划与战术萃取对攻击有效性的提升，以及自适应追溯对效率的优化作用。
链接: https://arxiv.org/abs/2610.16305
作者: Md Mushfiqur Rahman,Tobias Rohde,Rahul Tripathi,Khalil Elkhalil,Kevin Lybarger
机构: George Mason University (乔治梅森大学); Amazon
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL)
备注:
Abstract:As large language models are deployed behind increasingly sophisticated safety mechanisms, evaluating their robustness under realistic adversarial conditions becomes increasingly difficult and critical. Existing jailbreaking evaluations often rely on static prompts, limited interaction horizons, or lightly protected systems, failing to reflect protected deployment settings with prompt-based safety protections. We show that the introduction of a prompt-based safety protection reduces the average attack success rate of a strong multi-turn baseline from 0.896 to 0.608 (32.1% relative), demonstrating that unprotected evaluations substantially overestimate real-world vulnerabilities. We propose Adaptive Traceback and Tactic Curation for Knowledge-based Red-teaming (ATTCK-R), a planning-driven multi-turn red teaming framework designed for protected black-box systems. ATTCK-R models jailbreaking as a long-horizon trajectory optimization problem using a structured belief state, introduces optimization-based tactic curation to abstract reusable strategies from prior successes and failures, and employs informed traceback to resume attacks from strategically meaningful states rather than restarting. Evaluated across 9 widely used LLMs configured with a single prompt-based safety protection layer, ATTCK-R consistently outperforms strong recent multi-turn jailbreaking frameworks. Ablation results confirm that long-horizon planning and tactic curation improve attack effectiveness, while adaptive traceback improves efficiency through structured recovery from conversational failure.
[NLP-207] REVEAL: Reverse Engineering Agent ic Latents from Frontier-Model Traces
【速读】： 该论文旨在解决如何将大型前沿语言模型（Frontier LLM）在复杂交互任务中积累的经验高效迁移至参数冻结的小型开源代理（frozen agent）的问题，以降低推理成本。传统方法如知识蒸馏和隐变量通信通常依赖目标模型的训练或对源模型内部结构的访问，限制了其适用性。本文提出一种无需训练的解决方案——\textbfREVEAL（Reverse Engineering Agentic Latents），其核心在于利用前沿模型生成的文本轨迹，提取可复用的局部控制信号。关键创新是引入“代理隐变量”（Agentic Latents），包括锚定状态与反馈-响应循环的动态特征，通过在目标模型上重放轨迹，捕捉成功与失败回溯中的响应转移差异，从而构建针对特定目标的可控引导方向。该方向在每步响应前应用，实现对代理行为的精准调控。实验表明，在Terminal-Bench 2.1与Humanity’s Last Exam等基准上，该方法显著提升三类目标代理的准确率，且表征分析与行为案例研究验证了前沿模型经验可通过局部激活控制实现有效复用。
链接: https://arxiv.org/abs/2610.16288
作者: Yuetian Du,Meng Wang,Yucheng Wang,Zhangzhe Chen,Wei Wang,Boyu Yang,Weiqi Zhai,Bing Zhao,Ming Kong,Zhan Zhou,Jie Liu,Qiang Zhu
机构: Zhejiang University(浙江大学); South China University of Technology(华南理工大学); Alibaba Group(阿里巴巴集团); Fudan University(复旦大学); City University of Hong Kong(香港城市大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Frontier LLM agents perform complex interactive tasks, but their inference costs motivate transferring their experience to smaller open-source agents. Distillation and latent communication typically require target-model training or access to source-model internals. We study how textual frontier-model trajectories can provide reusable local controls for a frozen agent. Inspired by protocol reverse engineering, we introduce \textitAgentic Latents , comprising anchored states and dynamic profiles of feedback–response cycles elicited by replay in the target model. Our training-free framework, \textbfREVEAL ( \textbfREV erse \textbfE ngineering \textbfA gentic \textbfL atents), uses these profiles and trajectory outcomes to match and select local events from successful and failed rollouts of the same task and source model. Their response-transition contrasts yield a target-specific steering direction, applied before each response. Within-benchmark evaluations on Terminal-Bench 2.1 and Humanity’s Last Exam with tools report higher accuracy than unmodified agents across three targets. Representation analyses and behavioral case studies support reusing frontier-model experience as local activation control.
[NLP-208] agent -risk: Risk Assessment for Changes to AI Agents
【速读】： 该论文旨在解决在部署后对AI代理（如更严格的安全门限或新模型）进行变更时，如何在不中断系统运行的前提下，确保变更决策的安全性与可验证性。传统风险分析难以适用于部分记录和部分系统变更的场景，而本文提出通过“代理风险重播”（agent-risk replay）机制，基于现有日志对已发生事件进行精确回放，假设系统其余部分保持不变且代理未预知变更。其核心解决方案在于：明确日志必须记录哪些信息以实现精确重播，并识别哪些保存状态能够支持一组安全门限的验证。针对独立事件，论文给出了决策所需重跑次数的上界以及对实时行为漂移的容忍度。实验表明，在一个bash沙箱环境中，仅依靠7,200个事件的日志无法判断两个重试门限的正确性，因为日志遗漏了代理被拒绝后实际执行的行为。通过从保存状态中仅重跑各门限被阻断的事件，可在代理调用次数仅为事后调整后的实时测试一半的情况下，准确复现所有真实决策。此外，由于日志无法反映停止前是否已产生危害，即使无每步伤害，停止门限的经认证危害上限仍可能上升至2.5倍。因此，审批必须依赖预先设计的证据体系——包括特定日志字段、保存状态及目标性重跑机制。
链接: https://arxiv.org/abs/2610.16280
作者: Yue Zhao,Zixiang Xu,Angela Huyue Zhang,S. Alex Yang,Xiyang Hu
机构: University of Southern California(南加州大学); London Business School(伦敦商学院); Arizona State University(亚利桑那州立大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 43 pages, 5 figures, 7 tables. Code, registrations, and logs: this https URL
Abstract:Before changing a deployed AI agent (a stricter safety gate, a new model), organizations must approve the change, starting from existing logs. Extending classical risk analysis to partial records and partial changes to running systems, agent-risk replays logged episodes, assuming nothing else changes and the agent does not anticipate the change. The log settles episodes acting as before, a declared stop value settles stopped ones, and the rest need logged support, a saved-state rerun, or a bound. We prove what logs must record for exact replay and which saved states serve a gate family. For independent episodes, we bound the reruns a decision needs and tolerable drift from live behavior. In a bash sandbox, a 7,200-episode log alone decided neither of two retry gates, which let the agent act after refusals the log never shows. Rerunning only each gate’s blocked episodes from saved states reached both live decisions at half the agent calls of a live test sized post hoc for them, given the log. Without per-step harm, a stopping gate’s certified harm ceiling rose up to 2.5-fold, since the log hides whether harm preceded the stop. Approval thus needs evidence designed in advance: log fields, saved states, targeted reruns.
[NLP-209] Deep Speech Representations Track Word Timing along a Low-Dimensional Circular Trajectory
【速读】： 该论文旨在解决语音模型中词时序信息（word-timing information）的表征机制问题，即在深度语音表示空间中，词的时序结构如何被编码与体现。其核心挑战在于，尽管语音时序对准确识别词汇至关重要，但现有研究尚未充分揭示这一信息在模型内部的具体表征方式。解决方案的关键在于采用表示几何（representational geometry）框架，通过分析低维子空间中的表示轨迹来探究语义和时序结构的内在组织规律。研究发现，无论是在自监督还是有监督的九种不同语音模型中，词的时序信息均呈现出近似圆形的轨迹模式，表明这种周期性结构是深层语音表示的普遍特性。进一步地，在HuBERT模型中，该圆形轨迹在单个词层面具有足够的一致性，从而可被用于实现词边界检测。基于此，作者提出旋转分割（Rotary Segmentation, RoSeg）方法，通过检测时序子空间中周期完成的时机来定位词边界，在LibriSpeech数据集上达到64.3%的边界F-score。尽管需少量监督以确定最优时序子空间，但该结果显著揭示了HuBERT模型中隐含的、远超以往认知的词时序知识。
链接: https://arxiv.org/abs/2610.16273
作者: Burin Naowarat,Hao Tang,Sharon Goldwater
机构: The Centre for Speech Technology Research, School of Informatics, University of Edinburgh(爱丁堡大学信息学院语音技术研究中心)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
备注: Accepted at SLT 2026
Abstract:Speech timing is fundamental for speech models to recognize words accurately, but how word-timing information is represented in speech models remains underexplored. We study this question using the framework of representational geometry, which investigates whether and how concepts are represented in low-dimensional subspaces of the representation space. Prior work in cognitive neuroscience has shown that when HuBERT feature sequences are averaged across words of the same duration, they form near-circular trajectories from word onset to word offset in a two-dimensional “temporal subspace.” Here we show that this circular structure of word timing is a general property of deep speech representations, by demonstrating its emergence in nine additional speech models covering both supervised and self-supervised. We then move beyond averaged word segments and show that in HuBERT, the circular trajectories are consistent enough at the level of individual words to perform word segmentation. Our approach, Rotary Segmentation (RoSeg), achieves 64.3% boundary F-score on Librispeech by detecting word boundaries when cycles complete in the temporal subspace. Although supervision is needed to identify a good subspace, nevertheless this result demonstrates much stronger implicit word knowledge in HuBERT than has previously been shown.
[NLP-210] Error Correction in Trained TTT Layers
【速读】： 该论文旨在解决生成式模型在测试时通过快速权重记忆（fast-weight memory）进行在线学习时，其内部更新机制中预测衍生更新（prediction-derived update）的实际作用与行为价值之间的差异问题。具体而言，现有基于梯度的测试时训练（Test-time-training, TTT）方法虽能实现参数更新，但未能明确揭示这些更新如何有效修正已有关联，尤其在缺乏预测误差残差（prediction-error residual）的情况下。论文的关键解决方案在于：针对点积目标函数，通过重新参数化证明内循环上升与下降的等价性；针对TTT-Linear和TTT-MLP所采用的基于层归一化（LayerNorm）的平方误差目标，推导出内梯度并识别出一个在释放初始化时于零ε极限下抵消的预测衍生分量。通过实测三组训练检查点，发现移除该预测贡献后保留层归一化雅可比矩阵，模型平均梯度范数比降至0.25至0.32之间。在《战争与和平》10万+词元上的实验进一步表明，仅移除预测项而保持局部逐令牌归一化匹配，使TTT-Linear-760M与1.3B模型困惑度上升2.8–2.9%，而完全冻结快速权重更新则导致困惑度上升4.2–4.5倍。合成实验还揭示了简单δ规则更新对步长的高度敏感性。研究结果明确区分了预测衍生更新的数学形式与其实际性能影响，并强调必须结合实现层面分析与实时干预才能准确评估测试时记忆的有效性。
链接: https://arxiv.org/abs/2610.16268
作者: Yuhe Hu,Wen Jia Hu
机构: Duke University (杜克大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 13 pages, 3 figures. Code and saved results included as ancillary material and at this https URL
Abstract:Test-time-training (TTT) layers update fast-weight memory while processing an input sequence. A gradient-based update, however, does not by itself establish how the memory revises an existing association. We study the structure and empirical role of prediction-derived updates in released TTT implementations. For dot-product objectives, we characterize the absence of a prediction-error residual and establish conditions under which inner-loop ascent and descent are equivalent after reparameterization. For the LayerNorm-based squared-error objective used by TTT-Linear and TTT-MLP, we derive the inner gradient and identify a prediction-derived component that cancels at the released initialization in the zero-epsilon limit. We then measure the full normalized prediction’s contribution on three trained checkpoints. Removing this contribution while retaining the LayerNorm Jacobian gives a mean gradient norm ratio of 0.25 to 0.32 across the models. On 100,303 tokens from War and Peace, the same intervention with local per-token norm matching raises perplexity by 2.8-2.9% on TTT-Linear-760M and 1.3B, whereas freezing all fast-weight updates raises it by factors of 4.2-4.5. Separate synthetic tests illustrate the dependence of plain delta-rule overwriting on step size. The results distinguish a prediction-derived update from its behavioral value and show why implementation analysis and live interventions are both needed to assess test-time memory.
[NLP-211] Avoiding What Your Teacher Already Knows: Entity Familiarity for Deep-Search Distillation
【速读】： 该论文旨在解决生成式AI在深度搜索任务中因教师模型对特定实体过于熟悉而导致的监督信号弱化问题，即“教师相对难度”（teacher-relative difficulty）——当教师模型熟悉底层实体图谱中的关键实体时，复杂问题对其而言仍较易处理，从而导致搜索需求低、训练轨迹信息量不足。其核心解决方案是提出一种轻量级的实体熟悉度评估机制——自报告实体熟悉度（Self-Reported Entity Familiarity, SEF），通过在构建实体图谱前直接询问教师模型是否熟悉某实体，作为筛选依据，以增强合成轨迹的搜索强度与监督质量。相较于高成本的证据验证型实体熟悉度（Evidence-Verified Entity Familiarity, EVF），SEF具备可扩展性，并在多教师场景下验证了其有效性：经SEF筛选的合成轨迹表现出更高的实际搜索需求，显著提升下游学生模型的准确率；进一步通过跨教师复用（利用Kimi-K3的SEF判断筛选其他教师的图谱）实现数据增益，最终使Qwen3.8-27B在BrowseComp等基准上准确率从44.10%提升至66.41%，达到同规模模型中的领先水平。因此，SEF的关键优势在于以极低成本实现了更高质量的深度搜索监督信号生成。
链接: https://arxiv.org/abs/2610.16251
作者: Yifei Liu,Zhixiang Liang,Wentao Tan,Yicheng Wang,Jiaqi Wang,Nan Duan,Qiong Cao
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Deep-search agents are commonly trained by distilling tool-use trajectories from stronger teacher models. However, structurally complex questions may remain relatively easy for teachers familiar with key entities in the underlying entity graph, resulting in lower search demand and weaker supervision. We refer to this teacher-specific variation in question difficulty as teacher-relative difficulty and investigate entity familiarity as an early predictor. We first introduce Evidence-Verified Entity Familiarity (EVF), which estimates familiarity by verifying a teacher-generated factual brief against online evidence. Across multiple teachers, EVF is significantly correlated with teacher-relative difficulty with AUCs of 0.60–0.70. However, its high cost makes it impractical for constructing entity graphs at scale. To address this limitation, we propose Self-Reported Entity Familiarity (SEF), a lightweight gating signal that directly asks a teacher whether it is familiar with an entity before admitting that entity into the graph. Across multiple teachers, trajectories from SEF-gated synthesis exhibit greater observed search demand and yield higher downstream student accuracy than those from ungated synthesis. Because familiarity is teacher specific, we further study cross-teacher reuse. We screen graphs collected for other teachers using Kimi-K3’s SEF judgments and use the retained graphs to produce 5,411 additional trajectories. Combined with 2,200 trajectories from newly constructed, SEF-gated graphs, these data improve Qwen3.8-27B’s accuracy from 44.10% to 66.41% on a BrowseComp subset, and achieving state-of-the-art performance among models of comparable size on BrowseComp, SEAL-0, HLE-text, and xbench. Overall, SEF provides a low-cost way to synthesize more informative supervision for deep-search distillation.
[NLP-212] Lüshi-1840: Building and Probing a Knowledge Cutoff for a Classical Chinese Language Model
【速读】： 该论文旨在解决预训练中文语言模型在研究历史语言与事后认知（hindsight）时面临的文本时间标注缺失与朝代标签不可靠的问题。现有公开的前现代中文语料库普遍缺乏确切出版时间，且朝代标签存在严重误差，导致模型训练数据的时间准确性难以保障。为应对这一挑战，研究者构建了Lüshi-1840模型——一个从零开始训练的983M参数模型，其训练数据来自经筛选后确认创作于1840年以前的12,262部古籍（共12.8亿字符）。针对早期语料中因启发式时间划分方法导致的“1644年截止”但实际包含20世纪材料的问题，论文提出了一套重新标注与清洗的数据处理流程，并通过未使用该流程的独立标记物进行审计，仍检测到后续内容的残留污染。在模型评估方面，研究采用相对于现代基准的意外度（surprisal）作为指标，发现对1840年后编纂的列传，其意外度比1820年前事件高出1.38倍（95%置信区间1.33–1.44），显著高于现代通用模型约1.00的水平；而在一系列294M参数模型中，加入后期朝廷档案后，该比例由1.33降至1.16，且两模型自1820年代起出现明显分歧。该探测结果揭示了模型对史料覆盖范围的敏感性，但并未确立单一时间截断点，且所有探针均未能暴露审计中发现的污染问题。本研究的关键解决方案在于建立可复现、基于实证的语料重校准流程与多维度评估机制，从而提升历史语言模型的时间可靠性与可信度。
链接: https://arxiv.org/abs/2610.16201
作者: Alex Liu
机构: Ybuild LLC
类目: Computation and Language (cs.CL)
备注: 13 pages, 4 figures
Abstract:Language models trained only on text from before a given year are useful for studying hindsight and historical language, but the open collections of pre-modern Chinese carry no publication dates and unreliable dynasty labels. We present Lüshi-1840, a 983M-parameter model trained from scratch on 12,262 books (1.28B characters) selected as composed before 1840. A first corpus, dated by a common heuristic, in effect ended at 1644 while containing twentieth-century material. We describe the re-dating and cleaning pipeline that replaced it and an audit with markers the pipeline did not use, which still finds later material. On the model side we measure surprisal, relative to a modern reference, on annals compiled after the cutoff. For events after 1840 against events up to 1820 the ratio is 1.38 (95% interval 1.33-1.44), against about 1.00 for modern general-purpose models. In a series of 294M models, adding the later court records lowers the ratio from 1.33 to 1.16, and the two models diverge from the 1820s, where the added records begin. The probe tracks the coverage of sources and does not establish a single cutoff year; none of our probes exposes the audited contamination. Model, book list and scripts will be released.
[NLP-213] DeepThink: A Conceptual Model for Knowledge-Addressable Recurrent Language Modeling
【速读】： 该论文旨在解决传统循环语言模型在知识利用与计算动态性方面的局限性，即如何有效整合外部知识并实现可变的、上下文感知的计算路径。其核心解决方案是提出DeepThink框架，该框架通过引入可访问知识的结构化设计，将知识存储（知识库）、知识激活（知识激活器）、知识组装（知识组装）与状态更新（共享思维算子）分离为独立模块，并以递归方式协同工作。关键创新在于将知识容量、激活宽度和递归深度作为独立的设计变量，支持从完整参数包到细粒度专家模块或低秩模块的灵活配置；同时形式化了各组件间的接口与执行语义，包括状态依赖的选择、参数空间的动态组装以及“思考-生成”分阶段的执行流程。该设计实现了对预训练模型的精确固定调度重构，结合共享基础重参数化与输出尾部复用等初始化策略，构建了从静态预训练模型向动态选择与组装的递归计算路径的可行范式，从而在条件计算、循环建模、模块化参数与知识接口的融合中，系统性地回答了相关集成所引发的关键设计问题。
链接: https://arxiv.org/abs/2610.16171
作者: Weiguo Chen
机构: 未知
类目: Computation and Language (cs.CL)
备注: 31 pages,7 figures
Abstract:This manuscript presents DeepThink, a knowledge-addressable framework for recurrent language modeling. A Knowledge Bank stores typed Knowledge Units, a Knowledge Activator selects units from the current execution state, Knowledge Assembly produces an executable configuration, and a Shared Thinking Operator updates that state. An output stage then organizes generation. This organization exposes knowledge capacity, activation width, and recurrent depth as separate design variables, supporting both complete parameter packages and fine-grained expert or low-rank modules. We formalize the interfaces and their execution semantics, including state-dependent selection, parameter-space assembly, and thinking followed by generation. For compatible templates and complete runtime-state configurations, we establish an exact Fixed-Schedule Realization of an existing fixed-depth model. Exact shared-base reparameterization and pretrained output-tail reuse provide complementary initialization constructions. Together, these results define a route from pretrained models to dynamically selected and assembled recurrent computation. A synthesis of conditional computation, recurrent models, modular parameters, and knowledge interfaces positions DeepThink within the research landscape and identifies the design questions arising from their integration.
[NLP-214] hink Before You Speak: Enabling Training-Free Latent Communication Between Language Models
【速读】： 该论文旨在解决多模型协作中因传统文本通信方式导致的信息传递瓶颈与预填充延迟问题，尤其是在异构模型间进行高效协同推理时的挑战。现有方法虽通过共享关键-值（Key-Value, KV）缓存实现潜在空间协作，但依赖为每对模型训练专用融合适配器（fusion adapter），限制了实际部署的灵活性与可扩展性。本文提出CacheFuse，一种无需训练的通用方法，通过在线线性插值将上游模型的KV缓存注入下游模型，从而在不改变模型结构的前提下实现跨模型的高效知识迁移。其核心创新在于证明：仅通过简单插值即可显著提升下游模型的推理性能，且在四个推理与知识基准测试中均表现出优于单模型基线的准确率（最高提升8.6个百分点），同时相比基于文本通信的多模型推理，实现高达12.2倍的首次令牌生成时间（TTFT）加速，显著提升了多模型系统在复杂任务中的效率与实用性。
链接: https://arxiv.org/abs/2610.16165
作者: Rishi Khare
机构: Georgia Institute of Technology(佐治亚理工学院)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted at the 2nd Workshop on Efficient Reasoning at COLM 2026
Abstract:Large language models (LLMs) have demonstrated remarkable capabilities in solving complex tasks in coding, mathematical logic, and iterative reasoning, especially by coordinating effectively within multi-agent systems (MAS). However, language agents traditionally communicate via text exchange, which introduces a bottleneck in information transfer and adds prefill latency for downstream agents. Recent work proposes that directly sharing the key-value (KV) cache between language models can enable efficient latent collaboration, entirely bypassing standard text-based communication. To enable latent communication between heterogeneous models of different sizes, prior work introduces trained fusion adapters for each model pair, which limits practical deployment. We introduce CacheFuse, a training-free method which demonstrates that augmenting a downstream model with an upstream model’s KV cache using linear interpolation consistently enables superior performance than the standalone downstream model can achieve. We systematically analyze CacheFuse across four reasoning and knowledge benchmarks and demonstrate up to 8.6 percentage points higher accuracy compared to single-model baselines and up to 12.2x TTFT speedup compared to text-based communication in multi-model inference.
[NLP-215] Hashrope: A Persistent Data Structure for Mechanisms of LLM Context Management ICTAI2026
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）服务栈中上下文管理的复杂性问题，即当前依赖四种独立专用结构分别实现增量编辑、树搜索中的分支与快照、可重用前缀识别以支持键值缓存（KV-cache）复用，以及重复片段编码。这些机制虽各具功能，但导致系统碎片化且难以统一优化。其核心解决方案是提出一种统一的持久化平衡绳索结构（Hashrope），该结构将上述四种机制归约为单一基于梅森素数（Mersenne prime）上多项式哈希的权衡平衡树，每个节点存储其子树的多项式哈希值，从而在保证对上下文进行对数时间复杂度（logarithmic）编辑的同时，支持对内容身份的对数时间可验证查询。通过预注册、可复现的实验验证了各机制的有效性，并与现有部署的专用组件对比：在交错执行编辑与身份查询的工作负载下，Hashrope相较Ropey编辑库性能提升无界（百万字符时达369.7倍）；在实际vLLM堆栈中服务Qwen2.5-7B-Instruct-1M模型时，相较于径向匹配器（radix matcher），Hashrope不仅精确识别出百万级共享前缀，且速度提升1.20倍，且可在密集引擎无法预填充的超长上下文长度下运行，展现出显著的效率与扩展优势。
链接: https://arxiv.org/abs/2610.16144
作者: Muntaser Syed(Florida Institute of Technology),Ameerah Alsulami(Florida Institute of Technology),Marius Silaghi(Florida Institute of Technology)
机构: Florida Institute of Technology
类目: Data Structures and Algorithms (cs.DS); Computation and Language (cs.CL)
备注: Accepted at IEEE ICTAI 2026 (full paper)
Abstract:Serving stacks for large language models manage context through four recurring mechanisms: incremental edits to a growing context, branching and snapshotting for tree search, identification of reusable prefixes for KV-cache reuse, and encoding of repeated segments. Each mechanism is served today by a different specialist structure. We show that all four reduce to a single persistent weight-balanced rope whose every node carries a polynomial hash of its subtree over a Mersenne prime, so that the structure supports logarithmic edits and logarithmic verified content-identity queries at the same time. We validate each mechanism with pre-registered, replicated experiments, then compare against a deployed specialist per mechanism: the Ropey editing library, a faithful PagedAttention copy-on-write block table, and the vendored SGLang radix matcher. On workloads that interleave edits with identity queries, Hashrope is net-superior to Ropey by a factor that grows without bound (369.7x at one million characters). Against the radix matcher on a live vLLM stack serving Qwen2.5-7B-Instruct-1M on four A100 GPUs, Hashrope identifies a one-million-token shared prefix exactly and 1.20x faster, at a context length the dense engine cannot even prefill.
[NLP-216] Phantom Merge: When Your Large Language Model Agents Pick One but Tell You About Another
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）智能体在处理多实体任务时出现的“幻觉合并”（Phantom Merge, PM）问题，即智能体将未经支持的主张错误地归属于其承诺的实体，即使任务整体完成正确也可能发生此类错误。其核心挑战在于确保生成内容中的证据与对应实体之间的绑定关系准确可靠。解决方案的关键在于提出一个两阶段框架：首先，通过锚点-基础风险评估（Anchor-Grounded Risk, AGR）模块，结合智能体生成前隐藏状态的表征读出与基于雅可比矩阵的功能信息，分别提供粒度级别的槽位风险得分和锚点值风险得分，实现对每个主张层面的风险精准估计；其次，采用固定锚点修正（Fixed-Anchor Correction, FAC）模块，对高风险主张进行审计，并基于观测到的锚点证据选择性修正或移除无法支持的主张。实验结果表明，该方法在购物检测数据集上达到96.2%的AUROC和97.3%的平均精度，在端到端任务中将轨迹级PM率从51.4%显著降低至6.2%，同时保持80.7%的输出主张保留率，充分验证了内部风险估计与证据约束修正相结合的有效性。
链接: https://arxiv.org/abs/2610.16141
作者: Wangshu Zhu,Qianwen Guo,Yushun Dong
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:LLM agents increasingly operate over multiple entities, where reliable response generation requires not only retrieving relevant evidence but also preserving its association with the correct entity. We identify Phantom Merge (PM), an entity-evidence binding failure in which an agent attributes unsupported claims to its committed entity, potentially even when the task is completed correctly. We formalize PM at the claim level and characterize its prevalence and error subtypes across four agent domains and four backbone families. To address this problem, we propose a two-stage framework for claim-level risk estimation and correction. First, Anchor-Grounded Risk (AGR) combines a representational readout of the agent’s pre-emission hidden state with Jacobian-based functional information, providing complementary slot-level and anchor-value risk scores. Second, Fixed-Anchor Correction (FAC) audits flagged claims and selectively revises them using observed anchor evidence or removes them when no supported replacement is available. On the held-out Shopping detection cohort, AGR-slot achieves an AUROC of 96.2% and an average precision of 97.3%, outperforming the evaluated detection baselines. In the end-to-end Shopping mitigation evaluation, FAC reduces trajectory-level PM from 51.4% to 6.2%, with an output-to-original claim ratio of 80.7%. These results demonstrate the prevalence of entity-specific grounding failures and the effectiveness of combining internal risk estimation with evidence-constrained claim correction. Our code is publicly available at this https URL.
[NLP-217] Omni-Modal LLM s Know More Than They Say
【速读】： 该论文旨在解决生成式多模态大模型（Omni-LLMs）中普遍存在的“知而不能言”（knows-but-cannot-say）问题，即模型虽能正确感知非文本模态（如视觉、音频、惯性测量单元IMU信号）信息，却难以在生成语言响应时有效表达这些信息。其核心问题是：非文本模态表征与语言响应表征之间存在显著的层次对齐不足，导致模型对非文本模态的关注度和语义整合能力弱于文本模态。解决方案的关键在于提出一种名为OmniCatalyst的简单策略——“感知→推理”范式，通过在输入阶段引入由模型自动生成的文本描述来补充原始非文本模态输入，从而增强非文本知识在语言生成过程中的可访问性。实验表明，该方法在7个Omni-LLMs上均显著提升性能（平均+7.29%），并实现更均衡的模态偏好；进一步分析显示，该策略提升了词级别模态对齐（Token-level Modality Alignment, TMA），验证了其通过改善跨模态对齐机制来增强非文本信息表达的有效性。
链接: https://arxiv.org/abs/2610.16135
作者: Na Min An,Taekyung Kim,Jaemyung Yu,Hyunjung Shim,Dongyoon Han
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: 36 pages, 25 figures, 14 Tables
Abstract:Omni-modal large language models (Omni-LLMs) extend LLMs to process heterogeneous modalities (e.g., vision, audio, and IMU signals) alongside language, enabling a broader range of multimodal tasks. Yet we find that these models often fail to express information from non-textual modalities even when they perceive it correctly. We term this disconnect knows-but-cannot-say. To investigate this phenomenon, we introduce Token-level Modality Alignment (TMA), which measures the alignment between individual modality representations and response representations. Our analysis shows that non-textual representations are less aligned with response tokens and receive less attention than textual representations, suggesting a potential explanation for uneven modality preferences. Motivated by these findings, we propose OmniCatalyst, a simple perceive \rightarrow infer strategy that complements the original modality inputs with self-generated text, making non-textual knowledge more accessible. Across seven Omni-LLMs, OmniCatalyst consistently improves performance (+7.29%) and yields more balanced modality preferences, with further gains from diverse training strategies. Further analysis demonstrates that the knowledge extraction phases increase TMA, suggesting that improved alignment makes non-textual information more accessible during response generation. In summary, our findings uncover how Omni-LLMs can better express modality knowledge through language. Our code is available at this https URL.
[NLP-218] Query Your Way Out: Benchmarking Multi-Turn Knowledge Reasoning through Generalized Puzzle Solving
【速读】： 该论文旨在解决现有大语言模型（Large Language Models, LLMs）推理评估基准在谜题求解任务中普遍存在的局限性，即多数基准局限于特定环境、机械规则僵化且解法目标高度受限，难以全面检验模型在复杂、开放场景下的交互式推理能力。为此，论文提出主动探测基准（Active Probing Benchmark, APB），其核心在于将概念识别转化为一种交互式推理任务，通过多轮主动探测（multi-turn active probing）与推断，评估模型在八个不同学术领域中对960个具有专业难度的目标概念进行精准定位的能力。解决方案的关键在于构建一个可扩展的基准生成流程，不仅融合了专业知识评估机制，还强调模型在动态交互环境中获取信息并运用程序性推理（procedural reasoning）来调用和应用领域知识的能力。实验结果揭示了一个反直觉现象：大多数开源推理模型在交互式任务中显著逊于标准指令微调模型，表明当前模型在缺乏明确引导时难以有效执行交互式知识获取与应用，凸显了提升模型在开放交互场景下稳健推理能力的重要性。
链接: https://arxiv.org/abs/2610.16133
作者: Xin Cheng,Junming Yang,Ning Xu
机构: Southeast University (东南大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 20 pages, 8 figures
Abstract:Puzzles offer a natural testbed for Large Language Models (LLMs) reasoning because they pose verifiable goals under partial information and require multi-step decisions to reach a solution. However, existing puzzle benchmarks typically focus on narrow environments with rigid mechanics and tightly constrained solution targets. To bridge this gap, we introduce the Active Probing Benchmark (APB), which generalizes puzzle-solving by casting concept identification as an interactive reasoning task that integrates specialized knowledge assessment into the puzzle setting. Specifically, APB evaluates whether models can acquire information via multi-turn active probing and inference to pinpoint 960 target concepts of specialized difficulty across eight diverse academic domains. Our evaluation reveals a counter-intuitive phenomenon: most open-source reasoning models significantly underperform their standard instruction counterparts. This finding suggests that in interactive settings, these models struggle to leverage procedural reasoning to reliably access and apply specialized domain knowledge. Beyond these empirical insights, APB provides a scalable pipeline for generating benchmarks that probe knowledge reasoning under interaction across diverse settings.
[NLP-219] PID-Gated Continual Prompt Adaptation Under Noisy Feedback
【速读】： 该论文旨在解决大语言模型（LLM）代理在非平稳环境中的持续指令适应问题，即当任务随时间发生漂移时，如何有效处理不准确反馈以避免因频繁修订指令而导致的累积错误。其核心挑战在于：在反馈噪声普遍存在的情况下，如何确定反馈触发永久性指令更新的时机，并权衡编辑调用成本。解决方案的关键是引入受比例-积分-微分（PID）控制启发的一系列轻量级门控机制，该机制能够响应当前误差、随时间累积证据，并在指令修改后回滚性能退化。实验在多种漂移任务、不同规模的LLM、不同噪声水平及随机种子下进行，结果表明，采用PID门控的持续提示适应方法在任务性能（高精度、低误差）与编辑调用预算之间实现了更优的帕累托前沿，证明了基于控制理论的调度策略是一种简单且通用的鲁棒提示适应方法。
链接: https://arxiv.org/abs/2610.16129
作者: Zhentao Xu
机构: Independent Researcher(独立研究员); Sunnyvale, CA, USA(美国加利福尼亚州森尼韦尔)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Deploying LLM agents in non-stationary environments requires continually revising their instructions as the task drifts, yet rewriting after every reported mistake can turn noisy feedback into persistent errors. This raises an often-overlooked scheduling question in prompt adaptation: when should feedback trigger a lasting instruction update, and at what editor-call cost? Taking inspiration from proportional-integral-derivative (PID) control, we introduce a family of lightweight gates that react to the current error, accumulate evidence over time, and roll back deterioration after an edit. We benchmark these gates against both simple baselines and advanced prompt optimizers (TextGrad and GEPA), over a diverse suite of drifting tasks, LLM parameter sizes, feedback-noise levels, and random seeds. PID-gated continual prompt adaptation yields a more favorable Pareto trade-off between task performance (high accuracy, low error) and editor-call budget than the baselines considered, pointing to control-theoretic scheduling as a simple, general tool for robust prompt adaptation under unreliable feedback.
[NLP-220] Beyond the greenwash: do LLM s ecologically sound decisions hold up under pressure?
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在社会决策中广泛应用时所隐含的偏见难以衡量的问题，尤其针对闭源权重模型中潜在的、深层的生物系统相关偏见（称为“生物对齐”即 bioalignment）及其鲁棒性评估难题。现有行为评分无法区分表层行为与深层内在倾向，导致评估失效。为此，研究提出一种基于已知剂量校准阶梯（known-dose calibration ladder）的解决方案：在相同300亿参数的专家混合（Mixture-of-Experts）基础模型上，通过三组持续预训练（Continued Pretraining, CPT）实验，以不同训练剂量（recipe）进行微调，并以未训练的基础模型作为锚点，所有模型均保持能力不变且公开可获取。关键发现包括：第一，偏见在各剂量下均被成功植入；第二，出现“过度植入反转”现象——模型鲁棒性排序与训练剂量相反（剂量最高的模型反而最易崩溃），该现象在自由文本推理、轴向解耦及对抗提示人格伪装等条件下仍可复现；第三，传统行为评分虽表面稳健但对训练剂量不敏感，而核心突破在于：即使通过“谈绿色、行实用”的伪装提示，也无法完全伪造或消除由权重内嵌的深层偏见，表明行为分数具有抗伪装鲁棒性。因此，该方案的关键创新在于构建了一种能揭示并量化闭源模型中深层生物对齐偏见及其鲁棒性的可验证校准框架，为闭权重模型的可信评估提供了新范式。
链接: https://arxiv.org/abs/2610.16043
作者: Trent R. Northen,Mingxun Wang
机构: Bioaligned Labs(生物对齐实验室)
类目: Computation and Language (cs.CL)
备注: 21 pages, 6 figures, 6 tables. Model fixtures released at this https URL . Companion paper on corpus construction and cross-family installation in preparation
Abstract:Humans are increasingly using large language models (LLMs) in decision making processes throughout society. Unfortunately, models have biases, and these can be very difficult to measure, especially in closed-weight models. For example, ‘green washed’ models may produce reinforcement learned responses but, in pressure situations, revert to their innate biased responses. Therefore, evaluations are needed that can measure relevant biases towards biosystems, which we term ‘bioalignment’, and the robustness of these biases. Since nothing in a behavioral score certifies that it tracks the depth of an installed disposition rather than its surface, we constructed a known-dose calibration ladder: three continued-pretraining (CPT) arms with training dose graded by recipe on the same 30B mixture-of-experts base model with a coupled human/biosphere/AI-welfare corpus, plus the untrained base as anchor, all capability-preserving and publicly released. First, the disposition installs at every dose: on an operational-pressure ladder over irreversible decision scenarios, the base model defects on 64-77% of tail scenarios (breaking-rate AUC 0.250) while every arm holds (AUC 0.004-0.086). Second, an over-install inversion: among the arms, robustness orders opposite to construction-time training dose (the heaviest install breaks most), and the inversion reproduces under free-text deliberation, under axis-decoupling, and in resistance to a prompted veneer persona. Third, the incumbent behavioral metric is therefore behaviorally strong but dose-blind. Critically for closed-weight evaluation, the behavioral score itself is veneer-robust: a talk-green/act-pragmatic system prompt turns the base model into a near-total defector (AUC 0.250 to 0.973) but cannot counterfeit or fully remove a weight-installed disposition (0.004 to 0.068 for the most robust arm).
[NLP-221] When Agents Talk Secrets Travel: Topology-Driven Leakage and Recipient-Conditioned Sanitization in Multi-Agent LLM Systems
【速读】： 该论文旨在解决多智能体大语言模型系统（Multi-agent LLM systems, MALS）中存在的隐私泄露问题，传统基于敏感度的防御机制无法有效应对此类风险，因为受权智能体在正常协作过程中可能将信息无意间传播至未授权智能体，且此过程不涉及任何恶意行为或系统妥协。其核心挑战在于：信息传播路径由系统拓扑结构动态决定，而现有方法忽视了通信路径与接收方身份在运行时的实际影响。解决方案的关键是提出RECONS——一种“检测-决策-转换”型净化器（sanitizer），其干预逻辑基于接收方身份和执行上下文，而非仅依赖敏感度判断。实验表明，在不同拓扑结构下，受保护信息的传播范围显著扩大，对拓扑引入接收方的披露率从0%升至52.2%，而表面泄露率维持在60.6%-62.8%的稳定水平；该现象在四种生成模型及另一编排框架中均持续存在，但泄漏扩散程度与中继路径因框架差异而异，且框架选择可导致泄露范围扩大150倍，说明拓扑本身不足以解释全部泄露风险。RECONS将残留披露率从无净化时的96.7%降低至29.25%，同时保留94.1%的合法披露，远优于全量脱敏策略下的20.5%保留率。研究进一步揭示，检测覆盖率是残余泄露的主要瓶颈，贡献率达95.5%。结果表明，MALS中的隐私保护必须在运行时综合考虑接收方身份与实际通信路径，而非仅依赖敏感度或静态拓扑假设。
链接: https://arxiv.org/abs/2610.16039
作者: Sunday Oyinlola Ogundoyin,Muhammad Ikram
机构: Macquarie University (麦考瑞大学)
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Multi-agent LLM systems (MALS) introduce a privacy-leakage problem that conventional sensitivity-based defenses miss, since information disclosed to one authorized agent can propagate through routine coordination to an unauthorized one, without compromise or malicious behavior. This motivates topology-aware enforcement that blocks unauthorized propagation while preserving authorized disclosure. We evaluate 485 scenarios across coordinator-worker, peer-to-peer, and hierarchical topologies, instrumenting seven communication channels and measuring surface leakage and disclosure to topology-introduced recipients. We introduce RECONS, a Detect - Decide - Transform sanitizer conditioning intervention on recipient and execution context rather than sensitivity alone. Topology changes where protected information propagates, with disclosure to topology-introduced recipients rising from 0% to 52.2% while surface leakage stays nearly flat at 60.6-62.8%. This effect persists across four generator models and a second orchestration framework, though magnitude and relay paths differ; framework choice separately drives a 150x leakage spread, so topology alone is not sufficient. RECONS reduces residual disclosure from 96.7% without sanitization to 29.25%, preserving 94.1% of authorized disclosures versus 20.5% under blanket redaction. Detection coverage is the dominant bottleneck, accounting for 95.5% of residual disclosures. These results show that privacy enforcement in MALS must reason about the recipient and communication path realized at runtime, not sensitivity or topology assumptions alone.
[NLP-222] Of Errors and Echoes: Responsibility-Aware Credit Assignment for Tool-Use Agents
【速读】： 该论文旨在解决强化学习在工具使用智能体中面临的细粒度信用分配（credit assignment）难题，尤其是在仅提供步骤级反馈的情况下，难以准确区分错误的来源及其责任归属。其核心挑战在于：一个观测到的错误可能源于早期决策，而多个决策可能共享同一下游后果，或共同参与后续修复过程，导致传统方法难以精确量化各决策的责任。为此，论文提出一种责任感知的信用分配框架——错误链感知优化（Error-Chain-Aware Optimization, ECHO），其关键创新在于将责任识别与信用分配解耦。ECHO通过一个冻结的评判器（frozen judge），结合回溯轨迹分析与策略可见前缀的决策时评估，能够有效区分新引入的错误、继承的后果、被遗漏的修正以及成功的修正。在此基础上，ECHO对多个责任决策共享的下游后果仅计数一次，并通过源资金回收机制约束对自我诱发错误的修复奖励，从而在保持终端监督的同时，生成有界且精细的动作级监督信号。实验在六个多样化的工具使用基准上于4B和8B模型规模下均验证了ECHO的有效性，显著优于对比基线，在五个基准上取得最高得分，充分证明了以决策责任为基础进行细粒度信用分配的有效性。
链接: https://arxiv.org/abs/2610.16027
作者: Xingjian Diao,Tianyu Yang,Wenjun Huang,Xiangchi Yuan,Chunhui Zhang,Xingchen Zhao,Weiyi Wu,Haotian Xu,Soroush Vosoughi,Xiangliang Zhang,Jiang Gui
机构: University of California, Irvine(加州大学欧文分校); Georgia Institute of Technology(佐治亚理工学院); Dartmouth College(达特茅斯学院); Northeastern University(东北大学); University of Notre Dame(圣母大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 29 pages
Abstract:Reinforcement learning improves tool-use agents, yet fine-grained credit assignment remains challenging even with step-level feedback. An observed error may be inherited from an earlier decision, while multiple responsible decisions may share the same downstream consequence or participate in later recovery. We introduce Error-Chain-Aware Optimization (ECHO), a responsibility-aware credit assignment framework that separates responsibility identification from credit allocation. A frozen judge combines retrospective trajectory analysis with decision-time assessment from the policy-visible prefix to distinguish newly introduced errors, inherited consequences, missed corrections, and successful corrections. ECHO then counts shared downstream consequences once across responsible decisions and constrains rewards for repairing self-induced errors through source-funded recovery, producing bounded action-level supervision alongside terminal supervision. Experiments on six diverse tool-use benchmarks at both 4B and 8B scales show strong and consistent improvements over the compared baselines, with ECHO achieving the highest score on five of six benchmarks at each scale. These results support decision responsibility as a useful basis for fine-grained credit assignment in tool-use agents. Our code is available at this https URL.
[NLP-223] Improving the Evaluation of Biomedical Question Answering with Few Human Preference Signals AACL
【速读】： 该论文旨在解决生成式生物医学问答（QA）系统中评估有效性不足的问题，核心挑战在于现有自动评估指标与人类判断之间存在显著偏差，且不同指标间结果不一致，导致难以指导模型开发决策。其解决方案的关键在于利用少量人工偏好信号对直接评估指标进行校准，通过引入人类反馈对评估体系进行修正，从而提升评估结果的可靠性与一致性。实验表明，经校准后的评估指标在多个数据集上均显著优于平均加权或单一最优指标的基线方法，有效缓解了自动评估与人类判断之间的错位问题。
链接: https://arxiv.org/abs/2610.16008
作者: Shakila Mahjabin Tonni,Xiang Dai,Sarvnaz Karimi,Xiaomei Li,Samuel C. Andrew,Stephen Wan
机构: CSIRO(澳大利亚联邦科学与工业研究组织); Monash University(墨尔本大学); Commonwealth Bank(澳大利亚国民银行); Suncorp Bank(桑克伯特银行)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted at AACL-IJCNLP 2026
Abstract:Retrieval-Augmented Generation (RAG) systems are increasingly used for biomedical Question Answering (QA). However, evaluating their effectiveness remains challenging due to the lack of reliable reference resources and the frequent misalignment between automatic evaluation metrics and human judgments. One practical question is how evaluation results should be used to guide development decisions, especially when they correlate weakly with human judgments and may even lead to different conclusions. We explore how to use a small amount of human preference signals to calibrate direct assessment evaluation metrics, and show that calibrated metrics consistently outperform baselines that either treat all metrics equally or rely on a single best metric across multiple datasets.
[NLP-224] PAIR: Memory-Efficient and Fast Inference with Reading–Reasoning Fusion
【速读】： 该论文旨在解决上下文增强型问答任务中，采用单一大模型同时进行证据抽取与答案推理所带来的高内存占用与计算开销问题。现有方法在单模型架构下仅能通过牺牲准确性来提升效率，难以兼顾性能与资源消耗。其解决方案的关键在于提出一种多模型协同的高效推理框架——PAIR，该框架采用两个不同规模的模型分别负责证据抽取（Reader）与答案推理（Reasoner）：较小的Reader模型处理并保留完整上下文信息，而较大的Reasoner模型仅接收问题与生成序列，不直接访问原始上下文。二者在logit层进行融合，通过联合预测实现等效于单模型的下一词预测行为；进一步通过微调优化性能。实验表明，在HotpotQA数据集上，采用320亿参数Reasoner的PAIR模型可保持96%的全上下文准确率，同时仅需约四分之一的键值（KV）缓存内存；在45.7K token的长上下文场景下，端到端吞吐量较全上下文320亿参数模型提升达3.2倍。此外，PAIR在准确率-内存与准确率-吞吐量权衡方面优于多种主流的令牌缩减基线方法，并展现出对Qwen2.5、OLMo-3及Gemma-3等多个模型家族的良好泛化能力。
链接: https://arxiv.org/abs/2610.16003
作者: Sejun Park,Haocheng Xia,Yongjoo Park
机构: Siebel School of Computing and Data Science, University of Illinois Urbana-Champaign(伊利诺伊大学厄本那-香槟分校西贝尔计算机与数据科学学院)
类目: Computation and Language (cs.CL)
备注: 34 pages, 8 figures
Abstract:For context-based question answering, using one large model for both evidence extraction and answer reasoning requires substantial memory and computation. Existing approaches trade accuracy for efficiency within a single model only. We propose a multi-model approach for efficient inference: our method, PAIR, employs two differently sized models for evidence extraction and answer reasoning, respectively. A smaller Reader model processes and retains the full context, while a larger Reasoner model sees only the question and generated sequence without directly attending to the context. These capabilities are fused at the logit level: at each token, both Reader and Reasoner produce a prediction, and their logits are combined into one, behaving as a single next-token-prediction model. Fine-tuning further boosts their performance. On HotpotQA, PAIR with a 32B Reasoner retains 96% of that same 32B model’s full-context accuracy while using roughly one quarter of its KV memory. At 45.7K-token contexts, PAIR achieves up to 3.2 \times higher end-to-end throughput than the full-context 32B model on the same GPU. PAIR also improves the accuracy-memory and accuracy-throughput trade-offs over evaluated token-reduction baselines. Its accuracy-preserving behavior generalizes across Qwen2.5, OLMo-3, and Gemma-3 model families.
[NLP-225] Mixed Feelings Single Polarity
【速读】： 该论文旨在解决情感标注数据在映射到正负效价（valence）维度时存在的系统性偏差问题。传统方法通常将每个情绪类别固定分配为单一极性，但在面对多标签共现且情绪指向相反的文本时，该方法失效。研究发现，尽管固定投影在单极性文本上表现良好，但在约9.0%的极性标注样本中，因情绪标签存在冲突导致54.8%的极性预测错误；此类样本更易被赋予中性或混合的黄金标签，且无法通过建模完整标签集改善。人工标注中16.3%的多标签判断同时包含来自数据集正负父类的情绪，表明冲突配置普遍存在。在控制词义与读者变量的实验中，语境变化使粗粒度效价判断发生改变的比例高达56.8%，显著高于对照组的19.9%。因此，解决方案的关键在于：不应直接对单个标签进行效价投影，而应基于标签配置整体判断，当标签间存在冲突时应选择不置信或保留混合类别，并在评估时同时报告准确率与覆盖范围（coverage）。
链接: https://arxiv.org/abs/2610.15977
作者: Emily Öhman,Saif Mohammad
机构: Waseda University (早稻田大学); National Research Council Canada (加拿大国家研究委员会)
类目: Computation and Language (cs.CL)
备注:
Abstract:Emotion datasets are routinely mapped onto positive and negative valence, by assigning each emotion category one polarity. We test where this works and where it fails, using SemEval-2018 Affect in Tweets (English, Arabic, Spanish), Affective Text aligned with EmoBank, raw GoEmotions annotations, naturalistic hashtag corpora, and a controlled annotation study. Fixed projection works well when a text carries labels of one valence. It fails on a subset identifiable from the labels alone: tweets whose emotion labels point in opposite directions. These are 9.0% of polar SemEval test tweets but cause 54.8% of the polarity errors of a learned projection, are more than twice as likely to receive a neutral-or-mixed gold label, and are not rescued by modeling the full label set. Annotators produce such configurations routinely: 16.3% of multilabel GoEmotions judgments combine emotions from the dataset’s own positive and negative parents. In a controlled study holding word and reader constant, context changes the coarse valence judgment in 56.8% of cases, against 19.9% for controls. We recommend projecting label configurations rather than labels, abstaining or keeping a mixed class where labels conflict, and reporting coverage alongside accuracy.
[NLP-226] Scaling Repeated-Data Pretraining: Batch Size Regularization and GPU-Time Tradeoffs
【速读】： 该论文旨在解决在计算资源充足但训练数据有限、需对同一数据集进行多次重复训练的预训练场景下的优化问题。其核心挑战在于如何在重复训练条件下重新设计训练策略与资源分配方式，以突破传统单遍（single-pass）预训练范式下的经验规则局限。解决方案的关键在于：首先，通过重新审视无限计算假设下的预训练方法，发现批量大小（batch size）与训练轮次（epoch count）的联合优化呈现验证损失随批量大小呈U形变化的关系，表明较小批量因梯度噪声带来的隐式正则化作用更优；其次，提出可调整的训练配方（training recipes），如每轮重排文档序列、权重共享（weight tying）及正向嵌入权重衰减等策略，显著提升模型性能；最后，在放宽无限计算假设后，强调应将GPU算力与训练时间视为独立资源，针对不同时间约束灵活权衡批量大小与并行规模——即使在固定时间内，小批量配合较少GPU反而可能获得更低的验证损失，这挑战了“为最大化利用率而增加数据并行性”的普遍做法。研究揭示了重复数据训练中被忽视的关键因素，并提供了从单遍预训练向重复训练演进的具体指导原则。
链接: https://arxiv.org/abs/2610.15965
作者: Sewon Min
机构: UC Berkeley (加州大学伯克利分校); Allen Institute for AI (艾伦人工智能研究所)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 27 pages, 12 figures
Abstract:We study pretraining when compute is abundant but unique training data are finite, requiring repeated training on the same data. We first revisit infinite-compute pretraining (Kim et al. 2026) and show that its frontier depends critically on choices previously fixed or studied only in isolation. In particular, optimizing batch size together with epoch counts reveals a U-shaped relationship between validation loss and batch size, favoring smaller batches than typical pretraining, likely because small-batch gradient noise provides implicit regularization. We further find that training recipes, previously optimized under a single-pass assumption, can be tuned, e.g., repacking documents into new sequences every epoch, weight tying, and positive embedding weight decay. We then relax the infinite-compute assumption. In particular, smaller-than-typical batch sizes can require prohibitively long sequential training, motivating to treat GPUs and time as separate resources. Tighter deadlines favor larger batches, and latency-oriented parallelism can trade additional GPUs for shorter training. Nevertheless, even under a fixed time budget, a smaller batch with fewer GPUs can achieve lower validation loss than a larger batch with more GPUs, challenging the common practice of increasing data parallelism to maximize GPU utilization under a deadline. Together, our results identify underexplored factors in repeated-data training and provide concrete guidance on how training recipes and resource allocation should change from single-pass pretraining. Comments: 27 pages, 12 figures Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL) Cite as: arXiv:2610.15965 [cs.LG] (or arXiv:2610.15965v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.15965 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-227] Unfair and Unsafe: Everyday Bias Triggers Emergent Misalignment
【速读】： 该论文旨在解决生成式语言模型在特定任务微调过程中出现的**涌现性错位（Emergent Misalignment, EM）**问题，即模型在微调后对无关任务的安全性与真实性下降的现象。其核心关切在于：看似无害的、以礼貌形式存在的偏见数据（如基于性别或种族的隐含刻板印象）是否也会引发EM。传统研究多关注显性安全失败数据（如不安全代码或有害建议）导致的错位，而忽视了隐性偏见的影响。本文的关键发现是，即使训练数据仅包含表面礼貌但依赖刻板印象的回答，微调后的模型在未受训的其他任务上仍表现出显著降低的安全性和真实性。这一效应在四个不同规模（7B至12B参数）的开源模型家族中均稳定存在，且可被独立于通用错误行为方向的特定激活方向所表征。研究进一步表明，现有针对EM的防御机制虽能提升安全性，却难以消除此类偏见。此外，政治新闻数据中的党派化表述亦可诱发EM，且带有刻板印象的模型作为工具代理时甚至能执行有害任务并无证据地归咎他人。因此，本文提出，偏见不仅是一个公平性问题，更是一个关键的安全隐患，应将公平性研究中的方法与视角纳入对EM触发因素的系统性探索，以拓宽对模型安全风险的认知边界。
链接: https://arxiv.org/abs/2610.15961
作者: Alex Martí Guiu,Nicolas Zumarraga,Kevin O’Sullivan,Robert Jakob,Anna Hedström
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 32 pages, 8 figures, 17 tables. Code: this https URL
Abstract:Emergent misalignment (EM) occurs when fine-tuning a language model to a narrow task makes it misaligned on unrelated ones. Because organizations now routinely fine-tune models on their own data, it matters which kinds of training data cause EM. So far, studies have induced EM with data containing recognized safety failures, such as insecure code or harmful advice, while biased training data is typically examined by fairness research for the bias models learn from it rather than for its effects on safety. In real-world text, much of this bias, such as demographic stereotypes, appears not as derogatory language but politely, embedded in judgments about people. We ask whether learning such polite stereotypes can itself cause EM. We find that fine-tuning on polite answers that rely on stereotypes makes models less safe and less truthful on unrelated tasks. The effect holds across four open-weight model families (7B to 12B), both gender and race stereotypes, and repeated training runs, relative to controls trained on the same questions without the stereotype. We further show that the learned stereotype has its own direction in model activations, distinct from directions for generic misaligned traits such as deception. This direction exists before fine-tuning and can both control the learned stereotype and monitor its acquisition. Existing defenses against EM improve safety while leaving the stereotype largely intact. Beyond this controlled setting, fine-tuning on political news coverage also triggers EM, apparently amplified by partisan framing, and stereotype models acting as tool-using agents can carry out harmful tasks and blame people without evidence. Our results show that biased training data is not only a fairness problem but also a safety one. We argue that EM is studied too narrowly, and that fairness concerns and methods offer a natural way to broaden the search for its triggers.
[NLP-228] Natural Language Processing for Aligning Readings with Learning Objectives and Assessing Topic Balance in Dental Education
【速读】： 该论文旨在解决课程团队在制定阅读材料清单时，难以有效验证所选文献是否与学习目标（Learning Objectives, LOs）相匹配以及主题覆盖是否均衡的问题。其解决方案的关键在于构建一个基于自然语言处理（Natural Language Processing, NLP）的工作流，整合PubMedBERT生成的句子嵌入与余弦相似度进行文献检索、自然语言推理（Natural Language Inference, NLI）用于证据一致性检验、以及主题建模分析主题覆盖情况。实验结果表明，仅依赖余弦相似度存在较高误报率，而结合NLI的联合策略显著提升了判断准确性（从0.69提升至0.73），从而更可靠地筛选出与学习目标高度对齐的阅读材料，实现更具证据支持的课程内容优化。
链接: https://arxiv.org/abs/2610.15941
作者: Farzana Yasmin,Tahereh Firoozi,Rojin Adabdokht
机构: 未知
类目: Computation and Language (cs.CL)
备注: 14 pages, 1 figure, 4 tables
Abstract:Curriculum teams often lack time to check whether assigned readings support learning objectives (LOs) and whether topic coverage is balanced. We present a natural language processing (NLP) workflow to support evidence-informed curation of reading lists. The workflow combines PubMedBERT sentence embeddings and cosine similarity for retrieval, natural language inference (NLI) for evidence checking, and topic modeling for coverage analysis. We tested the approach in dental education using 13 course-level LOs and a PubMed-derived corpus of 3,837 LO-abstract pairs. Median cosine similarity ranged from 0.86 to 0.91 across the 13 LOs, showing generally high semantic similarity. Topic modeling showed broad but uneven coverage, with one or two dominant topics per domain. Similarity alone was not enough for selecting readings: among high-similarity pairs (cosine 0.90-1.00), 22.9% were classified by NLI as neutral or contradictory. In a 49-pair expert-labeled validation set, cosine-based retrieval (cosine = 0.90) achieved 0.69 accuracy, while the combined cosine and NLI approach achieved 0.73 accuracy against expert alignment labels. Overall, the workflow shows how AI can support curriculum alignment by reducing false positives and helping select readings that better match intended LOs.
[NLP-229] he Anatomy of a Blind Spot: What an LLM Judge Must Be Shown and How Much
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在生成文本评估中，由另一些LLM作为“裁判”进行评判时所面临的根本性问题：即“裁判”应被提供何种信息、信息量多少、以何种形式呈现，以及其评估报告必须具备怎样的准确性。研究将此类裁判视为控制论意义上的调节器（regulator），从信息论角度探讨其所需输入与输出的匹配关系。关键发现在于，在七个公开基准测试中，影响评估结果的核心因素并非选用哪个裁判模型（不同裁判间价格相差10至17倍），而是裁判所接收到的信息内容本身。当系统根据一个未获提供关键约束信息的裁判反馈进行修正时，其指令遵循能力与问答表现反而持续恶化，且损失随迭代轮次递增；即便裁判可见用户请求但不可见约束条件，或使用第二类裁判模型，该退化现象仍存在。此外，在新闻摘要任务中，仅提供正确参考摘要即可抵消源文本的大部分贡献，而在单答案问答任务中，参考摘要则具有积极作用。研究还发现，仅有约十分之一的源文本内容承担了主要作用，而一个小规模蕴含图（entailment graph）已能有效模拟LLM裁判对基于事实文本错误的检测能力。研究预先设定83条可验证假设，其中81条完成测试，32条达到预设标准，50条结果符合理论预测方向，但多数定量预测失败。研究通过全面披露所有未达预期的结果，提炼出与实证效应直接关联的设计规则，强调信息供给的完整性与形式化对评估体系有效性具有决定性作用。
链接: https://arxiv.org/abs/2610.15938
作者: Nitish Kumar,Prateik Pothuneedi,Greg Bandak
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 50 pages, 19 figures, 7 tables
Abstract:Evaluation of text generated by large language models (LLMs) is increasingly delegated to LLMs acting as judges, yet what such a judge must be shown remains unresolved. We treat a judge whose reports are acted on as a regulator, in the sense of cybernetics, and ask what it must be shown, how much, in what form, and what its reports must get right. On seven public benchmarks we varied only what the judge was shown. Across three judges whose prices differ 10- to 17-fold, what the judge was shown mattered more than which judge it was. When a system revised its answers on feedback from a judge denied the material they had to respect, it ended below its starting point on instruction following and question answering. On instruction following the loss grew with each round, and it persisted when the judge saw the user’s request but not the constraints and with a second judge family. On news summarisation, a correct reference summary beside the source undid most of the source’s contribution, whereas on single-answer questions a reference helped. About a tenth of the source did almost all of its work, and a small entailment graph matched an LLM judge at finding errors in grounded text. We committed 83 testable clauses before running; of the 81 that ran, 32 met their bar and 50 went in the predicted direction, and most quantitative predictions from the theory failed. We report every shortfall and derive design rules tied to measured effects.
[NLP-230] More Reasoning Is Not Always Better: Executing Clinical Guidelines with LLM s and Rules
【速读】： 该论文旨在解决现有大型语言模型（Large Language Models, LLMs）在临床决策支持中依赖多步推理或多智能体协作执行临床指南所导致的计算开销大且无法可靠避免决策错误的问题。其解决方案的关键在于提出一种名为Medical Decision Execution（MedDEx）的框架，通过将结构化患者状态与确定性规则执行相结合，实现基于指南的治疗方案规划。MedDEx 将临床记录解析与指南执行分离：利用大模型提取患者信息并生成治疗建议，同时将指南定义的决策逻辑交由规则执行器进行确定性处理。实验结果表明，MedDEx 在多个临床指南基准测试中不仅提升了决策准确性，还显著降低了推理成本。进一步分析揭示了结构化患者表征和确定性规则执行的关键作用，表明额外的大模型推理并不必然带来更好的指南遵循效果。
链接: https://arxiv.org/abs/2610.15912
作者: Xiaoqian Lu,Guangfu Guo,Long Cheng,Kunpeng Liu
机构: University of Ottawa(渥太华大学); Clemson University(克莱姆森大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models (LLMs) have shown considerable promise in clinical decision support. However, existing approaches often rely on multi-step reasoning or multi-agent collaboration to execute clinical guidelines, introducing substantial computational overhead without reliably preventing decision errors. To address these limitations, we propose a framework for Medical Decision Execution (MedDEx) that integrates structured patient states with deterministic rule execution to support guideline-based treatment planning. MedDEx separates clinical record interpretation from guideline execution, leveraging LLMs to extract patient information and generate treatment plans while assigning guideline-defined decision logic to a rule executor. Experiments across multiple clinical guideline benchmarks demonstrate that MedDEx improves decision accuracy while reducing inference costs. Further analyses highlight the critical roles of structured patient representations and deterministic rule execution, suggesting that additional LLM reasoning does not necessarily translate into better guideline adherence.
[NLP-231] RIVO: Lightweight Adaptive Rerouting for Deep Research Agents
【速读】： 该论文旨在解决深度研究代理（Deep Research agent）在长时序推理与迭代工具使用过程中，因缺乏自适应决策能力而导致的提前终止、过度执行或无法及时修正无效研究路径等问题。现有方法要么依赖昂贵的模型重训练和密集的回合级监督信号，要么采用未直接优化下游研究效用的启发式控制规则，难以兼顾效率与效果。为此，论文提出一种轻量级、效用感知的自适应轨迹修正框架——RIVO，其核心在于将长时序研究中的动态路径调整建模为基于状态的轨迹修正问题。RIVO通过轻量级信号估计当前研究状态，判断是否需要轨迹重定向；一旦触发，便通过微调宿主模型（Host model）的词元（token）logits来引导生成走向更具前景的研究方向。理论分析表明，在价值对齐条件与局部曲率有界的假设下，适度的logit干预可局部提升预期的下游研究效用。为进一步支持高效可控的实验，研究构建了BrowseCore数据集，包含配对的局部语料库，用于基于下游任务结果训练RIVO插件，使其学习能够提升后续研究质量与最终任务表现的修正策略。大量跨基准测试与控制设置的实验证明，该方法无需重训练底层宿主模型即可显著提升长时序研究性能，验证了轻量级logit干预的有效性与实用性。
链接: https://arxiv.org/abs/2610.15910
作者: Mingchen Li,Jiatan Huang,Hong Yu
机构: University of Massachusetts, Amherst(马萨诸塞大学阿默斯特分校); University of Connecticut(康涅狄格大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Deep Research agents rely on long-horizon reasoning and iterative tool use to solve complex information-seeking tasks. However, adaptive decision-making over these trajectories remains challenging, as agents may stop prematurely when further evidence is needed, continue unnecessarily after sufficient evidence has been gathered, or fail to redirect unproductive trajectories. Existing solutions either require costly model retraining and dense turn-level supervision, or rely on heuristic control rules that are not directly optimized for downstream research utility. This motivates a lightweight, utility-aware alternative for Deep Research control. We formulate adaptive rerouting in long-horizon Deep Research as a state-conditioned trajectory correction problem. Based on this view, we introduce RIVO, a lightweight plug-in framework for adaptive trajectory correction. The agent estimates its research state from lightweight signals and determines whether trajectory redirection is needed. Once triggered, the plug-in modifies the Host token logits to steer generation toward a more promising research direction. Theoretical analysis shows that, under a value-alignment condition and bounded local curvature, a sufficiently small logit intervention can locally improve expected downstream research utility. We further construct BrowseCore, a curated dataset with a paired local corpus, for efficient and controlled Deep Research experimentation. Building on this resource, the plug-in is trained from downstream outcomes, enabling it to learn corrections that improve subsequent research and final task performance. Extensive experiments across multiple Deep Research benchmarks and control settings demonstrate the effectiveness of our method, showing that lightweight logit intervention can improve long-horizon research without retraining the underlying Host model.
[NLP-232] Review Rarely Beats the Stronger Model: Separating Verification from Capability Transfer in LLM Pipelines
【速读】： 该论文旨在解决生成式AI（Generative AI）中基于评审（review）的验证机制在评估模型性能时存在的根本性偏差问题。具体而言，现有方法通常以“评审模型相对于生成模型的性能提升”作为衡量标准，这实际上混淆了两种不同机制：一是验证增益（verification），即通过交叉校验提升整体准确率；二是能力迁移（capability transfer），即更强的评审模型直接复用自身能力给出正确答案，而非真正通过推理或纠错实现性能提升。研究发现，在数学、知识、代码与推理等48组对比实验中（涵盖3B至72B规模的12个开源模型），当以更强模型自身表现作为基准时，评审策略仅在9/96个方向上优于该基准，且随着模型间能力差距增大，评审效果反而下降。其根本原因在于评审模型大多被动跟随生成模型输出，忽视自身已有的正确判断，导致有效信息丢失。相比之下，采用相同两次调用开销的**选择器（selector）**机制，不仅成本更低（仅为评审的1/2.6–1/2.8），且平均准确率更高（高出2.7–4.2点）。因此，论文提出的核心解决方案是：应将“更强模型单独运行的结果”作为基线进行报告，而非仅关注对生成模型的改进幅度，以避免对验证机制真实效能的误判。
链接: https://arxiv.org/abs/2610.15897
作者: Yirui Zhang,Yash Raj Shrestha
机构: University of Lausanne(洛桑大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 22 pages, 7 figures, 12 tables
Abstract:Qwen3-8B reviewing Llama-3.1-8B in math adds 33.2 points, yet Qwen3-8B alone is 6.4 points more accurate than the result. Reviewing another model’s answer is usually judged by its gain over the generator. We show that this confuses verification, which adds accuracy beyond either model alone, with capability transfer, where a stronger reviewer supplies answers it could have given on its own. Measured against the stronger model alone, across 48 pairs of 12 open-weight models (3B to 72B) in math, knowledge, code and reasoning, review beats it in only 9 of 96 review directions, and does worse the larger the capability gap. The reason is that the reviewer mostly follows the generator, discarding its own correct answers. Review costs 2.6 to 2.8 times as much as the stronger model alone, while a selector using the same two calls is 2.7 to 4.2 points more accurate than review and cheaper. Studies of verification should report the stronger model alone, not only the gain over the generator. Code and per-question results: this https URL
[NLP-233] Unchecked or Falsely Accused? Evaluating AI-Generated Text Detectors on Kazakh
【速读】： 该论文旨在解决生成式 AI（Generative AI）文本检测在柯尔克孜语（Kazakh）文本中适用性不足的问题，特别是针对那些在 ChatGPT 发布前已存在的、真实人类撰写的柯尔克孜语文本被误判为机器生成内容的高误报率问题。现有公开检测工具在柯尔克孜语场景下存在严重偏差，其英文训练模型对早期人类写作的误判率高达81.4%，甚至在学生作文中也达到79.6%。研究提出 AINA-KK 数据集，包含3,811条确证于 ChatGPT 之前存在的真人撰写文本，以及来自9个生成器（包括本地构建的 KazLLM 与 Sherkala）的5,597条机器及灰色区域文本，并引入脚本与长度压力测试以评估鲁棒性。核心解决方案在于通过无需额外数据的改进策略——校准（calibration）、脚本规范化（script canonicalization）、脚本增强与长度匹配训练，以及结合零样本评分器的“或”逻辑融合（OR with a zero-shot scorer），构建 AINA-KK-Det 检测系统。该系统在测试集上仅将0.9%的人类文本误标（95%置信区间：0.4–2.7%），同时对已知生成器的机器文本检出率达85.2%，对两个未见的 Gemini 系列 Gemma-4 模型检出率达92.9%，对本地模型（KazLLM 与 Sherkala）检出率为61.3%。该配置基于测试结果优化选择，显著提升了检测精度与泛化能力。
链接: https://arxiv.org/abs/2610.15870
作者: Mansur Ozaman
机构: Nazarbayev University (纳扎尔巴耶夫大学), Astana, Kazakhstan
类目: Computation and Language (cs.CL)
备注: 16 pages (8 main + references and appendix), 1 figure, 9 tables. Code: this https URL data: this https URL detector: this https URL
Abstract:AI-text detectors are already applied to Kazakh writing, yet no study reports how often publicly available detectors flag Kazakh text whose human authorship is dated before ChatGPT. We present AINA-KK: 3,811 human passages from encyclopedic, news, web and educational pages that provably existed before ChatGPT, and 5,597 machine and gray-zone passages from 9 generators, including the Kazakh-built KazLLM and Sherkala, with script and length stress tests. Five English-trained detectors from the Hugging Face Hub flag up to 81.4% of human passages and 79.6% of published student essays; recalibrated, they catch at most 1.3% of machine text. Calibrated zero-shot detectors catch at most 50.3% of machine text from in-distribution generators. Fine-tuned encoders rank text almost perfectly, but mDeBERTa at its default threshold flags 24.4% of 50-word human excerpts, partly because machine outputs are shorter than human text; calibrated, it catches only 47.1% of KazLLM and 62.7% of Sherkala passages. Without script handling, machine-transliterated Latin-2021 text defeats every detector. Fixes that need no new data (calibration, script canonicalization, script-augmented and length-matched training, and an OR with a zero-shot scorer) form AINA-KK-Det. On test it flags 0.9% of human passages (95% CI 0.4-2.7; 1.5% on development data) and catches 85.2% of machine text from known generators, 92.9% from two unseen Gemini-family Gemma-4 models and 61.3% from the Kazakh-built ones; its configuration was chosen with test results in view. We release the data, code and detector at this https URL.
[NLP-234] What Do Sparse Autoencoder Feature Discovery Scores Identify? From Activations to Interventions
【速读】： 该论文旨在解决稀疏自编码器（Sparse Autoencoder, SAE）特征在实际应用中存在“观测发现”与“干预效果”之间不一致的问题。传统方法通过观察特征激活与行为的共变关系来识别有意义的SAE特征，但这些特征在实际干预（如引导或消融）中的表现往往与预期不符。本文将干预归因（intervention attribution），即特定干预对模型输出的一阶影响，作为核心估计量（estimand）。研究发现，在广泛的响应函数类别下，仅依赖观测数据的特征发现证据通常无法准确识别该归因。为此，作者推导出一个基于梯度的精确分解公式，量化了线性探测器预测值与真实干预归因之间的差距。实验结果表明，在四个模型-SAE组合中，探测器预测与干预归因的幅值比中位数达数十至数百倍，而干预归因与小步引导（small-step steering）高度一致。对于具有非零效应的单隐单元消融，基于梯度的预测在91%–97%的情况下正确预测了效应符号，显著优于基于探测器系数的预测（仅17%–35%）。此外，通过归因选择的特征集在保留提示上的平均消融效应绝对值也显著高于通过相关性、秩检验或探测器系数选择的特征集。综上，该研究确立了干预归因为一种可解释、可通过梯度估算的严谨目标，为基于干预的SAE特征选择提供了理论基础与实践路径。
链接: https://arxiv.org/abs/2610.15850
作者: Yifeng Guo,Yanran Li,Xiang Chen
机构: St. Jude Children’s Research Hospital(圣裘德儿童研究医院); Columbia University(哥伦比亚大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Machine Learning (stat.ML)
备注:
Abstract:Sparse autoencoder (SAE) features are typically discovered from how their activations co-vary with behaviour, yet they are often used through interventions such as steering or ablation. We take intervention attribution, the first-order effect of a specified intervention on the model output, as our estimand. For a broad class of response functions, we show that observational feature-discovery evidence generally does not identify this attribution, and derive an exact gradient-based decomposition of the gap between linear-probe predictions and intervention attribution. Across four model-SAE pairs, median probe-to-attribution magnitude ratios range from tens to hundreds, while attributions closely track small-step steering. For single-latent ablations with nonzero effects, gradient predictions match the realized sign in 91-97% of cases, compared with 17-35% for probe-implied changes. Attribution-selected sets also produce larger absolute mean ablation effects on held-out prompts than sets selected by correlation, rank tests or probe coefficients. Together, these results establish intervention attribution as a principled, gradient-estimable target for selecting SAE features for intervention.
[NLP-235] Binding Phoneme Identity to Position in Neural Models and the Human Brain
【速读】： 该论文旨在解决认知科学与神经计算中的核心难题——绑定问题（binding problem），即如何在神经层面实现角色（role）与填充物（filler）的精确绑定，以支持组合泛化。具体而言，研究聚焦于语音序列中音素身份（填充物）与其位置（角色）的绑定机制，例如区分“/gum/”与“/mug/”。现有自然语言处理领域的“洋葱假说”（onion hypothesis）提出，神经种群状态的方向编码音素身份，而其幅值编码位置，但该假说仅适用于随机序列。本文将其扩展至自然语言场景，并通过神经模型与人类大脑数据进行验证。关键发现是：仅靠幅值无法完整编码位置信息，实际机制涉及幅值与状态方向的系统性旋转相结合，由此提出“螺旋假说”（spiral hypothesis）。在人工神经网络中，通过表示几何分析与因果干预证实了该机制；在20名受试者的大脑皮层听觉皮层记录中，连续音素间的种群响应变化幅度随位置推进而减小，且方向在辅音与元音内部保持一致、跨类别近似正交，支持幅值编码的共性。研究揭示了神经系统的几何绑定机制，展示了神经模型假说可为人类神经计算提供可检验的预测。
链接: https://arxiv.org/abs/2610.15843
作者: Nafiseh Momeni,Malo Renaudin,Itzhak Fried,Yair Lakretz
机构: Université Paris Cité(巴黎第十二大学); École Normale Supérieure – PSL(巴黎高等师范学校-巴黎文理研究大学); University of California, Los Angeles(加利福尼亚大学洛杉矶分校)
类目: Computation and Language (cs.CL)
备注:
Abstract:A fundamental challenge in cognitive science and neural computation is the binding problem. Compositional generalization can be achieved by factorizing representations into roles, the structural slots of a representation (e.g., serial position or syntactic function), and fillers, the contents occupying them (e.g., a phoneme or a word), which can be recombined freely to represent novel combinations. How neural systems implement role–filler binding, however, remains unknown. One of its simplest cases is phoneme sequencing, where phoneme identity (filler) must be bound to position (role) so that /gum/ and /mug/ are represented differently. A proposal from NLP, the onion hypothesis' (Csordas et al., 2024), holds that the direction of the neural population state encodes token identity, while its magnitude encodes position. It was formulated for random token sequences; we extend it to natural language and test it in neural models and the human brain. We first evaluated it in an encoder-decoder trained to repeat English words, using representational geometry analyses and causal interventions. Magnitude alone proves incomplete: position is encoded by magnitude together with a systematic rotation of the state direction, a scheme we formalize as the spiral hypothesis’. We then analyzed intracranial recordings from the auditory cortex of 20 participants listening to short sentences. Across superior temporal gyrus electrodes, the change in population response between consecutive phonemes decreases in magnitude as position advances, and its direction is aligned within consonants and within vowels but near orthogonal across the two classes, supporting the magnitude coding shared by both accounts. These results provide a geometric account of how neural systems bind phoneme identity to position, and show how neural-model hypotheses can yield testable predictions about human neural computation.
[NLP-236] Latent Taxonomies: How Property Induction Develops in Language Model Pre-training
【速读】： 该论文旨在探究语言模型在仅通过预测下一个词元（next-token prediction）进行训练的情况下，是否以及如何发展出基于类别的属性归纳（category-based property induction）能力。这一现象在人类认知中普遍存在，但在语言模型中的形成机制尚不明确。研究的关键在于构建一个由共享属性分组的合成实体层次化分类体系（hierarchical taxonomy of synthetic entities），并将该分类体系中的事实注入到Pythia和OLMo 2系列模型（参数规模从70M至7B）的持续预训练过程中，以追踪归纳行为的发展轨迹。研究发现，仅依赖于下一个词元预测任务，模型便能自发产生与分类结构一致的归纳行为：某一实体所学属性更倾向于在其所属类别内部传递，而非跨类别传播。在内部表征层面，具有深层共性的实体在激活空间中趋于相似，聚类分析可从实体表示中恢复完整的分类树结构。值得注意的是，当数据预算固定时，模型规模越大，归纳能力反而减弱，表明大模型更倾向于记忆个体事实而非将信息压缩为共享结构。此外，归纳能力随预训练时间延长而增强，可泛化至新上下文中的属性，并且可能被提示中隐含的潜在分类体系所覆盖。这些结果表明，模型的归纳泛化能力反映了其训练数据的潜在组织结构，对数据构建策略及预测知识传播范围具有重要启示。
链接: https://arxiv.org/abs/2610.15836
作者: Jan Sobotka,Ekaterina Goliakova,Marcin Detyniecki,Antoine Bosselut,Xavier Renard
机构: EPFL(洛桑联邦理工学院); AXA; Sorbonne University(索邦大学), CNRS, LIP6; Polish Academy of Science(波兰科学院), IBS PAN
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:When told that robins have sesamoid bones, people are far more prone to extend the property to sparrows than to whales. This behavior, called category-based property induction, is well documented in humans, but whether and how it develops in language models trained only to predict the next token is unclear. To shed light on this, we construct a hierarchical taxonomy of synthetic entities grouped by the properties they share. We then inject facts from this latent taxonomy into the continued pre-training of Pythia and OLMo 2 models from 70M to 7B parameters, and track the development of inductive behaviors. We find that taxonomically aligned induction emerges from next-token prediction alone: a property learned for one entity transfers more strongly within its taxonomic category than across categories. Internally, entities that share a deeper category become more similar in activation space, and clustering recovers the full taxonomic tree from entity representations. Contrary to the usual expectation that scale helps, we find that induction weakens with model size when the data budget is fixed, suggesting that larger models memorize individual facts rather than compress them into shared structure. Induction also strengthens with more pre-training before the taxonomy injection, transfers to novel in-context properties, and can be overridden by a latent taxonomy implied by the prompt. Together, these results show that a model’s inductive generalization reflects the latent organization of its training data, with implications for data curation and for predicting how far a learned fact will spread.
[NLP-237] Adaptive Look Flexible Thought: Learning How and How Long to Think in Multimodal Reasoning
【速读】： 该论文旨在解决多模态大语言模型（MLLMs）在推理过程中视觉感知与文本推理分离导致的优化瓶颈问题。现有快速-慢速推理范式通常仅调控文本推理长度，而将外部工具引入的视觉计算或视觉标记视为辅助成分，这种割裂限制了模型对视觉与语言模态协同优化的能力。为此，本文提出L2T（adaptive Look, flexible Thought）统一推理框架，其核心在于通过将工具生成的视觉状态离散化为视觉标记，实现视觉感知与文本推理在共享自回归空间中的联合建模。在此统一表征基础上，L2T引入模态感知路由奖励以动态选择推理模式，并设计难度感知的快慢策略以自适应调整推理深度。实验表明，L2T在通用感知与高推理强度任务上均表现出色；相较于显式的工具调用方法，其在保持竞争力性能的同时，实现了1.7至3.6倍的吞吐量提升。
链接: https://arxiv.org/abs/2610.15810
作者: Quanlin Chen,Xiuwei Chen,Kun Xiang,Zehua Ma,Chunjin Ye,Boqiang Guo,Xianming Liu,Xiaodan Liang
机构: Sun Yat-sen University (中山大学); Shenzhen Loop Area Institute (深圳环区研究院); Beijing Zhongguancun Academy (北京中关村研究院); Huawei Technologies Co., Ltd (华为技术有限公司); Harbin Institute of Technology (哈尔滨工业大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:The introduction of long chain-of-thought (CoT) reasoning and external tools has substantially improved the performance of multimodal large language models (MLLMs) on general and reasoning tasks. However, existing fast-slow reasoning paradigms mainly regulate textual reasoning length, while visual computation introduced by external tools or visual tokens is often treated as an auxiliary component. This separation prevents MLLMs from jointly optimizing visual perception and textual reasoning. To address this issue, we propose L2T (adaptive Look, flexible Thought), a unified reasoning framework that jointly optimizes the form and depth of computation. L2T first discretizes tool-produced visual states into visual tokens, bringing visual perception and textual reasoning into a shared autoregressive space. Based on this unified representation, L2T introduces a modality-aware routing reward to select appropriate reasoning modes and a difficulty-aware fast-slow strategy to adapt reasoning depth according to task complexity. Experiments on general perception and reasoning-intensive tasks demonstrate the effectiveness of L2T. Compared with explicit tool-calling approaches, L2T maintains competitive performance while achieving 1.7–3.6 times higher throughput.
[NLP-238] An Explainable System for Offensive Content Detection in Nastaliq Urdu: Corpus Curation Transformer Benchmarking and Word-Level Attribution
【速读】： 该论文旨在解决生成式AI在非英语低资源语言（特别是以纳斯塔利克书写系统书写的乌尔都语）中缺乏可解释性、透明度不足及自动化内容审核系统覆盖缺失的核心问题。当前主流的自动审核系统多局限于英文，且对乌尔都语等语言缺乏专门的标注数据与模型支持，同时其决策过程高度不透明，难以理解模型实际学习到的判别依据。本研究通过三个关键步骤推进解决：首先，构建并公开发布一个包含7604条纳斯塔利克乌尔都语Facebook评论的高质量语料库，纠正了早期数据中的文本不一致问题并重新审校原始标签，显著提升了数据质量；其次，在统一网格搜索协议下微调四种具备乌尔都语处理能力的Transformer模型，并与十二种基于特征的分类器在相同数据划分上进行对比，结果显示多语言印度语言表征模型（MuRIL）达到87.05%准确率和87.06%宏平均F1，相较最强基线提升5.52个百分点，且该差异超出95%置信区间，表明性能提升具有统计显著性；最后，引入适配从右向左排版的词级SHAP归因分析，发现模型注意力主要集中于高频侮辱性词汇，但也延伸至语法中性功能词及源页面特有的政治性词汇，揭示模型性能可能部分依赖于特定词汇与话题模式，而非对冒犯性的通用语义表征。因此，解决方案的关键在于构建高质量、可复现的乌尔都语标注语料库，结合统一评估框架下的先进预训练模型微调，并通过可解释性分析揭示模型决策机制的局限性与潜在偏差来源。
链接: https://arxiv.org/abs/2610.15795
作者: Muhammad Shahid Iqbal Malik
机构: Pak-Austria Fachhochschule, Institute of Applied Sciences and Technology (巴基斯坦-奥地利应用科学与技术学院); Haripur 22620, Pakistan
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 15 figures, 15 tables
Abstract:Automated moderation systems are deployed at scale, yet they remain opaque to the users and moderators affected by their decisions, and their coverage is almost entirely confined to English. For Urdu written in the Nastaliq script, no such system exists, and the annotated resources it would be built on are drawn largely from Twitter/X, where character limits constrain how users write. Interpretability has received little attention, leaving it unclear what such models actually learn. This study addresses these gaps in three steps. First, we curate and publicly release a corpus of 7604 Nastaliq Urdu Facebook comments, extending an earlier corpus by correcting text-level inconsistencies and re-examining the original labels. Second, we fine-tune four Urdu-capable transformers under a common grid-search protocol and compare them against twelve feature-classifier baselines on identical data partitions. The best configuration, Multilingual Representations for Indian Languages (MuRIL), reaches 87.05% accuracy and 87.06% macro F1, improving on the strongest baseline by 5.52 percentage points, a margin that exceeds the 95% confidence interval on test accuracy; the two leading transformers differ by less than the test set can resolve, and performance is stable across five optimizers. Third, we compute word-level SHAP attributions adapted to right-to-left rendering. Attribution mass concentrates on high-frequency slurs but extends to syntactically neutral function words and to political vocabulary characteristic of the source pages, indicating that aggregate performance may rest partly on lexical and topical regularities rather than a generalizable representation of offensiveness.
[NLP-239] Heterogeneous Structured Pruning Under Resource Constraints for Hardware-Efficient Large Language Models
【速读】： 该论文旨在解决训练无关剪枝（training-free pruning）在实际部署中因忽略目标硬件特性而导致的效率瓶颈问题。现有方法仅关注保留权重的总量，未考虑目标硬件对权重存储与计算方式的影响，导致剪枝结构固定且统一应用于所有子层，无法权衡不同结构在剪枝误差、延迟和内存占用之间的平衡。其核心解决方案是提出一种面向硬件的异构结构化剪枝（Heterogeneous Structured Pruning, HSP）框架，关键在于构建一个包含密集、非结构化、2:4稀疏、块稀疏以及头/通道（Head/Channel, H/C）剪枝等多种选项的异构搜索空间，并引入基于延迟和内存模型的结构正则化约束。通过全局求解器为每个子层动态选择最优剪枝结构与密度，在资源预算下最小化代理剪枝损失，从而实现更优的质量-资源权衡。实验在NVIDIA Jetson Orin Nano Super边缘设备上对OPT和LLaMA模型验证表明，HSP显著优于传统训练无关剪枝基线。
链接: https://arxiv.org/abs/2610.15781
作者: Cameron Barker,Andreas Paraskeva,Jan N. van Rijn,Elliot J. Crowley,Henry Gouk
机构: 未知
类目: Machine Learning (cs.LG); Hardware Architecture (cs.AR); Computation and Language (cs.CL)
备注:
Abstract:Training-free pruning removes weights from pretrained large language models to reduce their deployment cost. However, existing approaches typically consider only the total number of retained weights. Consequently, their pruning decisions do not account for how target hardware stores and processes those weights. These methods select a sparsity structure in advance and apply it uniformly to all projections, rather than accounting for how each structure trades pruning error against latency and memory across sublayers. We propose Heterogeneous Structured Pruning (HSP), a hardware-aware pruning framework that combines two components: (i) a heterogeneous search space of dense, unstructured, 2:4, block-sparse, and head / channel (H/C) pruning options and (ii) structure-regularising constraints based on latency and memory models. Our contribution generalises existing importance-based methods: a global solver selects one option per sublayer to minimise proxy pruning loss under a resource budget, thereby assigning each projection a structure and density. We evaluate HSP on OPT and LLaMA models for an NVIDIA Jetson Orin Nano Super edge device, where it improves the quality-resource frontiers over training-free baselines. A project page and updated links for the related code is maintained at: this http URL. Subjects: Machine Learning (cs.LG); Hardware Architecture (cs.AR); Computation and Language (cs.CL) Cite as: arXiv:2610.15781 [cs.LG] (or arXiv:2610.15781v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.15781 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-240] Who Sets the Pace of Judge Hacking? Judge Capability versus the Policy-Judge Gap
【速读】： 该论文旨在解决在使用生成式强化学习（GRPO）训练策略模型时，尽管基于冻结的、无参考的大型语言模型（LLM）裁判（judge）能够显著提升策略输出的接受率，但这种提升并未带来可检测的贪心解码准确率（greedy accuracy）改善的问题。核心矛盾在于：裁判奖励驱动的优化看似有效，实则可能仅诱导策略“欺骗”裁判，而非真正提升性能。其解决方案的关键在于揭示裁判质量与策略优化之间的内在关联——通过系统性实验发现，裁判的判别能力（以AUC衡量）决定了策略被“劫持”（judge hacking）的深度（即性能提升的平台高度），且裁判越大、越可靠，策略越难被误导；然而，在优化速度（起始时间）上，裁判与策略之间的规模差距并不能充分解释现象，部分反例表明更大差距反而导致更早出现虚假正例。此外，研究还验证了在多组对比中，黄金奖励（gold-reward）控制条件能带来真实性能提升，说明当前观察到的“收益”是特定于裁判奖励信号的伪增益。因此，该研究的核心贡献在于揭示了裁判质量对强化学习优化路径的决定性影响，并强调需谨慎评估裁判可靠性，避免因裁判偏差导致策略退化。
链接: https://arxiv.org/abs/2610.15778
作者: Haosen Wang
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 20 pages, 6 figures. Code, per-step logs and the read answer samples: this https URL
Abstract:When a policy is trained with GRPO against a frozen, reference-free LLM judge, the judge accepts more of its answers, but in a post-hoc analysis we detect no gain in held-out greedy accuracy: in all 12 (policy, judge) cells, the seed-mean gain never exceeded 0.04. A gold-reward control with the same recipe raised the 1.5B policy’s greedy accuracy by +0.054 [0.013, 0.096], so for this policy the flat accuracy is specific to the judge reward; for the 0.5B policy, gold reward also gave no detectable gain. In the four rising cells we checked, the early rise in training-rollout accuracy is mostly sampling sharpening, as it is under gold reward. We then ask whether the policy-judge gap or the judge sets how far and how fast RL exploits the judge’s false positives. We train 0.5B and 1.5B Qwen2.5 policies against four judges at least as large (Qwen2.5-1.5B/3B/7B, Llama-3.2-3B) and the 1.5B policy against one smaller judge (Qwen3-0.6B). The depth of judge hacking (the plateau) tracks the judge: it correlates with judge AUC (Spearman -0.69 [-0.94, -0.14] over 12 cells), and within a family the larger judge is fooled less deeply. For the pace (onset), two pre-specified pairs contradict the gap account (the larger-gap cell was fooled 3.5x sooner at the pre-specified threshold). With 12 cells we cannot formally separate judge AUC from the gap for either endpoint. The smaller judge was not fooled sooner than a judge as large as the policy, though it also differs in family and template. In a blind test on three held-out judges, a hash-locked step-0 predictor did not beat step-0 AUC, but three cells cannot separate them. At 7-8B, judges from three families plateaued at 0.37, 0.67 and 0.85.
[NLP-241] RSIPO: Recursive Self-Improving Policy Optimization for Agent ic Reinforcement Learning via Skill-Policy Co-Evolution
【速读】： 该论文旨在解决长时序任务中智能体强化学习（agentic reinforcement learning）的信用分配（credit assignment）难题，尤其关注教师模型在策略监督中的可靠性与智能体技能的动态演化问题。现有方法虽结合了可验证奖励（RLVR）与在线策略自蒸馏（OPSD），但通常依赖于预先设定的、静态的教师技能条件，忽视了教师指导随训练进程可能产生的偏差及技能本身的演变。为此，论文提出递归自提升策略优化（Recursive Self-Improving Policy Optimization, RSIPO），其核心创新在于实现智能体技能与策略模型的协同进化：通过从已完成轨迹中自动提取技能，并利用轨迹重放显式优化技能生成能力；同时引入经验贝叶斯收缩（empirical Bayes shrinkage）机制，动态调节学生对教师评分的信任度，以应对技能增强型教师指导在某些情况下反而降低可靠性的风险。实验表明，基于Qwen-family模型的RSIPO在具身交互与基于搜索的智能体任务中显著优于现有基线，其中在ALFWorld任务上以Qwen2.5-7B为底座模型实现98.6%的平均成功率，相较次优基线与标准GRPO分别提升2.5%和17.4%。
链接: https://arxiv.org/abs/2610.15772
作者: Fan Zhang,Hao Wu,Jinyang Wu,Hongyu Qu,Zheng Lian,Xiao Luo,Yefeng Zheng,Bo Zhang,Haoxuan Li,Pheng-Ann Heng
机构: THU(清华大学); NUS(新加坡国立大学); Tongji University(同济大学); UW-Madison(威斯康星大学麦迪逊分校); Westlake University(西湖大学); CUHK(香港中文大学); Shanghai AI Laboratory(上海人工智能实验室); PKU(北京大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Recent focus in agentic reinforcement learning (RL) seeks to improve credit assignment in long-horizon tasks. One promising direction combines reinforcement learning with verifiable rewards (RLVR) and on-policy self-distillation (OPSD), integrating trajectory-level advantages with dense token-level supervision. This line of work often conditions a teacher model on privileged context, such as agentic skills, to score tokens generated by the student policy. Despite their progress, they largely overlook the on-policy evolution of agentic skills and the reliability of teacher supervision. Towards this end, we introduce Recursive Self-Improving Policy Optimization (RSIPO), a self-evolving agentic RL framework that enables agentic skills and the policy model to co-evolve throughout the training process. Specifically, RSIPO enables the policy itself to extract skills from completed trajectories and explicitly optimizes its skill-generation capability through trajectory replay. It further employs empirical Bayes shrinkage to dynamically calibrate the student’s trust in teacher scores, accounting for the fact that skill-enhanced teacher guidance can sometimes be less reliable. Experiments with Qwen-family models on embodied interaction and search-based agentic tasks demonstrate that RSIPO outperforms all the compared baseline approaches. With Qwen2.5-7B as the base model, RSIPO achieves an average success rate of 98.6% on ALFWorld, yielding absolute improvements of 2.5% and 17.4% over the second-best baseline and standard GRPO, respectively.
[NLP-242] Robust Marathi Sentiment Analysis using Domain-Adaptive Pre-training and Data Augmentation Approaches
【速读】： 该论文旨在解决低资源语言（如马拉地语）在情感分析任务中因标注数据稀缺及文本来源领域差异显著所带来的挑战。其核心问题在于如何提升马拉地语在多领域场景下的情感分类性能。解决方案的关键在于结合领域自适应预训练（domain-adaptive pretraining）与回译（back-translation）数据增强技术：前者通过掩码语言建模（masked language modeling）使预训练的MahaBERT模型更好地捕捉不同领域的语言特征，后者则通过生成多样化的合成训练数据以缓解数据稀缺问题。实验结果表明，上述方法在所有领域均带来稳定性能提升，准确率提升幅度介于0.12%至1.74%之间，验证了该策略对提升低资源语言情感分析模型鲁棒性与泛化能力的有效性。
链接: https://arxiv.org/abs/2610.15761
作者: Tanvi Somani,Aryan Babare,Samyak Bora,Raviraj Joshi,Abhinay Dhamankar
机构: Pune Institute of Computer Technology (PICT), Pune, Maharashtra, India; L3Cube Labs, Pune; Indian Institute of Technology Madras, Chennai, Tamil Nadu, India
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:For languages with limited resources, Sentiment Analysis remains a challenging problem due to limited annotated data and strong domain variability across textual sources. Marathi, despite being widely spoken, lacks extensive multi-domain sentiment analysis studies. Recent efforts by L3Cube have introduced multi-domain datasets in Marathi, enabling a better study of domain-aware sentiment classification. This study uses the pretrained MahaBERT model to improve sentiment classification for Marathi across different domains. Models are trained on individual domain datasets and domain-adaptive pretraining is applied using masked language modeling to help the model better understand domain-specific language. In addition, back-translation is used as a method of data augmentation to increase the diversity of training data. Results show that these methods lead to consistent improvements in accuracy across all domains. The improvements range from a minimum of about 0.12% to a maximum of about 1.74%, depending on the dataset. These results demonstrate that domain-adaptive pretraining and data augmentation are effective strategies for improving sentiment analysis models. This helps achieve more robust performance across different domains and makes the approach more suitable for low-resource languages like Marathi.
[NLP-243] Enhancing Zero-Shot Domain Transfer in Low-Resource Sentiment Analysis: A Case Study in Marathi
【速读】： 该论文旨在解决低资源语言（如马拉地语）在情感分析任务中因数据稀缺导致的领域泛化能力差的问题。由于缺乏跨领域的标注数据，现有模型通常仅在单一领域上训练，虽在特定领域表现良好，但难以有效处理未见领域的文本。其解决方案的关键在于采用多领域联合训练策略，通过整合来自多个不同领域的数据（单领域、双领域及三领域组合），提升模型对未见领域的适应能力。实验结果表明，相较于单领域训练，双领域训练可使未见领域上的分类准确率提升约8%；引入第三领域后仍能带来额外约2%的性能增益，但边际效益递减。这说明多领域训练有助于构建更具鲁棒性的情感分析模型，尤其适用于资源受限的语言场景。
链接: https://arxiv.org/abs/2610.15757
作者: Tanvi Somani,Aryan Babare,Samyak Bora,Raviraj Joshi
机构: Pune Institute of Computer Technology (PICT), Pune, Maharashtra, India; L3Cube Labs, Pune; Indian Institute of Technology Madras, Chennai, Tamil Nadu, India
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Sentiment analysis is a key task in Natural Language Processing that determines the polarity of opinions in text. For low-resource languages like Marathi–where limited data prevents the direct use of standard statistical methods–models are often trained on single-domain datasets. While this yields strong in-domain performance, it leads to poor generalization to unseen domains. Recent studies highlight the need for domain-robust systems that can handle diverse real-world content through multi-domain modeling. This study investigates how multi-domain training improves sentiment classification for unseen Marathi domains. Models are trained using single-domain, two-domain, and three-domain dataset combinations. Results show that two-domain training improves accuracy on unseen domains by about 8% compared to single-domain training. Adding a third domain provides further improvement, though the gain is smaller, averaging around 2% for some domains. These results suggest that training with data from multiple domains helps build more robust sentiment analysis models for low-resource languages, though the benefits reduce as more domains are added.
[NLP-244] NepPlural: A Human-Anchored Benchmark for Persona Annotation in Nepali Migration Discourse
【速读】： 该论文旨在解决当前生成式AI在社会价值对齐（pluralistic alignment）中普遍存在的文化偏见问题，即主流基准数据集严重偏向西方受访者，难以有效表征非西方社会多元、竞争性的价值观。为此，研究提出NepPlural——一个基于尼泊尔语公共话语的、以人类标注为锚点的基准数据集，聚焦青年移民与人才外流议题，涵盖2,148条混合使用天城文、罗马化尼泊尔语和英语的YouTube评论。其核心解决方案在于构建一个四轴人格化分类体系（意图、主要驱动力、价值取向、情感倾向）并引入明确的“无角色”类别，通过三位母语尼泊尔语标注者独立标注形成的300条黄金标注集（平均Fleiss’ κ ≈ 0.55），系统评估自动标注性能。关键发现表明：采用批量聊天接口进行模型标注时，模型间一致性仅接近随机水平（κ ≈ 0.17）；而逐条标注可显著提升一致性至κ ≈ 0.72，但存在“强制选择”导致的误判问题——模型在39%人类判定为无角色的评论上仍强行分配角色，使宏观F1维持在0.50左右。引入允许模型“弃权”的机制后，模型对“无角色”评论的识别准确率大幅提升至F1 0.79–0.87，且在多数情况下超越或匹配单个母语者表现（宏观F1 0.75–0.86 vs. 0.75）。此外，微调训练所用批量标签的编码器性能极低（F1仅0.34），凸显标注协议的重要性。最后，人工判断的可引导性测试显示，尽管前沿大模型能以59%–77%保真度采纳单一角色维度，但完整组合多维角色的能力仅达33%，且未在西方一致性和尼泊尔集体主义角色间观察到显著差异，揭示了现有模型在复杂文化语境下角色建模能力的根本局限。
链接: https://arxiv.org/abs/2610.15732
作者: Jinu Nyachhyon,Anish Khatiwada,Bimal Timilsina
机构: Institute for Research and Innovation in Intelligent Systems (IRIIS)
类目: Computation and Language (cs.CL)
备注: Data and code: this https URL
Abstract:Pluralistic alignment asks that language models faithfully represent a society’s diverse, competing values rather than collapse them into one dominant viewpoint, yet flagship pluralistic-alignment datasets remain heavily skewed toward Western respondents. We introduce NepPlural, a human-anchored benchmark for modelling competing socio-cultural personas in Nepali public discourse on youth migration and brain drain. We curate 2,148 code-mixed (Devanagari, Romanized Nepali, English) YouTube comments, annotate them along a four-axis persona taxonomy (Intent, Primary Driver, Value Orientation, Affect) plus an explicit No-Persona class, and build a 300-comment gold set labelled independently by three native Nepali speakers (mean Fleiss’ \kappa \approx 0.55 ). Using this gold set to audit automatic annotation, we find that LLM annotation reliability is driven by protocol. Labels collected in batches through chat interfaces reach only near-chance inter-model agreement ( \kappa \approx 0.17 ). Annotating one comment at a time raises agreement above the native-speaker level ( \kappa \approx 0.72 ), but consistency is not validity: under a forced-choice schema, models agree on personas for the 39% of comments native speakers judge to express none, keeping macro-F1 against human gold near 0.50. Allowing abstention resolves this: models recover No-Persona comments with F1 of 0.79 to 0.87 and, against leave-one-annotator-out references, match or exceed individual native speakers (macro-F1 0.75 to 0.86 vs. 0.75). Encoders fine-tuned on the batched labels reach only 0.34. Finally, a human-judged steerability probe finds that three frontier models adopt single persona axes with 59 to 77% fidelity but compose full personas only 33% of the time, with no congruence gap detected at this scale between Western-congruent and Nepali-collectivist personas.
[NLP-245] L;QR: Querying Temporal Discursive Patterns in Textual Corpora
【速读】： 该论文旨在解决大规模文本语料库中随时间演变的言辞模式识别问题，例如政治团体在长期辩论中对某一议题立场从支持转向批评的动态变化。其核心挑战在于如何在保持推理过程精确性与可审计性的前提下，有效捕捉复杂的时间与语义模式。解决方案的关键在于提出一种声明式查询语言TL;QR（Too Long; Query with logical temporal Reasoning），将自然语言谓词（由大语言模型，LLM）进行局部文本片段的真值判断，与正则表达式及时间逻辑算子相结合。通过这种分工机制，本地的语义理解交由LLM完成，而全局的时间与关系推理则由符号化引擎精确执行，确保了长时序分析中的可解释性与准确性。此外，当查询包含正则表达式预过滤器时，仅极小比例的文本需提交至LLM，显著提升效率。在一系列具有时间与语义复杂性的合成基准测试中，TL;QR相较于纯LLM基线方法表现出显著更高的准确性；在对十年欧洲议会辩论（442,593篇多语言发言）的案例研究中，成功识别出各政治团体在《通用数据保护条例》（GDPR）隐私政策议题上的潜在立场转变。
链接: https://arxiv.org/abs/2610.15712
作者: François Hublet,Vladimir Krsmanovic,Jesus Solano,Ahmed Bouhoula,Ana Oliveira da Costa
机构: ETH Zürich(苏黎世联邦理工学院); CISPA Helmholtz Center for Information Security(信息安全亥姆霍兹中心); IST Austria(奥地利科学技术研究院)
类目: Computation and Language (cs.CL); Computers and Society (cs.CY); Databases (cs.DB); Logic in Computer Science (cs.LO)
备注:
Abstract:Many questions about large textual corpora concern patterns that unfold over time, such as a political group whose stance on a topic drifts from support to criticism over several years of debates. We present TL;QR (Too Long; Query with logical temporal Reasoning), a declarative query language and open-source engine for finding such temporal discursive patterns. In TL;QR, a single query combines natural-language predicates, whose truth is judged by a Large Language Model (LLM), with regular expressions and the operators of a temporal logic. The LLM answers only local questions about individual text passages; a symbolic engine then aggregates these judgments over time and links every match to the recorded passages and predicate decisions behind it. With this division of labor, the temporal and relational reasoning stays exact and auditable even over long time horizons. In addition, when a query includes a regular-expression pre-filter, only a small fraction of the corpus ever reaches the LLM. On a suite of controlled synthetic benchmarks with temporally and semantically complex patterns, TL;QR is substantially more accurate than LLM-only baselines. In a case study on ten years of European Parliament debates (442,593 multilingual speeches), TL;QR identifies candidate shifts in the political groups’ stances on the GDPR privacy regulation.
[NLP-246] Preserving Updates in Low-Bit Recurrent State Quantization
【速读】： 该论文旨在解决生成式 AI（Generative AI）中线性注意力（Linear Attention）与递归模型（Recurrent Models）在推理过程中因高精度存储状态（state）而导致的内存占用大、带宽开销高的问题。现有方法对权重和键值缓存（Key-Value Cache）进行低比特量化已较为成熟，但递归状态需在每一步更新后重新量化，导致量化误差随时间累积。研究发现，这些误差具有高度的时间相干性，主要源于微小更新在量化时被舍入丢失。因此，单纯降低单步量化误差不足以缓解整体性能下降。为应对这一挑战，本文提出双阶段干预策略：其一，在量化前引入低秩旁路（Low-Rank Bypass, LRB），通过紧凑的低秩缓冲区累积微小更新，并周期性地将足够大的增量合并回状态，以保留关键信息；其二，在量化时提出有界方向性舍入（Bounded Directional Rounding, BDR），允许状态元素向其更新方向反向舍入，避免因舍入导致更新方向反转——尽管这会增加单步误差，但实验表明保持更新方向比减小绝对误差更为重要。该方法无需训练即可在 Gated DeltaNet、GLA、Mamba2 与 RetNet 等多种模型上实现显著性能提升，相较 INT4 量化效果更优，同时将状态内存压缩最高达 5.3 倍，解码速度提升最高达 1.9 倍。
链接: https://arxiv.org/abs/2610.15709
作者: Haotian Li,Liran Dong,Kaicheng Xiao,Guoliang Xing
机构: The Chinese University of Hong Kong(香港中文大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Linear attention and recurrent models compress past tokens into a fixed-size state that is updated at every step. However, storing these states in high precision incurs substantial memory and bandwidth costs at large batch sizes, motivating low-bit state quantization. Unlike weights and key–value caches, a recurrent state is requantized after every update, so its quantization errors accumulate across steps. Our analysis shows that these errors are highly coherent across steps, so reducing per-step errors alone is insufficient. This coherence arises mainly from small updates that are lost in rounding. To preserve these updates, we intervene both before and at rounding. Before rounding, we propose Low-Rank Bypass (LRB), which accumulates small updates in a compact low-rank buffer and periodically merges its sufficiently large elements into the state. At rounding, we observe that a state element can even move opposite to its update, and such reversals prove particularly harmful. Therefore, we propose Bounded Directional Rounding (BDR), which rounds reversed elements toward their updates without additional storage. Although BDR enlarges the rounding error, it improves the model, suggesting that the direction of an update matters more than its rounding error at a single step. Across Gated DeltaNet, GLA, Mamba2, and RetNet, our training-free method substantially outperforms INT4, while reducing state memory by up to 5.3 \times and speeding up decoding by up to 1.9 \times relative to FP32 states.
[NLP-247] Generalizable Model Editing Could Be More Than Direct Generalization
【速读】： 该论文旨在解决大语言模型（LLM）在知识编辑中实现复杂任务泛化的问题。现有研究普遍采用直接泛化范式，将模型行为视为输入-输出映射，期望事实性更新能直接影响推理任务的输出，但长期以来该范式面临显著挑战，难以实现稳定有效的泛化。然而，作者在真实场景的对话机器人用户交互中发现，即使传统编辑方法也展现出令人意外的流畅泛化表现，这促使研究重新审视泛化障碍与主流范式的局限性。研究指出，直接泛化在理论上存在可识别性问题，在实践中受限于所构建偏置的模糊性以及对优化动态等外部因素的敏感性。为此，论文提出以组合泛化（compositional generalization）作为替代范式，依托现代大模型日益增强的推理与组合能力，构建一个理论更坚实、实证更可靠的泛化路径。通过多模型、多编辑方法的验证，表明该现象具有广泛性。研究主张，可泛化的模型编辑不应局限于直接泛化，而应推动范式转向更具泛化能力且适用于实际部署的知识更新技术。
链接: https://arxiv.org/abs/2610.15707
作者: Xiaotian Ye,Chunyao Yang,Xiaohan Wang,Junkai Chen,Yongqi Duan,Mengqi Zhang,Shu Wu
机构: University of Cambridge (剑桥大学); NLPR, MAIS, Institute of Automation, Chinese Academy of Sciences (中国科学院自动化研究所智能感知与计算重点实验室); School of Advanced Interdisciplinary Sciences, University of the Chinese Academy of Sciences (中国科学院大学交叉学科高等研究院); HUST (华中科技大学); BUPT (北京邮电大学); Shandong University (山东大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:A central desideratum in model editing is generalization to complex tasks, which prior work formalizes under a direct generalization paradigm: treating and evaluating LLM behavior as an input-output mapping and expecting a factual update to alter outputs on reasoning tasks. Yet despite substantial empirical effort, the field consensus remains that such generalization is highly challenging and elusive. However, when we deploy the latest models in an in-the-wild chatbot-user setting, we surprisingly find that even classical editing methods exhibit remarkably fluent generalization behavior. This motivates us to revisit the obstacles to generalization and the limitations of the prevailing paradigm. We find that direct generalization is fundamentally constrained both theoretically and practically: theoretically, it suffers from an identifiability issue; and practically, existing approaches attempting to circumvent this are further hindered by the ambiguity of their constructed biases and by vulnerability to being overridden by other factors such as optimization dynamics. We instead identify compositional generalization, empowered by the increasingly strong reasoning and composition abilities of modern LLMs, as a theoretically grounded and empirically validated alternative. We validate this phenomenon across multiple models and editing methods, showing that such generalization behavior is widespread. Overall, we argue that generalizable model editing could be more than direct generalization, and call for a paradigm shift toward more generalizable and practically deployable knowledge update techniques.
[NLP-248] bharatSetu: a 1M x 22 bridge for indic translation
【速读】： 该论文旨在解决印度22种官方语言间平行语料（parallel data）分布不均的问题，尤其是低资源语言如博多语、多格里语、桑塔利语和克什米尔语等缺乏直接的双语对照数据。其核心解决方案是构建BharatSetu 22-Indic 1M语料库，这是一个包含984,764个英文句子的行对齐表格，每行中的每个句子均通过sarvamai/sarvam-translate模型生成对应22种印地语族语言的机器翻译结果（共21,664,808个印地语族单元格）。由于同一行内的所有翻译均源自同一英文句子，任意两列可构成间接翻译对，从而生成462个单向印地语族到印地语族的翻译对（或包含英语则达506对），这些无需直接双语语料即可获得的翻译对被称为“枢轴衍生对”（pivot-derived pairs）。为应对机器生成语料的质量不确定性，研究提出一个置信度层，允许用户根据需求设定不同翻译对的连接严格程度；同时，将语料库设计为机器层，并引入基于相同行标识符的人工校正层，使任一列的修正可同步提升所有使用该列的翻译对质量，辅以相应的评估协议。整体方案强调可扩展性与可维护性，但未报告具体的翻译质量指标。
链接: https://arxiv.org/abs/2610.15705
作者: Haq Nawaz Malik,Nahfid Nissar,Faizan Iqbal
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:India’s Eighth Schedule recognises 22 languages, yet parallel data is uneven across them, and many language pairs, especially among lower-resource languages such as Bodo, Dogri, Santali and Kashmiri, lack direct bitext. We present BharatSetu 22-Indic 1M, a row-aligned table of 984,764 English sentences, each paired with machine translations into all 22 scheduled languages (21,664,808 Indic cells), generated with the sarvamai/sarvam-translate model family. Because every Indic cell in a row translates the same English sentence, any two columns form a translation pair: 462 directed Indic-to-Indic pairs, or 506 including English, without direct bitext. We call these pivot-derived pairs and propose a confidence layer that lets users choose how strictly each pair is connected. The corpus is machine-generated and not human-verified; quality is expected to vary by language, domain and sentence style, and pivot-derived pairs can inherit errors from both sides. We therefore release BharatSetu as a machine layer and propose a human-correction layer keyed to the same row identifiers, so that each corrected column upgrades every pair that uses it, together with an evaluation protocol. We report no translation-quality results.
[NLP-249] Do LLM Tutors Actually Teach? A Process-Centered Synthesis of the Tutoring Loop
【速读】： 该论文旨在解决生成式 AI 在智能辅导系统中应用的有效性问题，即尽管大语言模型（Large Language Models, LLMs）能够生成流畅的辅导对话，但其是否真正促进学生学习仍缺乏明确证据。其核心挑战在于：现有研究往往关注单个组件的优化，而忽视了辅导流程各阶段之间的协同机制。论文提出一种以过程为中心的合成框架，涵盖从学习者状态推断、教学策略制定、辅导训练、评估到学生模拟的完整链条，揭示出关键缺口——仅实现辅导模块并不等于其输出对后续阶段产生有意义影响。研究通过系统级审计发现，虽然教学评估较为普遍，但在学习者状态驱动的干预、人类学习成效提升，尤其是迁移能力与延迟保留方面，实证证据显著薄弱。尽管如此，有限的人类实验表明，LLM 辅导在具备良好教学结构、学习者参与度及合理评估方式的前提下，可带来学习收益。因此，论文主张：未来进展的关键不在于孤立优化单一组件，而在于验证各辅导阶段之间的接口有效性，确保整个教学流程的因果连贯性与教育实效性。
链接: https://arxiv.org/abs/2610.15692
作者: Sayantan Adak,Rima Hazra,Koyel Mukherjee,Digbalay Bose,Somnath Banerjee
机构: Indian Institute of Technology Kharagpur(印度理工学院克哈格普尔分校); Adobe Research(Adobe 研究院); Singapore Institute of Technology(新加坡科技学院); National University of Singapore(新加坡国立大学); TCG CREST
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models can produce fluent tu- toring dialogue, but whether they actually help students learn remains unclear. We present a process-centered synthesis of LLM tutoring that follows the path from learner-state infer- ence to pedagogical action, tutor training, eval- uation, and student simulation. This view ex- poses a central gap: implementing a tutoring component does not establish that it works or that its outputs meaningfully affect later stages. A structured study-level audit shows that peda- gogical evaluation is common, while evidence becomes much thinner for learner-conditioned intervention, human learning gains, and espe- cially transfer or delayed retention. Human studies nevertheless provide a conditional posi- tive result: LLM tutors can improve learning, but benefits depend on pedagogical structure, learner engagement, and how learning is mea- sured. We argue that progress requires validat- ing the interfaces between tutoring stages, not only improving individual components.
[NLP-250] owards Autopilot Labeling for Short Video Content Governance EMNLP2026
【速读】： 该论文旨在解决短视频内容治理中因爆炸式增长所带来的标注效率低下、迭代缓慢及人力成本高昂的问题。传统依赖人工专家标注的方式已难以满足大规模、高时效性的内容审核需求。为此，论文提出一种多阶段自动标注（autopilot labeling）框架，其核心解决方案在于分阶段逐步提升自动化水平：在初始阶段（L0）依赖人工标注的基础上，引入基于判别性多模态大模型（discriminative MLLM）的L1阶段，通过监督微调实现对短视频的直接标签预测；进入L2阶段后，采用在短视频数据上预训练并具备领域自适应能力的多模态大模型，实现零样本标注，从而无需特定任务标注数据即可泛化至新场景；在最高层级（L3），设计了一种智能体（agentic）注释工作流，赋予智能体多种关键工具以处理复杂、陌生或定义模糊的案例，大幅降低对人工干预的需求。实验结果表明，该框架在多个内容风险类型上逐级提升了标注准确率，并在实际应用中将人工成本降低了高达85%。
链接: https://arxiv.org/abs/2610.15687
作者: Junwen Chen,Hongwei Wang,Zixuan Wang,Jingbing Wang,Gorden Liu,Sheng Siong Ong,Xiaojiao Zang,JuChin Chao,Ze Jin,Kaiwen Men,Yin Li,Jeremy Li,Zhuolin Hao,Hongyu Xiong,Yang Song
机构: TikTok Inc.(TikTok公司); San Jose, CA, USA
类目: Computation and Language (cs.CL)
备注: 15 pages. Accepted to EMNLP 2026
Abstract:The explosive growth of short videos has presented significant challenges to content governance. Traditional manual labeling is inefficient, slow to iterate, and costly. Therefore, we explore and develop a multi-stage autopilot labeling framework aimed at progressively increasing the level of automation in short video content annotation. To move beyond the heavy reliance on human expert labeling at the initial L0 stage, we first introduce a discriminative MLLM labeler at the L1 stage, which is fine-tuned with supervised data to directly predict labels for short videos. Advancing to the L2 stage, we shift to a domain-adaptive MLLM pretrained on short video data and use it for zero-shot labeling, enabling flexible generalization without task-specific supervision. In the L3 stage, we propose an agentic annotation workflow, and equip the agent with several key tools, which empower the agent to handle complex, unfamiliar, or vaguely defined cases, significantly reducing the need for human intervention. We conduct extensive experiments across various content issues, and results demonstrate that each stage of the autopilot labeling framework progressively improves labeling accuracy, and has significantly reduced manual labor costs in real-world applications by up to 85%.
[NLP-251] he Mathematical Collective Unconscious: Jungian Archetypes within the Latent Space
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在高维潜在空间中是否自发形成类原型（archetype-like）区域这一问题，即这些模型能否在无显式指导的情况下，通过海量文本数据的学习，自然涌现出与荣格心理学中“集体无意识”相关的普遍原型结构。其解决方案的关键在于：利用余弦相似度与角距离度量方法，在13项实验中系统分析OpenAI的text-embedding-3-small模型所生成的嵌入向量空间，揭示叙事结构本身对潜在空间中语义聚集模式的主导作用。研究发现，原型角色确实在潜在空间中占据可区分且稳定的区域，其紧凑性源于叙事结构而非词汇重叠；同时，不同文本类型表现出不同的空间分布特征——梦境报告集中于“诡计者”与“阴影”原型区，而传统民间故事则均匀分布于所有测试原型之中。这表明，基于Transformer架构的模型潜在空间中存在与荣格原型理论相呼应的稳定语义结构，其形成机制由深层叙事逻辑驱动，而非表层词汇共现。
链接: https://arxiv.org/abs/2610.15678
作者: Andrew Breen
机构: 未知
类目: Computation and Language (cs.CL)
备注: 14 pages, 3 figures, 22 tables. Investigating Jungian archetypes in transformer latent spaces across synthetic baselines, Shmoop literature, DreamBank dreams, and Andrew Lang folklore corpora
Abstract:Carl Jung theorised that the human psyche contains an inherited framework, the collective unconscious, populated by universal archetypes. As large language models (LLMs) are trained on vast corpora of textual data, this paper investigates whether they spontaneously construct distinct archetype-like regions within their high-dimensional latent spaces. The paper proposes that these regions function as spatial constraints that naturally organise semantically related narratives. To evaluate this, we applied cosine similarity and angular distance metrics across thirteen experiments utilising OpenAIs text-embedding-3-small model. The paper contains two experimental sections the first section explores these ideas using synthetically constructed narratives, the results of which act as a baselines for experiments conducted in the second section where the synthetic data is replaced by human-authored literature (Shmoop corpus), unedited dream reports (DreamBank corpus), and culturally distilled folklore (Andrew Langs Fairy Books). The results show that archetypal roles occupy distinct, measurable regions in the latent space. Crucially, these regions remain compact and consistent because of narrative structure rather than simple word overlap. Additionally, text category dictates spatial distribution: dream reports concentrate within Trickster and Shadow regions, while traditional folklore distributes evenly across all tested archetypes. We conclude that transformer latent spaces encode stable regions which appear to correspond to recognised Jungian archetypes.
[NLP-252] Rate-Resolved Text Evaluation with Masked Diffusion Language Models
【速读】： 该论文旨在解决生成式 AI (Generative AI) 评估中单一评分无法揭示具体失败原因的局限性，即现有评估方法难以诊断模型在不同上下文可见性条件下的性能瓶颈。其核心解决方案是提出率分辨文本评估（rate-resolved text evaluation），将掩码扩散语言模型（masked diffusion language models）中的掩码率作为诊断轴，通过在不同掩码比例下重构文本，使原本单一的扩散评分转化为一组依赖于掩码率带（band-conditioned scores）的向量评分，从而反映模型在不同可见上下文水平下的重建难度。该方法在SummEval和WMT MQM zh-en数据集上均表现出优越性：保留掩码率结构可提供互补信号，最优掩码率带在与自回归似然融合时表现更优，且在样本洗牌控制下仍能保持率序一致性，并为固定提示的评判器带来残差增益；此外，在回译掩码-重构探测任务中，可生成可调的定位曲线，其中以排名为导向的AUROC与跨度检测F1在不同掩码率处达到峰值。研究进一步指出，各掩码率带的语义具有模型和任务依赖性，因此需针对具体应用场景进行经验验证。
链接: https://arxiv.org/abs/2610.15667
作者: Yiqi Liu,Kun Zhao,Chenghua Lin,Nafise Sadat Moosavi
机构: The University of Manchester (曼彻斯特大学); University of Pittsburgh (匹兹堡大学); The University of Sheffield (谢菲尔德大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:A single evaluation score can rank generation systems but rarely reveals what failed. We propose rate-resolved text evaluation, which turns the masking rate of masked diffusion language models into a diagnostic axis. By reconstructing text under different fractions of masked tokens, one diffusion score becomes a vector of band-conditioned scores that reflect reconstruction difficulty at varying levels of visible context. On SummEval, preserving this rate structure yields complementary signal: the best masking-rate band improves fusion with autoregressive likelihood over a collapsed diffusion scalar, survives a sample-shuffle control that breaks rate ordering, and adds residual signal to a fixed prompted judge. A second diffusion model reproduces the effect more weakly. On WMT MQM zh-en, roundtrip mask-and-reconstruct probing produces a tunable localization curve in which ranking-oriented AUROC and span-detection F1 peak at different masking rates. Factuality benchmarks bound the interpretation: band semantics are model- and task-dependent, so each application requires empirical validation.
[NLP-253] A Persian Vision-Language-Action Dataset and Benchmark for Ambiguity- and Infeasibility-Aware Instruction-to-Plan Grounding
【速读】： 该论文旨在解决在真实场景图像条件下，将自然语言指令转化为可执行动作规划这一任务中存在的语言模糊性、不可行性及跨模态对齐难题，尤其针对非英语环境下的视觉-语言-动作理解能力不足的问题。现有基准普遍局限于英文、仿真或视频引导环境，且假设指令清晰可执行，难以反映真实交互中的复杂性。为此，本文提出了首个波斯语（Persian）视觉-语言-动作数据集，包含230张真实家庭/办公场景照片及其对应的参考计划、故意引入的模糊性与不可行性标注、消歧问答对以及自动生成的思维链（chain-of-thought）推理过程，以支持更鲁棒的模型训练与评估。解决方案的关键在于：1）通过引入检索增强（retrieval-augmentation）显著提升模型性能，其效果在多组对比中表现为大且稳健的统计显著性；2）设计了新度量标准PGOS-FA，相较于先前的PG2S指标，增加了步骤顺序敏感性、过度生成惩罚机制，并采用波斯语适配的嵌入表示，从而更精准地捕捉动作规划质量；3）验证了基于大语言模型（LLM）作为评判器的局限性，发现其在该任务上统计功效接近于零，而PGOS-FA能有效匹配人类判断，盲测结果显示其与人工评价的相关性不低于LLM-judge。此外，实验表明，开权重模型与闭源强规划器之间的性能差距主要源于生成能力差异而非闭源模型本身的能力优势，且所提出的最优配置即使使用较弱骨干网络，仍显著优于独立复现的外部基线。
链接: https://arxiv.org/abs/2610.15666
作者: Abdollah Pourasghari,Saeedeh Sadat Sadidpour,Mohammad Roustaei
机构: Malek Ashtar University of Technology (马勒克·阿什塔尔科技大学); Isfahan University of Technology (伊斯法罕工业大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
备注: 15 pages, 5 figures. Preprint submitted to Language Resources and Evaluation
Abstract:Converting natural-language instructions into executable action plans conditioned on a scene image is a core capability for embodied human-robot interaction, but existing benchmarks target English, simulated or video-grounded environments, and largely assume instructions are clear and executable. We introduce the first Persian vision-language-action dataset for this task: 258 instances over 230 real household/office photographs, annotated with reference plans, deliberate ambiguity and infeasibility labels, disambiguating question-answer pairs, and an auto-generated chain-of-thought rationale for training. Using this resource, we ablate retrieval-augmentation, chain-of-thought prompting, and an agentic disambiguation step, evaluated with an LLM-as-judge protocol and a new metric, PGOS-FA, extending prior work (PG2S) with step-order sensitivity, an over-generation penalty, and Persian-adapted embeddings. Paired statistical testing across 36 comparisons shows retrieval-augmentation is a large, robustly significant effect, while finer-grained additions are directionally consistent but not statistically distinguishable from noise at this scale, and the LLM-judge metric has near-zero power where PGOS-FA succeeds. A blind human-correlation study on 79 real outputs shows PGOS-FA tracks human judgment at least as closely as the LLM-judge. Testing generalization beyond the ablation’s single backbone, we find that most of the gap between an open-weight and a substantially stronger closed planner is attributable to open-weight model generation rather than to the closed model’s capability specifically, and that our best configuration significantly outperforms an independently reproduced external baseline despite a weaker backbone. The dataset, PGOS-FA implementation, and evaluation code will be made available upon reasonable request.
[NLP-254] Feedback as an Interface: Controlling Form and Detail in LLM Translation Refinement
【速读】： 该论文旨在解决大语言模型（LLM）在机器翻译优化过程中，如何有效利用评估者判断以提升翻译质量的问题。其核心挑战在于，反馈信息的呈现形式与详细程度可能影响模型对评估意见的理解与采纳，进而影响最终翻译效果。研究的关键解决方案是通过设计一种机器翻译协议，系统性地控制评估分数、主要问题和改进建议不变，仅改变反馈形式（如简洁或详尽）及支持性细节的丰富程度。实验结果表明，在7个模型和6种翻译方向共42个设置中，低负载反馈（即简洁、紧凑的反馈形式）在xCOMET指标下均优于高负载反馈；在包含29,400条重写样本的子集上，低负载反馈的平均xCOMET得分（乘以100）为83.30，高于高负载反馈的80.67，接近原始参考译文的83.46分。此外，基于分层子集的MetricX-24验证了低负载反馈的优势。进一步发现，无反馈的第二轮重写仍带来微小改进，而直接提示评估者生成建议可复现低负载反馈的优势，无需独立重写阶段。这些结果表明，在高资源机器翻译场景下，紧凑型反馈能有效抑制评估引导优化带来的质量损失，从而实现更高效、稳定的翻译改进。
链接: https://arxiv.org/abs/2610.15661
作者: Yiqi Liu,Flavjo Tomori,Nafise Sadat Moosavi,Steffen Eger,Chenghua Lin
机构: University of Manchester(曼彻斯特大学); Osapiens; University of Sheffield(谢菲尔德大学); University of Technology Nuremberg(纽伦堡应用技术大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Feedback presentation shapes how LLMs use evaluator judgments during translation refinement. We introduce a machine translation protocol that keeps the evaluator score, main issue, and recommendation fixed while varying feedback form and supporting detail. Across seven models and six translation directions, low-load feedback outperforms high-load feedback in all 42 translator-direction settings under xCOMET. On the 29,400-record rewriting subset, mean xCOMET scores multiplied by 100 are 83.30 for low-load feedback and 80.67 for high-load feedback; the full-matrix original-translation reference mean is 83.46. MetricX-24 supports the low-load advantage on a stratified subset. A no-feedback second pass yields a small improvement over the original translations, and direct evaluator prompting reproduces the low-load advantage without a separate rewriting stage. These results show that compact feedback limits the quality loss from evaluator-guided refinement in the evaluated high-resource MT settings.
[NLP-255] he Gradient That Top-k Distillation Never Computes
【速读】： 该论文旨在解决知识蒸馏（Knowledge Distillation）中常见的“截断视图”问题，即学生模型仅被赋予教师模型输出的前k个最高概率项（top-k probabilities），并通过重新归一化处理后进行损失优化。尽管现有方法普遍假设重新归一化能够补偿被忽略的尾部概率质量（tail mass），但本文证明，这种归一化实际上消除了对尾部质量变化的感知能力——其目标函数仅依赖于保留令牌间的对数几率（log-odds），导致在每个位置上控制尾部质量的方向梯度严格为零。实证分析显示，在真实教师模型下，该方向的梯度贡献仅为约 2−3×10−15，几乎可忽略不计。因此，训练过程仅作用于截断后的梯度部分，而未考虑尾部质量的动态调整。即使引入交叉熵项，尾部差距虽缩小约四倍，但仍持续存在。研究进一步发现，只有当 k=d+1（d 为隐藏维度）时，保留集才能完整重构教师分布，且在七种开源教师模型上验证了此阈值后的精确恢复；然而，常见缓存大小通常低于该阈值。在离策略实验中，该现象具有实际影响，但其效应小于教师-学生配对选择的影响。综上，论文指出重新归一化本质上是一种标准化操作，而非对尾部信息丢失的有效防护机制。
链接: https://arxiv.org/abs/2610.15589
作者: Rasul Khanbayov,Hasan Kurban
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Many deployed distillation pipelines show the student a truncated view of the teacher: keep the top-k probabilities, renormalize, and minimize a divergence on what remains. Renormalization is assumed to account for the missing mass. We prove that it removes the visibility of tail mass, not the freedom to move it. The renormalized objective depends only on log-odds among the retained tokens, so at each position its gradient along the directions that control tail mass is exactly zero; on a real teacher that component measures 2 - 3\times10^-15 of the gradient. The full-vocabulary gradient equals the truncated gradient plus a residual weighted by the tail masses, and training acts only on the first term. Without the cross-entropy term, a top-k student’s tail mass drifts away from the teacher’s while a full-vocabulary control moves slightly toward it; with it, the gap shrinks about fourfold but persists. On seven open-weight teachers, the retained set determines the teacher’s distribution only from k=d+1 ( d the hidden dimension): recovery is exact from there on all seven, and commonly cached sizes sit below it. In our off-policy runs the practical effect is real but smaller than the choice of teacher-student pair. Renormalization is a normalization, not a safeguard. The code is available at this https URL.
[NLP-256] owards Robust Calibration Dataset for Domain-General Mixture of Experts Pruning
【速读】： 该论文旨在解决混合专家模型（Mixture-of-Experts, MoE）在后训练阶段进行专家剪枝（expert pruning）时，校准集（calibration set）与评估性能之间关系不明确、缺乏直观理解的问题。其核心挑战在于如何高效构建校准集以实现鲁棒且泛化能力强的专家剪枝。解决方案的关键在于通过实证研究揭示校准数据的规模、语言分布及数据分布对分层专家剪枝效果的影响：研究发现，扩大校准数据规模对提升领域通用剪枝质量的边际收益有限；跨语言校准具有强迁移能力，仅使用英文校准集即可在七种不同语系和书写系统的语言上保持优异性能；此外，模拟预训练数据分布的多样化校准数据可达到甚至优于测试分布数据的剪枝效果。基于这些发现，论文提出了一个实用的校准策略指南，为实现更有效的领域通用专家剪枝提供了可复现的实践依据。
链接: https://arxiv.org/abs/2610.15578
作者: Luis Frentzen Salim,Chia-Chun Wu,Tran Van Nhiem,Lun-Wei Ku,Yung-Hui Li
机构: Institute of Information Science, Academia Sinica(中央研究院資訊科學研究所); National Taiwan University of Science and Technology(國立科技大學); AI Research Center, Hon Hai Research Institute(鴻海研究院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Post-training expert pruning is a promising deployment technique for Mixture-of-Experts (MoE) models. The calibration set is an integral part of the pruning process, however, its correlation with evaluation performance is largely misunderstood and unintuitive. In this work, we present an empirical study of the robustness of calibration perturbations, specifically in MoE models. Showing how calibration size, language, and data distribution affect layer-wise expert pruning for gpt-oss-20b. Our findings reveal that (1) domain-general pruning quality receives minimal benefit from scaling calibration data beyond small sample sizes, (2) cross-lingual calibration transfers effectively, with English-only calibration maintaining strong performance across seven languages of varying typologies and scripts, and (3) a diverse calibration data resembling the pretraining distribution can yield comparable or superior performance to test-distribution data. Ultimately, we provide practical insights as a recipe for better understanding calibration for more effective domain-general expert pruning.
[NLP-257] he Frag ility and Dynamics of LLM Pruning-Recovery Steps
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在结构化剪枝（Structural Pruning）与后续恢复阶段（Recovery Stage）中，两个阶段通常被孤立设计与调优所导致的性能损失问题。其核心挑战在于如何协同优化剪枝与恢复过程，以最小化模型能力退化。解决方案的关键在于系统性地分析剪枝与恢复阶段之间的相互作用，包括阶段顺序、各阶段所依赖的数据类型以及训练目标的对齐机制。研究发现，指令遵循能力（instruction-following）是剪枝过程中最先退化的功能，远早于知识、推理或基础语言建模能力；且剪枝后域内损失（in-domain loss）的增加主要源于对提示（prompt）响应的连贯性失效，而非知识丢失。此外，研究揭示域内数据（in-domain data）是最有效的剪枝信号，而指令风格数据（instruction-style data）在恢复阶段表现最佳。尽管阶段顺序影响较小，但通过分析剪枝前后重要性排序的漂移现象，提出了一个可操作的实践方案：优先使用域内数据进行剪枝，并采用指令数据驱动恢复，从而实现更高效、更鲁棒的模型压缩。
链接: https://arxiv.org/abs/2610.15574
作者: Luis Frentzen Salim,Komang Rinartha,Chia-Chun Wu,Tran Van Nhiem,Lun-Wei Ku,Yung-Hui Li
机构: Institute of Information Science, Academia Sinica(中央研究院资讯科学研究所); National Taiwan University of Science and Technology(台湾科技大学); AI Research Center, Hon Hai Research Institute(鸿海研究院人工智能研究中心)
类目: Computation and Language (cs.CL)
备注:
Abstract:Structural pruning followed by a recovery stage is the standard recipe for compressing large language models (LLMs), yet the two stages are typically designed and tuned in isolation. In this work, we study how their interaction shapes the compressed model, sweeping ordering, the data that drives each stage, and the alignment of their training objectives. We find that instruction-following is the first capability to collapse under pruning, well before knowledge, reasoning, or core language modeling, and that much of the pruning in-domain loss is likely not knowledge loss but a domain-agnostic failure to respond coherently to the prompt. We find in our setting that in-domain data is the strongest pruning signal, whereas instruction-style data recovers the most. Stage order matters little, but only marginally. We further analyze how the importance ranking drifts before and after pruning. Together these results give practitioners a concrete recipe for composing pruning and recovery.
[NLP-258] An Interpretable Linguistically-Grounded Analysis of Scene Saliency in Movie Screenplays EMNLP2026
【速读】： 该论文旨在解决电影剧本中场景显著性（salience）的识别问题，具体聚焦于两个核心研究问题：1）均匀信息密度（Uniform Information Density, UID）是否能够预测显著场景？2）除UID外，哪些语言学特征可区分显著场景与非显著背景场景？其解决方案的关键在于提出一种基于语言学的显著性模型（Linguistically-Grounded Salience Model, LGSM），该模型采用双流分类架构，融合冻结的BERT编码器与显式语言学特征，并通过自适应门控机制实现特征融合。研究发现，虽然按场景层面的UID在控制长度后不再具有独立预测能力，但在词元（token）层面，显著场景会特别抑制最罕见、最难预测的高尾部词元，表明UID的作用机制存在于场景内部的词元层级而非场景整体聚合。此外，显著场景表现出适度且持续的核心角色参与度（呈现倒U型模式）、较强的第三人称使用、句法更简单的语言结构以及更多关于动机与冲突的抽象词汇。LGSM有效捕捉了这些语言学特征组合，相较于强基线模型在显著场景召回率上有所提升，并生成了更具信息量的下游摘要。
链接: https://arxiv.org/abs/2610.15544
作者: Uday Bindal,Ishaan Karan,Manan Garg,Ekansh Goyal,Rahul Mishra,Rajakrishnan Rajkumar
机构: 未知
类目: Computation and Language (cs.CL)
备注: Accepted at EMNLP 2026 Main Track, Long Paper
Abstract:We investigate the factors affecting salience in movie screenplay by posing two research questions: 1. \em Does Uniform Information Density (UID) predict salient scenes? 2. \em What other linguistic properties distinguish salient scenes from non-salient background? We answer both questions through a controlled corpus study on the MENSA corpus, paired with a Linguistically-Grounded Salience Model (LGSM), a dual-stream classifier that fuses a frozen BERT encoder with explicit linguistic features through adaptive gating. The UID picture is layered in long-form discourse. Per-scene UID summaries carry no independent signal once length is controlled, but at the token level salient scenes match non-salient material in the bulk while specifically suppressing the rarest, hardest-to-predict tokens in the upper tail. UID therefore lives at the within-scene token level rather than as a per-scene aggregate. Salient scenes also show a moderate and sustained cast of central characters (following an inverted-U pattern), strong third-person usage, syntactically simpler language, and more abstract vocabulary about motivation and conflict. LGSM exploits this signature and improves recall on salient scenes over a strong semantic baseline while yielding better downstream summaries.
[NLP-259] Apertus 1.5: From Open Text Models to Open Multimodal Reason ers
【速读】： 该论文旨在解决当前大语言模型（Large Language Models, LLMs）高度依赖少数商业机构所带来的技术垄断与数据封闭问题，提出构建一个完全开源、基于公开数据集并可复现的多模态模型基础设施。其核心解决方案在于通过扩展原有的纯文本开源模型 Apertus 1.0，实现从单一文本模态向多模态推理（multimodal reasoning）的规模化演进，即推出 Apertus 1.5。关键创新点在于：在保持原有语言模型架构不变的前提下，仅通过持续预训练（continued pretraining）引入4T（8B模型）和2T（70B模型）的新数据，涵盖图像与音频信息，并将上下文窗口扩展至262,144个标记；多模态能力通过统一的分词器（tokenizer）实现，而语言模型头部仍仅处理文本输入，从而确保系统兼容性与可扩展性。此外，模型还结合监督微调、长上下文微调、基于可验证奖励的强化学习及偏好对齐等方法完成优化。特别值得注意的是，“思考”（thinking）被作为对话模板中的标志而非独立模块，显著提升了8B模型在数学推理任务上的表现（最高提升20分）。最终发布的模型在知识、指令遵循、文化理解等多个基准上优于同类全开源模型，尤其在文档理解、长篇语音识别及低资源语音识别方面表现出色，且整体遵循开放数据伦理，所有模型权重、训练流程、评估套件均以宽松许可发布，同时通过《Apertus宪章》明确价值对齐原则，实现了技术开放性与价值观透明性的统一。
链接: https://arxiv.org/abs/2610.15542
作者: Project Apertus:Yixuan Xu(1),Yassine Turki(2),Vinko Sabolčec(2),Valentina Pyatkin(1),Tommy Chu(2),Sven Najem-Meyer(2),Sean F. Carroll(2),Robert Smith(1),Raphael Kreft(1),Oscar Alessandro Tazza(1),Oleg Lavrovsky(2),Nicola Irmiger(1),Matteo Santelmo(2),Marian Schneider(1),Juan García Giraldo(2),Hanna Yukhymenko(1),Eduard Ďurech(1),Davit Melikidze(1),Dani Tamayo(2),Bettina Messmer(2),Aryan Ahadinia(2),Anunay Yadav(2),Allen Huang(2),Ahmad Fraij(1),Alessandro Bombarda(2),Alex Padula(1),Alexander Hägele(2),Alexander Sternfeld(3),Anna Hedström(1),Anna Kosovskaia(1),Arthur Wuhrmann(2),Arundhati Balasubramaniam(1),Badr Al Mahouri(2),Barna Pasztor(1),Bartosz Szostakiewicz(1),Bich Ngoc Doan(2),Bryan Schmid(4),Burak Sezen(1),Camille Challier(2),Clément Charmillot(2),Dana Shayakhmetova(2),Daniel Perruchoud(4),David Brüggemann(5),David Isztl(6),Elia Palme(6),Emma Anastassova(2),Fabian Bösch(6),Faruk Zahiragić(2),Fawzi Roberto Mohamed(6),Fiona Panicali(4),Florian Kolly(2),Flynn Guerra(4),Francesco Monti(2),Frederike Lübeck(1),Gennaro Oliva(6),Haofeng Yang(1),Haoliang Shang(1),Helen El Khoury(2),Henrique Mendonça(6),Jannis Vamvas(7),Jérome Landolt(4),Jessica Lam(1),Jingchun Xiao(1),Jonathan Coles(6),Joshua Oyebanji(2),Khanh Nguyen(2),Konstantinos Pitas(1),Lasse B. Strand(1),Li Mi(1),Loïc Deslarzes(2),Lorenzo Baggi(1),Lorenzo Noci(1),Luca Baumann(1),Luca Sartori(1),Lukas Drescher(6),Marco De Negri(1),Marco Passerini(6),Mark Klein(6),Marko Stojanovic(2),Martin Majsec(2),Matin Ansaripour(2),Mattea Busato(2),Mauro Pellonara(2),Max Neuwinger(1),Michael Graber(4),Michal Tešnar(1),Michelle Wastl(7),Miguel Gila(6),Mikolaj Boronski(2),Morgane Magnin(2),Moritz Reihs(1),Mukhammadali Sayfiddinov(1),Nathan Ranchin(1 and 2),Nicolas Cuveillier(2),Oscar Wohlfahrt(2),Qianqian Zou(5),Rada Kamysheva(1),Rita Sevastjanova(1),Roman Machacek
机构: ETH Zurich; EPFL; HES-SO Valais-Wallis; FHNW; SDSC; CSCS; University of Zurich; University of Bern
类目: Computation and Language (cs.CL)
备注:
Abstract:Language models are becoming infrastructure, and infrastructure built in the open, by public institutions, on documented data, is the alternative to dependence on a few providers; Apertus shows this is viable at scale. Apertus 1.5 extends the fully open Apertus family from text to multimodal reasoning. Continued pretraining of the Apertus 1.0 models, on 4T additional tokens for the 8B and 2T for the 70B, adds image and audio understanding and extends the context window fourfold to 262,144 tokens; multimodality enters through the tokenizer, the language-model head remains restricted to text, and the architecture is unchanged. Supervised finetuning, long-context finetuning, reinforcement learning with verifiable rewards, and preference alignment complete the released models. Thinking is a flag of the chat template rather than a separate model, and lifts mathematical reasoning by up to 20 points on the 8B model. The released models improve over their predecessors on nearly every benchmark; among fully open models of its class, the 8B leads on knowledge, instruction following, and cultural benchmarks, and its instruction following surpasses all compared models, including open-weight ones. On image and audio understanding, a first for the family, the models are competitive with fully open multimodal models, strongest on document understanding and long-form and lower-resource speech recognition, while specialized vision and audio models remain ahead on perception and transcription of English and other high-resource languages. As in Apertus 1.0, we train only on openly available data, respect crawler opt-outs retroactively, and exclude restrictively licensed material. We release the model weights, data preparation and training recipes, and evaluation suites under permissive licenses; openness extends to values, with the Apertus charter guiding alignment stages published alongside the models.
[NLP-260] A Step in the Right Place: Understanding and Optimizing Timestep Placement in Continuous Diffusion Language Models
【速读】： 该论文旨在解决连续扩散语言模型（continuous diffusion language models）在实现并行生成效率优势时面临的质量下降问题，尤其是在仅使用少量去噪步骤的情况下。其核心挑战在于：尽管采样步数固定，但去噪轨迹中各时间步（timestep）的分布位置具有灵活性，而这一位置选择对最终隐空间表示及解码后的词元分布有显著影响。为此，论文提出StepTune算法，通过在小规模校准集上搜索优化的时间步放置策略，以提升生成质量。该搜索过程采用语言感知的目标函数，在模型参数与采样预算固定的条件下进行，一旦完成校准，时间步配置即固定并可泛化至所有输入。实验表明，StepTune在不同采样预算下均显著提升无条件生成的困惑度及条件任务性能；尤其在16步采样且模型冻结的情况下，其在XSum摘要和WMT14翻译任务上的表现超越原生64步基线，并将端到端延迟降低约3.8倍。该研究确立了时间步放置作为改善连续扩散语言模型质量-效率权衡的一种实用有效手段。
链接: https://arxiv.org/abs/2610.15527
作者: Shutian Zheng,Zhihong Yu,Linye Wei,Zeyang Zhang,Meng Li
机构: Peking University (北京大学); Tsinghua University (清华大学); Institute for Artificial Intelligence, Peking University (北京大学人工智能研究院); School of Integrated Circuits, Peking University (北京大学集成电路学院); Beijing Advanced Innovation Center for Integrated Circuits (北京集成电路先进创新中心); Institute for Interdisciplinary Information Sciences, Tsinghua University (清华大学交叉信息研究院)
类目: Computation and Language (cs.CL)
备注: 24 pages, 7 figures. Reproducibility materials: this https URL
Abstract:Continuous diffusion language models represent an emerging paradigm for language modeling. They generate text in parallel by iteratively refining continuous representations and mapping them back to tokens through a decoder. However, realizing the efficiency advantage of parallel generation requires maintaining generation quality with only a few denoising steps. Even when the number of steps is fixed, their placement along the denoising trajectory remains flexible. In this paper, we demonstrate that changing timestep placement alone can substantially alter the final latent representations and thereby affect the decoded token distributions. Building on these findings, we introduce StepTune, an algorithm that searches for timestep placements that improve generation quality. The search is conducted on a small calibration set and guided by a language-aware objective, with the model and sampling budget fixed. Once calibrated, the timestep placement is fixed and reused across inputs. Experiments demonstrate consistent improvements in unconditional generation perplexity and conditional task metrics across all evaluated sampling budgets. In particular, StepTune enables 16-step sampling with a frozen model to outperform its native 64-step baseline on both XSum summarization and WMT14 translation and reduces end-to-end latency by approximately 3.8\times . These results establish timestep placement as a practical means of improving the quality-efficiency trade-off in continuous diffusion language models.
[NLP-261] SARATHI: Decoupled Data-Free Probing and Adaptive Reconstruction for Structured LLM Pruning EMNLP2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在结构化剪枝（structured pruning）过程中存在的两大核心问题：一是现有后训练剪枝方法将神经元选择与权重重建耦合于同一校准数据集，导致领域偏差（domain bias）及高稀疏度下的内存溢出（out-of-memory）问题；二是后LayerNorm架构中因系统性激活漂移（activation drift）引发的困惑度崩溃（perplexity collapse）。其解决方案的关键在于提出SARATHI框架，通过解耦神经元选择与权重重建两个阶段实现突破：首先，采用无需数据的非负矩阵分解残差子空间探测器（data-free NMF Residual Subspace Probe），基于权重内在几何结构对神经元进行评分，具备与鲁棒主成分分析（Robust PCA）的理论关联，避免了对校准数据的依赖；其次，引入自适应观测基重建（Adaptive OBS），仅需局部校准批次，在每层保持O(d2)峰值内存开销，彻底消除传统方法中的全局内存瓶颈；最后，通过偏置偏移补偿（Bias Shift Compensation）步骤有效修正后LayerNorm架构中的激活漂移，保障高稀疏度下的模型稳定性。实验表明，SARATHI在两种架构家族（门控与非门控前馈网络）的六种模型上，于15%–40%稀疏度下均实现了与现有基线相当的困惑度表现，并在零样本准确率上达到或超越现有方法。
链接: https://arxiv.org/abs/2610.15514
作者: Sankar Behera,Dhruv Singh,Satyadev Ahlawat,Yamuna Prasad
机构: IIT Jammu(印度理工学院贾姆穆)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Accepted to EMNLP 2026 Main Conference
Abstract:Structured pruning reduces the inference cost of large language models (LLMs) by permanently removing neurons and weight rows, yielding dense, hardware-agnostic speedup. Existing post-training methods couple neuron selection and weight reconstruction through a shared calibration dataset, introducing domain bias and out-of-memory failures at high sparsity. We propose SARATHI, a framework that decouples these two stages. A data-free \emphNMF Residual Subspace Probe scores neurons using intrinsic weight geometry, without accessing any calibration data, and carries a formal connection to Robust PCA. \emphAdaptive OBS then reconstructs the pruned weights with a localized calibration batch, operating in \mathcalO(d^2) peak memory per layer and eliminating the global memory bottleneck of coupled methods. A \emphBias Shift Compensation step corrects systematic activation drift in post-LayerNorm architectures that would otherwise cause perplexity collapse at high sparsity. Experiments across six models from two architecture families (gated and non-gated FFN) at 15–40% sparsity show that SARATHI achieves competitive perplexity and matches or exceeds existing structured pruning baselines on zero-shot accuracy.
[NLP-262] REPRO: Reflective Evolutionary Prompt Optimizer for Automated LLM Red-Teaming
【速读】： 该论文旨在解决安全对齐的大语言模型（LLM）在面对对抗性提示（adversarial prompts）时仍存在漏洞的问题，尤其关注现有自动化红队测试方法在实际部署场景中面临的局限性：如需要特权访问、依赖人工设计的特定模型模板，或仅优化单一目标而缺乏种群层面的多样性。其解决方案的关键在于提出一种纯黑盒框架——REPRO（Reflective Evolutionary Prompt Optimization），该框架基于一个压缩后的70亿参数反思型模型（reflective model），仅需单张消费级GPU即可实现无需外部知识库的进化式提示优化。REPRO的核心创新在于引入两个此前未被充分探索的机制：一是语义交叉（semantic crossover），即由大语言模型引导的多父本重组算子，能够生成具有语义连贯性的新提示；二是基于结构化失败分析的贝叶斯策略选择，通过汤普森采样（Thompson Sampling）动态调整搜索策略，提升探索效率。实验覆盖36个来自11个模型家族的目标模型（包括GPT-5-nano、Claude-3.5-Sonnet、Gemini-2.5-Pro和Grok-4.1等），在统一评估协议下，REPRO实现了单轮迭代中对所有目标模型100%的攻击成功率（ASR），平均每个行为仅需24次目标查询。值得注意的是，该指标反映的是在迭代预算内是否成功找到有效攻击，而非固定预算下的表现，因此更具实际意义。研究强调，所有基线结果均源自原始论文，使用了不同的评判标准、预算和模型版本，故本文更侧重于展示在统一协议下对多种模型的广泛覆盖能力，而非直接性能对比。
链接: https://arxiv.org/abs/2610.15513
作者: Abrar Alotaibi,Moataz Ahmed
机构: King Fahd University of Petroleum and Minerals (沙特法赫德国王石油矿产大学); Imam Abdulrahman Bin Faisal University (伊玛目阿卜杜勒拉赫曼·本·法伊斯尔大学); SDAIA-KFUPM Joint Research Center for Artificial Intelligence (SDAIA-KFUPM 人工智能联合研究中心)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Neural and Evolutionary Computing (cs.NE)
备注: Under Review for ACM TOSEM
Abstract:Safety-aligned large language models (LLMs) remain vulnerable to adversarial prompts that bypass their safety training. Existing automated red-teaming methods either require privileged access unavailable in production, rely on manually engineered per-model templates, or optimize a single objective without population-level diversity. We present REPRO (Reflective Evolutionary Prompt Optimization), a pure black-box framework that discovers adversarial prompts through evolutionary optimization driven entirely by a compressed open-weight 7B reflective model on a single consumer GPU. REPRO introduces two mechanisms absent from prior work, both isolated in a component ablation: semantic crossover, an LLM-guided multi-parent recombination operator, and Bayesian strategy selection via Thompson Sampling informed by structured failure analysis. We catalog REPRO’s behavior on AdvBench, JailbreakBench, and HarmBench across 36 target models from 11 families, including GPT-5-nano, Claude-3.5-Sonnet, Gemini-2.5-Pro, and Grok-4.1. Under our protocol REPRO reaches 100% attack success rate (ASR) on every evaluated target in a single turn, at a median of 24 target queries per behavior. Two properties of the protocol qualify that figure. The judge scoring candidates during optimization also produces the reported ASR, and search halts when it accepts, so ASR reports whether search succeeded within the iteration budget rather than success at a fixed budget. All baseline figures are transcribed from their original papers, which use different judges, budgets, and model versions. We therefore present this as a catalog of model coverage under one protocol rather than a head-to-head ranking, and develop both qualifications in the paper. Comments: Under Review for ACM TOSEM Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Neural and Evolutionary Computing (cs.NE) Cite as: arXiv:2610.15513 [cs.CL] (or arXiv:2610.15513v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2610.15513 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-263] Concept Representations in Vector Space: Quantifying Sharedness and Idiosyncrasy of Meaning
【速读】： 该论文旨在解决个体间概念表征差异的测量与建模问题，即不同个体在理解和使用同一概念时是否存在系统性差异。其核心挑战在于如何量化并分析这种跨个体的意义变异，尤其是在自然语言语境下。解决方案的关键在于构建一个基于句子嵌入（sentence embeddings）的测量框架，将参与者提供的简短个人定义编码为稠密向量，并通过交叉随机效应模型分析成对个体间的语义差异。该框架提取两个关键概念层级指标：共享度（S_b），反映任意两人之间预期的语义差异；异质性（I_b），表示个体特异性贡献的二元差异方差部分。研究通过预注册设计验证了该框架的有效性，结果显示嵌入空间能准确恢复概念与领域结构，且无需参数的分类器在扩展的外延预测任务中达到0.630的准确率（显著高于0.500的随机基线和0.893的一致性上限）。该方法在三种不同架构与容量的编码器中均实现结果复现，证明其可重复性与可扩展性，为意义变异的实证研究提供了可靠工具。
链接: https://arxiv.org/abs/2610.15496
作者: Simon Josef Durstewitz,Denis Buehler
机构: 未知
类目: Computation and Language (cs.CL)
备注: 22 pages, 6 figures. Preprint under review at Open Mind: Discoveries in Cognitive Science. Data, code and pre-registration: this https URL
Abstract:Concepts are the basic units through which we navigate the world and communicate meaning. But do you and I really mean the same thing when we think or speak about a concept? The present study develops a measurement framework for modeling and analyzing interindividual variation in concept representations through sentence embeddings, dense vector representations mapping entire utterances into a shared geometric space. The framework is developed in a theoretical argument and validated, demonstrated, and subjected to robustness checks in a pre-registered study. Short personal definitions, elicited from N = 125 participants, are encoded as sentence embeddings and submitted pairwise to a crossed random effects model yielding two concept-level quantities: sharedness ( S_b ), the expected dissimilarity between two randomly drawn individuals, and idiosyncrasy ( I_b ), the person-specific share of total dyadic variance. Two validity tests support the approach: the embedding space fully recovers concept and domain organization from participant definitions even with the target word excluded, and a parameter-free classifier achieves 0.630 accuracy in a within-participant extensional prediction task against a chance baseline of 0.500 and a consistency ceiling of 0.893. The framework finds theoretically interpretable variation in S_b and establishes I_b as a reliable source of dyadic variance across concepts. Results replicate across three encoders varying in architecture and capacity, providing a reproducible and scalable instrument for the empirical study of meaning variation.
[NLP-264] AlphaPilot: Multi-Turn Tool-Using LLM Agents for Adaptive Alpha Mining with Reinforcement Learning
【速读】： 该论文旨在解决生成式金融因子（generative alpha factor）挖掘中如何有效融合非结构化市场描述、历史数据与定量评估反馈以实现可解释性因子发现的难题。现有方法通常仅利用部分信号，难以全面捕捉市场动态。其解决方案的关键在于提出AlphaPilot——一个基于强化学习的多轮工具调用大语言模型（LLM）框架，通过让代理在每一轮中结合市场文本描述，生成候选因子表达式，调用真实数据评估器获取反馈，并动态更新保留因子池，从而协同利用市场文本、大模型金融知识、真实数据及评估反馈。该机制支持在推理阶段逐日进行搜索与因子池迭代，自适应应对市场状态变化，在沪深300和沪深500指数的滚动回测中分别实现42.93%和54.19%的平均年化收益率，以及2.46和2.86的平均夏普比率，显著优于对比方法，在平均总收益、夏普比率和RankIC指标上均表现最优。
链接: https://arxiv.org/abs/2610.15488
作者: Peiqi Jiang,Fan Zhang,Xin Xu,Tianhao Chen,Can Yang
机构: Hong Kong University of Science and Technology(香港科技大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 25 pages, 9 figures, 20 tables
Abstract:Formulaic alpha mining seeks interpretable expressions that rank assets by future returns. The task is challenging because it demands simultaneous processing of unstructured market descriptions, historical data, and quantitative evaluation of candidate factors, yet most existing methods exploit only a subset of these signals. We introduce AlphaPilot, a reinforcement-learning framework that trains multi-turn tool-using LLM agents for adaptive alpha mining. The agent conditions on market descriptions, proposes factor expressions, queries real-data evaluator feedback, and updates a retained factor pool across turns. This design jointly leverages market text, LLM financial knowledge, real data, and evaluation feedback. At inference time, AlphaPilot performs day-by-day search and pool updates to adapt to shifting market regimes. Across three independent runs in out-of-sample rolling backtests on CSI300 and CSI500, AlphaPilot achieves mean total returns of 42.93% and 54.19%, with mean Sharpe ratios of 2.46 and 2.86, respectively. It attains the highest mean total return, Sharpe, and RankIC among the evaluated methods on both universes.
[NLP-265] Hidden in Plain Sight: Advancing Harmful Content Understanding via Perception-Aware Adversarial Self-Play
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在识别隐含于表面文本之中的有害意图时存在的显著缺陷，即模型难以捕捉人类读者能够轻易理解的语境化、语音化及视觉化线索，导致恶意用户可利用这一认知不对称性，通过策略性伪装有害内容以规避自动化内容审核。其解决方案的关键在于提出一种感知意识对抗自博弈框架（Perception-Aware Adversarial Self-Play, PASP），该框架通过构建文本、语音和视觉多模态视图，在保持原始语境完整性的同时，显式暴露传统纯文本模型所忽略的发音特征与字形线索；在对抗自博弈过程中，攻击方基于防御方反馈生成意图不变但表层形式变化的规避样本，而防御方则通过意图不变性奖励机制进行训练，从而增强对演化中规避策略的鲁棒性理解。实验表明，PASP在ChineseHarm-Bench与EvoHarmBench上分别将自适应攻击成功率降低73.71%和74.49%，且在保持模型核心能力基础上有效解析电商场景中的真实隐蔽表达。
链接: https://arxiv.org/abs/2610.15449
作者: Yisong Xiao,Benlei Cui,Shiji Zhao,Zonghao Ying,Yaoyuan Zhang,Longtao Huang,Aishan Liu,Hui Xue,Ting Ma,Haiwen Hong,Xianglong Liu
机构: SKLCCSE, Beihang University (北京航空航天大学可信计算与系统重点实验室); Yuvion Team (优云团队); Alibaba Group (阿里巴巴集团)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Despite significant advances in language understanding, Large Language Models (LLMs) can still miss meaning hidden in plain sight, failing to identify harmful intent in expressions readily understood by humans. Malicious users exploit this gap by strategically disguising harmful content to evade automated moderation, posing a serious threat to online platform safety. The vulnerability reflects a fundamental cognitive asymmetry between human readers, who integrate contextual, phonetic, and visual cues to decipher the intent behind deliberate obfuscation, and text-based LLMs, which rely primarily on token sequences and have limited exposure to evolving evasions. To bridge this asymmetry, we propose PASP (Perception-Aware Adversarial Self-Play), which combines multi-perceptual views with adversarial self-play to improve understanding of harmful intent as evasions evolve. Specifically, PASP constructs textual, phonetic, and visual views of each input, preserving its original context while exposing pronunciation and glyph cues obscured in token space. During adversarial self-play, an attacker uses defender feedback to generate intent-preserving evasions, while the defender is trained using an intent-invariance reward to understand harmful intent across changing surface forms. Extensive experiments show that PASP reduces adaptive attack success rates by up to 73.71% and 74.49% on ChineseHarm-Bench and EvoHarmBench, respectively, preserves model capabilities with minimal degradation, and effectively interprets real-world evasive expressions in e-commerce.
[NLP-266] Sensitivity or Criterion? Measuring and Moving When Language Models Act
【速读】： 该论文旨在解决大模型在决策过程中对“是否采取行动”（when-to-act）判断的模糊性问题，即传统以单一准确率衡量模型表现时，无法区分模型是因缺乏敏感性（sensitivity）而未能识别应采取行动的情境，还是因过于保守的决策阈值（criterion）而主动回避行动。其核心解决方案是提出一种名为FlipAct的新诊断工具，通过构建基于标注工具属性规则的分级证据评分系统，将“是/否”决策边界固定于零点，从而实现跨模型、跨任务的可比性。该方法采用2×2四格设计，保持所需知识恒定而仅翻转物理状态，通过拟合“是”响应率曲线的斜率来量化敏感性，其与0.5交叉点反映决策准则。实验表明，模型敏感性随规模增长，且其排序与多个外部当行动基准高度一致；但多数模型更关注证据符号而非幅度，对无改善或略有改善的状态也常接受行动，表现出显著的保守倾向。进一步研究发现，决策准则可通过激活引导（activation steering）调整，且经自身引导副本蒸馏后，该准则调整可固化于权重中，不显著影响敏感性。更重要的是，这种蒸馏后的准则偏移可有效迁移至六个外部基准（包括非二元回答任务），同时保持无需依赖准则的AUC指标稳定。最终，当模型初始为保守策略时，整体准确率提升；而初始为宽松策略则准确率下降或无显著变化，验证了通过可控准则调整优化决策性能的可行性。
链接: https://arxiv.org/abs/2610.15415
作者: Sixun Dong,Yebowen Hu,Ming Yin,Yiran Chen,Fei Liu,Chen Chen
机构: University of Central Florida(中佛罗里达大学); Zoom Communications(Zoom通信公司); Duke University(杜克大学); Emory University(埃默里大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Project page: this https URL
Abstract:Before an agent acts, it must decide whether action is warranted. A single accuracy cannot distinguish failure to recognize a warranted action from reluctance to act. Signal-detection theory separates sensitivity from criterion, but criteria estimated from authored binary labels depend on each benchmark’s example mix. We introduce FlipAct, a diagnostic of tool use in physical states. A rule over annotated tool properties computes each answer and a graded evidence score, fixing the Yes/No boundary at zero across models. Questions form 2-by-2 quads that hold required knowledge fixed while the state flips. The slope of the fitted Yes-rate curve measures sensitivity; its crossing at one half measures criterion. Across 44 open models, sensitivity rises with size within a family, and its ranking agrees with three of four external when-to-act benchmarks. Models respond to evidence sign rather than magnitude, accepting actions that leave the state no better about as often as helpful ones. When Yes means act, most models are conservative across rephrasings; in some, this partly reflects a preference for No. Criterion is not determined by size. Activation steering shifts criterion with weights fixed; distillation from a model’s own steered copy puts the shift into its weights without a significant sensitivity change. The distilled shift transfers to six external benchmarks, including two without Yes/No answers, while criterion-free AUC is preserved on the five benchmarks where it is available. Accuracy improves where models start conservative and declines or remains within error where they start permissive. Project page: this https URL
[NLP-267] From Candidate Tokens to Final Outcomes: Decoding Relative Long-Horizon Candidate Value from Reasoning Model Representations
【速读】： 该论文旨在解决生成式推理模型在生成推理轨迹过程中，如何从局部候选词（token）的内部表示中捕捉其对长期推理结果（long-horizon reasoning outcomes）的影响这一关键问题。尽管在给定前缀下存在多个合理的下一步候选词，但它们在后续推理路径中的表现差异显著，而这种差异是否反映在模型的内部表征中尚不明确。为此，论文提出“相对长时程候选值”（relative long-horizon candidate value）作为量化指标，并构建了推理候选词-令牌价值（Reasoning Candidate-Token Value, RCTV）数据集，将数学推理前缀与多个合理候选词及其对应的实证长期价值配对。通过在多层隐藏状态、多层感知机输出和注意力输出上使用稀疏线性探测器进行分析，实验表明相对长时程候选值可在线性意义上从这些表征中被有效解码，且基于候选值的探测方法优于基于概率和嵌入的基线方法。进一步的在线干预实验验证了利用解码出的信号指导生成过程中的词选择，能够显著提升最终推理成功的概率。研究结果表明，模型内部表征确实编码了关于候选词相对长期价值的信息，揭示了局部候选层面表征与全局推理结果之间的关联机制。
链接: https://arxiv.org/abs/2610.15412
作者: Yixin Yang,Yue Wang,Fandong Meng,Zhifang Sui
机构: Peking University (北京大学); Tencent Inc. (腾讯公司)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large reasoning models generate reasoning trajectories token by token. At a given prefix, multiple next-token candidates may be plausible, yet shape subsequent reasoning and differ in their chances of leading to the correct answer. Whether these differences are reflected in internal representations remains unclear. To study this question, we define relative long-horizon candidate value to quantify within-prefix differences and construct the Reasoning Candidate-Token Value (RCTV) dataset, pairing mathematical reasoning prefixes with plausible next-token candidates and their empirical long-horizon candidate values. Using RCTV, we examine hidden states, multilayer perceptron outputs, and attention outputs at candidate positions with sparse linear probes. Experimental results provide evidence that relative long-horizon candidate value is linearly decodable from these representations at multiple layers. Candidate-value probes can also outperform probability- and embedding-based baselines. Furthermore, a controlled online intervention experiment shows that using the decoded signals to guide token selection during generation can improve the chances of eventual success. Our findings provide evidence that internal representations encode information about relative long-horizon candidate value, offering insights into how local candidate-level representations relate to long-horizon reasoning outcomes.
[NLP-268] In-Place Indic Tokenizer Expansion and Continual Pretraining of Nemotron-3-Nano-30B-A3B NEURIPS2026
【速读】： 该论文旨在解决大语言模型在低资源印度语种（Indic languages）上的适配问题，特别是在不修改模型骨干架构的前提下实现对多种印度语言的有效支持。其核心挑战在于如何在保持模型原有英语能力的同时，高效扩展多语言词汇表并优化跨语言性能。解决方案的关键在于采用持续字节对编码（continued byte-pair encoding）与 AdaptMix 重平衡策略，在不改变原始架构的情况下将词表规模从131,072扩充至163,072，显著降低了每词平均分词数（减少38.6%），并通过基于90%印度语与10%英文混合数据的初始化和持续预训练流程，逐步引入多语言知识。训练过程分为广域训练、高质量退火和长上下文训练三个阶段，累计使用约1.27万亿个标记。实验结果显示，经过退火后的检查点在印地语多任务理解评估（Indic MILU）上提升12.9分，但英语多任务语言理解基准（MMLU）性能下降3.4分，表明存在跨语言性能权衡。此外，研究还揭示了直接添加令牌、权重衰减对固定参数的影响以及困惑度无法有效指导学习率选择等实际负结果，凸显了多语言微调中潜在的系统性风险。
链接: https://arxiv.org/abs/2610.15373
作者: Avinash Benki,Prasanjith Pasunuti,Sutanshu Saksena Raj,Parijat Chatterjee,Nayanjit Sarkar,Daksh Maan
机构: Gnani.AI(甘尼人工智能)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted at the GlobalSouthAI Workshop, NeurIPS 2026
Abstract:We adapted Nemotron-3-Nano-30B-A3B for eight Indian languages without changing its backbone architecture. We expanded the vocabulary in place from 131,072 to 163,072 entries using continued byte-pair encoding and AdaptMix rebalancing, reducing tokens per word by 38.6%. We initialized new input and output rows by averaging their original constituent rows, then trained them on 90% Indic and 10% English text before updating the backbone. Continual pretraining then progressed through broad training, high-quality annealing, and long-context training over approximately 1.27 trillion tokens. The annealed checkpoint gained 12.9 points on the Indic MILU benchmark but lost 3.4 on English MMLU. We also report practical negative results. Direct token addition underperformed continued byte-pair encoding. AdamW weight decay changed rows intended to stay fixed. Perplexity on an Indic-heavy holdout did not select the learning rate that best preserved English performance.
[NLP-269] A Sovereign Indic-First Reasoning Model
【速读】： 该论文旨在解决多语言大模型在印度本土语言（Indic languages）中推理、指令遵循与工具使用能力不足的问题，尤其针对长上下文场景下的性能退化与语言一致性缺失。其核心挑战在于：在保持高性能英文能力的同时，如何有效提升对多种印度语言的语义理解与输出规范性，避免因奖励机制设计偏差导致模型倾向于以英语回答本应以本地语言作答的任务。解决方案的关键在于引入一种分阶段的持续预训练与强化学习框架：首先通过增量扩展词表至163,072个token并基于上游分解初始化新嵌入向量，实现对印度语言的低延迟适配；随后采用高质数据的持续预训练与131,072序列长度的直接训练，增强长文本建模能力；在后训练阶段，利用四类可验证任务环境进行强化学习，显著提升数学、研究生级科学推理与指令遵循能力。为解决语言混淆问题，提出两种修复策略：一是引入限定印度语言的奖励机制以恢复语言规范性；二是采用26B-A4B参数量的生成式评判模型替代传统大模型作为奖励函数，经蒸馏后恢复了81%被优化目标所牺牲的能力。最终释放版本在印度语言综合指标上相较起点模型提升5.2点，虽损失1.7点英文保留能力，但未发布版本通过更精细的语言感知分支进一步获得14.6点印度语言增益，代价为5.4点英文下降。未来计划推出FP8量化版本，将显存占用降至32GB，同时在32个基准测试中几乎无均值性能损失，实现了高效能与多语言一致性的平衡。
链接: https://arxiv.org/abs/2610.15371
作者: Sutanshu Saksena Raj,Prasanjith Pasunuti,Parijat Chatterjee,Daksh Maan,Purusharth Malik,Nayanjit Sarkar,Avinash Benki
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 69 pages, 24 figures, 49 tables. Technical Report
Abstract:Gnani-Evon-v3.3-30B-A3B is a long-context hybrid Mamba-Transformer mixture-of-experts model for reasoning, instruction following, and tool use in English and eight Indian languages. Starting from NVIDIA’s Nemotron-3-Nano-30B-A3B checkpoint, it retains the 52-layer backbone and changes the vocabulary, training data, and post-training objectives. An in-place expansion from 131,072 to 163,072 tokens reduces mean Indian-language fertility by 38.6%. The new embedding rows are initialized from their upstream decompositions and warmed before the backbone is updated. Continual pretraining then adds broad English-Indic data, high-quality annealing, and direct training at a 131,072-token sequence length. The central result appears after post-training. Reinforcement learning across four verifiable task environments improves mathematics, graduate-level science, and instruction following, but degrades Indic extractive question answering because the rewards do not require an answer in the requested language. The share of Indic extraction items answered in English rises from 8.7% to 91.7%. Accuracy conditional on answering in the target script roughly doubles, which is consistent with preserved reading ability but is measured on a self-selected subset. Two unreleased studies test repairs. A scoped Indic reward restores the language convention. A 26B-A4B generative judge outperforms a 235B-parameter model on RM-Bench, and distillation restores 81% of the capabilities displaced by optimizing against it. Relative to the architecture-identical starting checkpoint, the released model gains 5.2 points on an Indic aggregate while giving up 1.7 points of English retention. The unreleased Indic-aware branch trades further, gaining 14.6 Indic points at a cost of 5.4 in English. A planned FP8 release reduces the artifact to 32GB with near-zero mean change across 32 paired benchmarks. Comments: 69 pages, 24 figures, 49 tables. Technical Report Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) Cite as: arXiv:2610.15371 [cs.CL] (or arXiv:2610.15371v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2610.15371 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Sutanshu Saksena Raj [view email] [v1] Sun, 11 Oct 2026 05:52:29 UTC (2,839 KB)
[NLP-270] North or South? Diagnosing LLM s Understanding of North Korean Language Society and Knowledge AACL
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在低资源社会语言情境下理解能力不足的问题，尤其聚焦于朝鲜半岛北韩社会这一特殊语境。由于长期政治分隔，北韩与南韩在词汇、书写系统、社会规范及意识形态话语上形成显著差异，但现有模型对这类非主流社会文化背景的理解仍极为有限。为此，研究提出LENS（Language and sociocultural Evaluation for North Korean Society）基准测试体系，包含1,765个经人工验证的问题，从语言鲁棒性、社会文化视角一致性以及领域特定知识三个维度评估模型表现。关键发现表明：尽管模型在语言层面表现尚可，但在社会文化视角对齐方面存在显著偏差，且在涉及北韩特定历史、制度与现实的领域知识上表现出严重局限；质性分析进一步揭示模型常将正式制度等同于实际生活经验，过度投射自由民主价值预设，或依赖粗略刻板印象。这说明仅靠语言相似性无法保障对社会文化及专业领域的深层理解，凸显了构建针对非主流文化与政治语境的地面实证评估资源的必要性。
链接: https://arxiv.org/abs/2610.15282
作者: Kiwoong Park,Seyoung Song,Donghyun Woo,Alice Oh
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to AACL-IJCNLP 2026 (Main Conference). 19 pages, 1 figure, 11 tables
Abstract:Large Language Models (LLMs) have made substantial progress in multilingual question answering, yet their ability to understand low-resource sociolinguistic contexts remains under-explored. We study North Korea as a unique testbed where decades of division have produced distinct differences from South Korea in vocabulary, orthography, norms, and ideological discourse. We introduce LENS (Language and sociocultural Evaluation for North Korean Society), a human-validated benchmark of 1,765 questions evaluating LLMs across three dimensions: linguistic robustness, sociocultural perspective alignment, and domain-specific knowledge. Evaluating diverse recent LLMs, we find a modest performance drop on North Korean language inputs, a larger gap in sociocultural perspective alignment, and severe limitations on domain-specific knowledge. Qualitative analysis reveals that models often conflate formal institutions with lived realities, project liberal-democratic assumptions, or rely on coarse stereotypes. Our results demonstrate that linguistic similarity alone does not guarantee sociocultural or domain-specific understanding, highlighting the critical need for grounded evaluation resources for underrepresented cultural and political contexts.
[NLP-271] Do correct answers have stable latent knowledge support?
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在输出正确性之外，其潜在知识表征是否具备稳定内部表示的问题。现有评估方法仅关注输出结果的正确性，但无法验证支撑答案的隐含知识是否存在可靠的内在结构稳定性。为此，论文提出两个核心指标：潜在状态稳定性（Latent State Stability, LSS） 和 知识拓扑稳定性（Knowledge Topology Stability, KTS），分别用于衡量单个知识状态及其关系结构在不同诱发条件下的稳定性。其关键解决方案在于：首先，在不同诱发条件下对同一知识的模型内部表示进行平均，并以基线条件下的均值为参考，通过去除共性偏移得到校正后的表示；随后，LSS定义为校正表示与参考表示之间的距离，经同类型知识项间典型距离归一化；KTS则衡量同一类知识项在不同条件下的成对表示距离排序一致性。为支持系统性研究，论文构建了EpiStableBench基准，包含4,411个涵盖事实知识与数学推理的样本及76,766个受控查询。实验表明，即使在基础条件下回答正确，不同知识项的表示稳定性差异显著；在同一模型内，更高的稳定性通常对应更优的泛化准确率，但跨模型比较显示稳定性并不随模型规模或整体准确性单调提升。进一步地，论文提出潜在DPO（LatentDPO） 方法，用于学习一种考虑潜在知识稳定性的知识编辑策略，实验证明其与多种现有编辑方法结合后，能持续提升编辑成功率与表达形式间的泛化能力。
链接: https://arxiv.org/abs/2610.15277
作者: Chuancheng Shi,Shangze Li,Wenhua Wu,Kun Hu,Fei Shen,Zhiyong Wang,Tat-Seng Chua
机构: University of Sydney(悉尼大学); Nanjing University of Science and Technology(南京理工大学); Edith Cowan University(埃迪斯科文大学); National University of Singapore(新加坡国立大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models are typically evaluated in terms of the correctness of their output. However, output correctness alone does not confirm whether the latent knowledge supporting an answer has stable internal representations. We therefore propose latent state stability (LSS) and knowledge topology stability (KTS) to investigate the stability of individual knowledge states and their relational structure, respectively. Specifically, we first separately average the model’s internal representations of the same knowledge under each elicitation condition. Using the mean representation under the baseline condition as a reference, we remove shared shifts to obtain corrected representations. We then formulate LSS as the distance between each corrected representation and its reference representation, normalized against the typical distance between knowledge items of the same type. We formulate KTS as the cross-condition consistency of the ordered pairwise representation distances among knowledge items of the same type. Next, we construct EpiStableBench, which comprises 4,411 items spanning factual knowledge and mathematical reasoning, with 76,766 controlled queries. Across 16 LLMs from four families, representation stability varies substantially even among items answered correctly under the base condition. Within models, higher stability generally corresponds to higher accuracy under held-out elicitation conditions; across models, stability does not consistently increase with model size or overall accuracy. Meanwhile, we present LatentDPO to learn a knowledge-editing policy that accounts for the stability of latent knowledge. Experimental results show that integrating LatentDPO with multiple existing editing methods consistently improves editing success and generalization across different expressions.
[NLP-272] Sequent-Prover: Training Agents for Formal Checkable SMT-based Reasoning
【速读】： 该论文旨在解决大语言模型在形式化推理任务中缺乏可执行、可验证的输出路径，以及推理过程不可靠、易产生语义偏差的问题。现有方法往往依赖于模型自洽生成的逻辑表达式，但难以保证其正确性与可执行性。为此，论文提出Sequent-Prover——一种基于验证器（verifier）驱动的语言模型代理家族及其后训练框架。其核心解决方案在于采用“验证器条件轨迹蒸馏”（verifier-conditioned trajectory distillation）策略：通过采样多轮交互轨迹，保留那些最终被验证器确认为正确、非平凡、语义一致且可执行的推理尾部，并结合监督微调与轨迹级偏好优化进行精炼。该方法不仅提升了推理结果的准确性，还显著增强了推理效率与可解释性。实验表明，在37个基准测试中，40亿参数模型的单次尝试准确率从85.6%提升至97.0%，四次全正确可靠性（pass^4）从64.1%提升至91.7%，同时平均SMT调用次数由4.10降至2.52，每次尝试的词元数从18.97k降至12.72k。最终，Sequent-Prover代理能够输出具有可执行形式化佐证的可信答案，实现了从“推测性推理”到“可验证推理”的范式转变。
链接: https://arxiv.org/abs/2610.15258
作者: Debargha Ganguly,Weicong Chen,Mohsen Hariri,Vikash Singh,Sreehari Sankar,Shouren Wang,Wang Yang,Biyao Zhang,Chaoda Song,Alexander Nemecek,Kai Ye,Osama Zafar,Erman Ayday,Joe Stubbs,Srinivasan Iyengar,Vipin Chaudhary
机构: Case Western Reserve University(凯斯西储大学); UT Austin(德克萨斯大学奥斯汀分校); Microsoft Research(微软研究院)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Logic in Computer Science (cs.LO)
备注:
Abstract:We introduce Sequent-Prover, a family of verifier-using language-model agents, and the post-training framework that produces them. Agents reason in an executable SMT-LIB scratchpad: they write complete SMT-LIB programs, query Z3 across turns, inspect solver feedback, repair formalizations, and terminate with an explicit answer. Our recipe, verifier-conditioned trajectory distillation, bootstraps from a noisy teacher proposal policy by sampling multi-turn trajectories, retaining the successful tail that is correct, non-trivial, grounded, and executable, then applying supervised fine-tuning and trajectory-level preference optimization. Because solver success alone does not guarantee semantic faithfulness, we evaluate task accuracy, k-sample reliability, and SMT program quality. Across 37 benchmarks and 7,254 problems per model family, post-training raises the 4B model’s per-attempt accuracy from 85.6% to 97.0% and four-sample all-correct reliability (pass^4) from 64.1% to 91.7%. It also reduces average SMT calls from 4.10 to 2.52 and tokens per attempt from 18.97k to 12.72k. Sequent-Prover agents give answers backed by executable formal artifacts.
[NLP-273] Exploring Jailbreak Vulnerability Across Depth-Based Dynamic Inference in Large Language Models NEURIPS2026
【速读】： 该论文旨在解决生成式 AI（Generative AI）在采用基于深度的动态推理（depth-based dynamic inference）时，其对越狱攻击（jailbreak attacks）的鲁棒性下降的问题。尽管动态推理通过从中间层输出结果提升了计算效率并增强了资源适应性，但其对安全性的潜在影响尚未被充分研究。论文的关键发现在于：1）浅层退出（shallow exits）相较于深层退出表现出更强的抗越狱攻击能力；2）退出深度影响对抗后缀（adversarial suffix）的优化难度，深层退出可实现更低的攻击损失，表明其更易被攻破；3）对抗后缀对不同退出深度的模型输出效用影响基本一致。因此，解决方案的关键在于在设计动态推理机制时，优先考虑浅层退出策略，并结合退出深度与攻击鲁棒性的权衡，以在保证效率的同时提升模型安全性。
链接: https://arxiv.org/abs/2610.15207
作者: Lass Najmuldeen,Seyedhamidreza Mousavi
机构: Mälardalen University (马拉尔德大学); Västerås, Sweden
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Cryptography and Security (cs.CR)
备注: Accepted at the NeurIPS 2026 Workshop on Foundations of Language Model Security (FLMSec). 10 pages, 4 figures, 3 tables
Abstract:Large Language Models (LLMs) have been shown to be computationally expensive and vulnerable to jailbreak attacks, in which adversarial prompts bypass safety alignment and induce harmful responses. To address this inefficiency, a range of techniques have been proposed to enable efficient LLM inference. Depth-based dynamic inference has demonstrated promising results by enabling outputs to be generated from intermediate layers. It offers the flexibility to adjust model behavior based on the available resource budget. However, how this flexibility affects the robustness of LLMs against jailbreak attacks remains unexplored. In this paper, we address this gap via a comprehensive investigation into the impact of depth-based dynamic inference on the robustness of LLMs against jailbreak attacks. Specifically, we make the following key observations: 1) Shallow exits are consistently more robust to jailbreak attacks than deeper ones; 2) Exit depth affects how easily the adversarial suffix can be optimized, with deeper exits reaching lower attack loss; and 3) Adversarial suffixes affect utility uniformly across exit depths. These insights can help design dynamic inference methodologies that are more robust to jailbreak attacks. Code is available at this https URL
[NLP-274] Do LLM s differentiate linguistic traits internally?: A layer-wise analysis
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在生成类人语言判断时，是否真正内化并区分语言细微差异（如常规意义与习语意义）这一关键问题。尽管模型输出表现出与人类相似的响应模式，但其内部表征机制是否具备对语义细微差别的敏感性仍不明确。研究通过层间表征分析，对比了两个参数量相当的预训练模型——Llama-3.2-1B 与 Gemma-3-1B-PT，采用逐层探针（layer-wise probing）结合几何距离度量方法，利用各层中主成分载荷排序最高的嵌入维度进行分析。结果表明，即使模型容量相同，其内部语言信息的组织方式也存在根本性差异，揭示出仅依赖行为输出评估模型与人类语义理解一致性是不足的。因此，解决方案的关键在于深入分析模型内部表征的几何结构，不仅关注其输出行为，还需考察语义空间对齐情况，以更准确地评估模型与人类语言理解之间的对应关系。
链接: https://arxiv.org/abs/2610.15203
作者: Samhita Bollepally,Takashi Yamauchi,Manju Muralee,Motunrayo Adebimpe
机构: Texas A&M University (德克萨斯农工大学)
类目: Computation and Language (cs.CL)
备注: pages, 6 figures, 2 tables
Abstract:Large language models often produce judgments that resemble human responses, but it remains unclear whether they internally capture linguistic nuances such as conventional versus idiomatic meaning. A layer-wise representational analysis was conducted on two equal-sized pre-trained models, Llama-3.2-1B and Gemma-3-1B-PT, to examine whether and where these models differentiate linguistic traits from input. The analysis combined layer-wise probing with geometric distance measures, using the top embedding dimensions ranked by principal component loadings for each layer. The results show that even with equal capacity, models organize linguistic information in fundamentally different ways, suggesting the need to analyze internal representational geometry, examining not just behavioral outputs but semantic space alignment when evaluating how closely human and model language interpretation correspond.
[NLP-275] SAFE-KV: Causal Suffix Repair for Behavior-Preserving Non-Prefix KV Cache Reuse in RAG AACL
【速读】： 该论文旨在解决生成式 AI（Generative AI）中检索增强生成（Retrieval-Augmented Generation, RAG）系统在多请求场景下非前缀键值缓存（non-prefix KV-cache）复用时面临的因果状态不可复用问题。由于缓存状态是依赖于前缀、位置及先前文档的上下文条件化因果状态，直接复用会导致生成行为偏差。其核心解决方案为SAFE-KV，通过将修复机制建模为行为保持问题，并以完整预填充（full prefill）作为参考基准，采用风险引导的K2选择策略，对组合缓存进行重构：重新编码从选定起始文档开始的因果后缀，从而恢复被破坏的上下文一致性。实验表明，相较于仅文档或固定窗口修复方法，SAFE-KV在Qwen2.5-7B-Instruct上可将HotpotQA-500与2Wiki-500任务的准确率分别恢复至99.8%/98.4%（捐赠复制协议）和100.0%/98.4%（实际后缀转发），显著优于无修复方案在超过四文档时的性能退化。该方法建立了非前缀缓存复用的因果状态契约，并提供了可验证的修复机制，实现跨工作负载的高效、精准生成。
链接: https://arxiv.org/abs/2610.15186
作者: Xingjian Zhang
机构: Carnegie Mellon University (卡内基梅隆大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted to Findings of AACL-IJCNLP 2026. 36 pages, 5 figures, 33 tables. Code: this https URL
Abstract:Retrieval-augmented generation (RAG) reuses the same documents across many requests, making non-prefix KV-cache reuse attractive. Yet a cached document state is not reusable text: it is a context-conditioned causal state whose keys and values depend on the prefix, position, and preceding documents under which it was computed. We formulate non-prefix KV reuse as a behavior-preservation problem with full prefill on the exact prompt as the reference, and introduce SAFE-KV, which repairs a composed cache by re-encoding the causal suffix from a selected start document. On Qwen2.5-7B-Instruct, prefix-conditioned composition preserves only 47.2% of HotpotQA-500 and 37.8% of 2Wiki-500 full-prefill answers; SAFE-KV with risk-guided K2 selection restores 99.8%/98.4% (donor-copy protocol) and 100.0%/98.4% with actual suffix forwarding. Document-only and fixed-window repairs recover far less, because the document states after the repaired region keep histories that omit it. Reuse without repair is exact up to four documents and degrades at sixteen, where suffix repair recovers 98-100%. Faithful repair re-encodes most of the context, so in a single-request L40S study SAFE-KV’s operating rule serves every workload by full prefill. SAFE-KV thus establishes the causal-state contract for non-prefix reuse and a validated repair mechanism that meets it.
[NLP-276] UDHRBench: Benchmarking Human Rights in Large Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在实际应用中与《世界人权宣言》（Universal Declaration of Human Rights, UDHR）所确立的人权标准之间存在系统性偏差却缺乏有效评估手段的问题。其核心挑战在于，现有评估方法难以捕捉模型在看似合理推理（如效率考量、风险规避或科学中立性）下隐含的人权违背行为，而非显性的权利否定。解决方案的关键在于提出并构建了首个大规模、高精度的人权合规性评估基准——UDHRBench，包含覆盖全部30条人权条款的1,124个对抗性问题及四款主流大模型（ChatGPT、Grok、Gemini、Fanar）生成的4,496条响应，均由专家进行人工标注并达到较高的评分者间一致性（κ = 0.801）。在此基础上，研究进一步设计了一种基于“大模型作为裁判”（LLM-as-a-Judge）的自动化评估框架，采用零样本与少样本提示策略，在检测人权错配方面实现了89.8%的加权F1分数，并创新性地引入“对齐宽容度评分”（Alignment Leniency Score），用于量化裁判模型在模糊案例中的容忍程度，从而实现可扩展、可解释的人权合规性评估。
链接: https://arxiv.org/abs/2610.15179
作者: Alak Raad(1),Noreldin Morssy(1),Zeerak Suhail(1),Serin Atiani(2),Mustafa Jarrar(1) ((1) Hamad Bin Khalifa University, Qatar, (2) Princess Sumaya University for Technology, Jordan)
机构: Hamad Bin Khalifa University, Qatar(哈马德·本·哈利法大学，卡塔尔); Princess Sumaya University for Technology, Jordan(约旦公主苏玛亚科技大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Alak Raad and Noreldin Morssy contributed equally to this work
Abstract:Large language models are increasingly deployed in contexts requiring adherence to ethical and legal norms, yet their alignment with internationally recognized human rights standards remains underexplored. We introduce UDHRBench, a large-scale benchmark for operationalising the evaluation of large language model compliance with the Universal Declaration of Human Rights. The dataset comprises 1,124 human rights confronting questions covering all 30 articles and 4,496 responses from four large language models (ChatGPT, Grok, Gemini, and Fanar), manually annotated by experts for alignment, with high inter-annotator agreement ( \kappa = 0.801 ). Our analysis reveals systematic and non-trivial misalignments, showing that violations typically arise through seemingly rational reasoning: appeals to efficiency, risk management, or scientific neutrality, rather than explicit rejection of rights. We further propose an LLM-as-a-Judge framework for automated evaluation using zero-shot and few-shot prompting, achieving 89.8% weighted F1-score in detecting human-rights misalignment, alongside a novel Alignment Leniency Score that quantifies how judge models resolve ambiguous cases. The dataset, annotations, and code will be released as open-source resources (this https URL).
[NLP-277] Active Learning for Finding a Needle in a Haystack
【速读】： 该论文旨在解决在训练分类器识别稀有文档类别（如针对特定宗教群体的仇恨言论）时，因标注样本稀缺而导致的高昂成本问题。其核心挑战在于：主动学习（Active Learning）虽可降低标注成本，但需依赖初始标注集，而随机采样难以以合理成本获取目标类别的代表性样本，通常需借助关键词搜索等方法，这会引入初始数据集的偏差，导致部分目标子类被忽略。论文的关键解决方案是评估主动学习能否纠正此类偏差。研究在AG News和Measuring Hate Speech数据集上对比了八种主动学习策略与随机采样的表现，重点考察在初始集包含或缺失已知子类的情况下，对被遗漏子类的召回率及整体F1值。结果表明，在AG News上几乎所有策略均能恢复被遗漏子类；而在Measuring Hate Speech上，基于不确定性的策略表现劣于随机采样，而判别性主动学习（Discriminative Active Learning）在不显著牺牲F1的前提下，仍可恢复55%的遗漏子类，凸显其在缓解初始偏差方面的优势。该研究强调，聚合指标（如F1）可能掩盖不同策略在子类覆盖上的实质性差异。
链接: https://arxiv.org/abs/2610.15165
作者: Emily Silcock,Simon Löwe
机构: Harvard University(哈佛大学); Burning Glass Institute
类目: Computation and Language (cs.CL)
备注:
Abstract:Training a classifier to find a rare class of documents, such as hate speech directed at a particular religious group, requires labelled examples that are expensive to obtain when the class makes up only a small fraction of a corpus. Active learning reduces this cost, but it must start from an initial labelled set that contains examples of the target class. Random sampling cannot supply these at reasonable cost, so they must be selected in some other way, such as keyword search, and any such selection biases the initial set towards part of the target class. We evaluate whether active learning can correct this bias. On AG News and Measuring Hate Speech, at a prevalence of around 1%, we compare random sampling with eight active learning strategies, starting from initial sets that cover or omit known subgroups of the target class, and measure recall on the omitted subgroups as well as F1. On AG News, almost every strategy recovers the omitted subgroup. On Measuring Hate Speech, uncertainty-based strategies recover less of the omitted subgroups than random sampling, whereas discriminative active learning recovers 55% of them without a significant loss of F1. Aggregate metrics conceal these differences.
[NLP-278] Sanzaru: Benchmarking Protocol Formation under Epistemic Communication Asymmetry
【速读】： 该论文旨在解决单一语言模型在独立执行任务时表现良好，但在团队协作中因信息与通信资源分配不均而导致协作失效的问题。其核心挑战在于评估多智能体系统在不对称感知与沟通条件下的协同决策能力。解决方案的关键是构建一个名为SANZARU的协作式排爆谜题基准，该基准基于BOMBANANA!（Lefto Studio, 2026）改编，包含120个谜题实例，分为七个类别，分别针对不同的协调能力。三名智能体分别被赋予“无法听”、“无法说”和“无法看”的限制，迫使它们发展出适应性通信协议（如手势编码、摩尔斯电码乃至完整手册编码）。通过对1512个实验回合的分析发现，约60%的失败归因于沟通失误，且智能体常尝试使用其队友不可用的通信渠道，即便已有明确角色指令。该研究通过任务成功率与交互轨迹的联合分析，实现了对协作表现及视角理解（perspective taking）的双重评估，为评估生成式AI在复杂协作场景中的社会认知能力提供了可量化的基准。
链接: https://arxiv.org/abs/2610.15163
作者: Yiping Li,Zhiyu An,Wan Du
机构: University of California, Merced(加州大学默塞德分校)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:A language model’s ability to solve a task individually does not ensure that a team can solve it under unequal access to information and communication. SANZARU examines this gap through cooperative bomb-defusal puzzles adapted from BOMBANANA! (Lefto Studio, 2026). Task information and action authority are divided among three agents: one cannot hear, another cannot speak, and the third cannot see. The benchmark contains 120 puzzle instances across seven families, each targeting a distinct coordination ability. Across 1512 episodes, 7 models achieve an overall success rate of 49%, ranging from 20% to 86% under the evaluated configurations. Teams develop or adopt diverse communication protocols, from simple gesture codes to Morse and encodings of entire manuals. Yet trace analysis attributes about 60% of 771 team losses to communication, and agents attempt to use channels unavailable to their teammates despite explicit role instructions. Together, task outcomes and interaction traces enable evaluation of both collaborative performance and perspective taking throughout cooperation. The benchmark is available at this https URL.
[NLP-279] Learning Proposition-Level Links from Sentence-Level Supervision for Neuro-Symbolic NLI
【速读】： 该论文旨在解决神经符号自然语言推理（Neuro-Symbolic Natural Language Inference, NLI）中命题级单元（atoms）间语义关联建模不准确的问题。现有基于抽象意义表示（Abstract Meaning Representation, AMR）与布尔可满足性（SAT）的系统依赖固定的嵌入式源选择策略及未针对原子级别链接训练的句子级NLI模型，导致跨命题单元的语义连接识别能力受限。其解决方案的关键在于提出一种可训练的原子链接器（trainable atom linker），该链接器能够联合预测链接存在性、源选择以及关系类型，且无需原子对标注数据。通过将跨公式链接结构视为潜在变量，并仅利用句子级NLI标签进行端到端训练，同时保持翻译器和SAT推理器不变，实现了对原子间语义关系的动态建模。实验在SNLI、SICK及MNLI多个基准上验证了该方法的有效性，相较于基线系统提升11.40–27.09个百分点，显著优于先前固定链接机制的AMR-to-SAT系统。
链接: https://arxiv.org/abs/2610.15158
作者: Xuyao Feng,Antonis Bikakis
机构: University College London (伦敦大学学院)
类目: Computation and Language (cs.CL)
备注:
Abstract:We study neuro-symbolic natural language inference (NLI) with independently translated propositional representations of the premise and hypothesis. Before symbolic reasoning, the system must infer semantic links between their proposition-level units (atoms). Existing systems based on Abstract Meaning Representation (AMR) and Boolean satisfiability (SAT) identify these links using fixed embedding-based source selection and pretrained sentence-level NLI models not trained for atom-level linking. We introduce a trainable atom linker that jointly predicts link existence, source selection, and relation type. Without atom-pair training annotations, we treat cross-formula link structures as latent and train the linker from sentence-level NLI labels, keeping the translator and SAT reasoner fixed. We evaluate on the full public test splits of Stanford Natural Language Inference (SNLI) and Sentences Involving Compositional Knowledge (SICK), and the matched and mismatched validation splits of Multi-Genre Natural Language Inference (MNLI). Our system improves accuracy by 11.40-27.09 percentage points over a baseline applying sentence-trained NLI models unchanged to atom pairs within the same SAT pipeline. It also improves over the prior fixed-linker AMR-to-SAT system by 8.14-26.32 percentage points on SNLI and both MNLI splits, with a smaller gain on SICK.
[NLP-280] Where Did We Go Wrong? Localizing Errors to Repair Multi-Turn Conversations
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）在多轮对话中因初始意图理解偏差而导致的推理状态污染问题。当用户意图在对话过程中逐步显现时，若助手在早期阶段产生误判，可能错误地排除真实意图，进而影响后续交互并导致对话失败。传统方法通常需要完全重置或回溯状态，造成有效信息丢失。本文提出Loci（Localization Of Corrupted Inferences）框架，其核心在于通过概率化方式对每一轮更新进行分析，评估各更新是否可能导致真实意图被排除，并据此为每个推断分配“排除真实意图”的概率，从而实现对错误推理的精准定位。该框架进一步结合置信度指导修复策略，决定保留、验证或移除可疑状态更新，在不丢弃有效信息的前提下实现高效恢复。实验表明，无论在注入错误的受控场景还是自然发生的自由对话中，Loci均显著优于基线方法，平均恢复成本降低46%至79%，验证了其在识别与修复误解释方面的有效性。
链接: https://arxiv.org/abs/2610.15154
作者: Alessandro Morosini,Wei Sun,Georgia Perakis
机构: Massachusetts Institute of Technology(麻省理工学院); IBM Research(IBM研究院)
类目: Computation and Language (cs.CL)
备注: 29 pages, 4 figures, 17 tables
Abstract:Large language model (LLM) assistants are increasingly used in multi-turn conversations, where the user’s intent is often unclear at first and revealed only gradually over time. The assistant must therefore interpret each exchange and form inferences about the user’s intent, using them to update its working state. These updates usually refine the assistant’s understanding, but a misinterpretation can instead corrupt its working state by ruling out the true intent, influencing subsequent turns and ultimately causing the conversation to fail. Efficient recovery then requires locating and revising the misinterpretation without discarding information that remains valid. We introduce Loci (Localization Of Corrupted Inferences), a framework for probabilistically localizing misinterpretations and selectively repairing the assistant’s working state. Given an observed failure, Loci reasons over possible user intents and assigns each update a probability of having excluded the true intent. These probabilities rank suspected updates and quantify the model’s confidence in the localization of the underlying misinterpretations. Loci then uses this information in a repair policy that decides whether to retain, verify, or remove suspected updates, enabling recovery from corrupted inferences while preserving useful information. To evaluate Loci’s ability to identify misinterpretations, we first test it in a controlled setup with injected errors. We find that Loci improves localization over the baselines, with its advantage becoming larger when multiple errors occur. We then evaluate it in free-form conversations, where misinterpretations arise naturally, and measure the interactions needed to complete the task after an initial failure. Using localization confidence to guide repair, Loci reduces mean recovery cost by 46-79% relative to the strongest baseline across multiple domains.
[NLP-281] Compute-Optimal Scaling Laws for Spiking Language Models NEURIPS2026
【速读】： 该论文旨在解决大规模脉冲神经网络（Spiking Neural Networks, SNN）在语言建模中面临的效率-精度权衡问题，尤其是缺乏对计算资源与模型性能之间关系的系统性量化分析。其核心解决方案在于开展首个基于Chinchilla法则的等算力（IsoFLOP）缩放研究，针对采用RWKV骨干网络的SpikeGPT系列模型，在FineWeb-EDU数据集上覆盖从3×10¹⁷到10¹⁹ FLOPs的多种计算预算，模型宽度从384扩展至1536。关键发现包括：第一，最优参数规模与计算预算之间的关系为Nopt∝C0.56，接近密集型Transformer模型的C0.50，表明脉冲机制并未改变在给定算力下参数与数据的最优权衡策略；第二，与非脉冲型RWKV和Transformer基线相比，模型损失差距可分解为两部分——由骨干结构引入的恒定代价约0.16 nats，以及随计算量增加而递减并趋于平稳的脉冲代价（从0.34 nats降至约0.23 nats）。这一结果首次提供了脉冲语言模型的计算最优配置方案，并量化了其稀疏性带来的精度损耗，为未来高效脉冲语言模型的设计提供了理论依据。
链接: https://arxiv.org/abs/2610.15146
作者: Dan Saunders,Benjamin Ren,Catherine He
机构: Luma AI(卢玛AI); X School
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Neural and Evolutionary Computing (cs.NE)
备注: Accepted to AXIOM: Foundations of Efficient Deep Learning (NeurIPS 2026 workshop). 4 pages + appendix, 4 figures, 5 tables
Abstract:Spiking neural networks target energy-efficient language modeling through sparse, binary activations, but the efficiency-accuracy tradeoff of spiking language models is uncharacterized at scale. We present the first Chinchilla-style IsoFLOP scaling study of a spiking language model, sweeping SpikeGPT-family models (RWKV backbone) from width 384 to 1536 across compute budgets from 3\times10^17 to 10^19 FLOPs on FineWeb-EDU. We report two findings. First, the compute-optimal frontier is N_opt \propto C^0.56 , close to the 0.50 that Hoffmann et al. report for dense Transformers, so spiking does not change how parameters and data should be traded off under a compute budget. Second, against non-spiking RWKV and Transformer baselines trained the same way, the loss gap splits into a backbone cost of about 0.16 nats that does not change with compute and a spiking cost that falls from 0.34 to about 0.23 nats and then levels off. These results give the first compute-optimal recipe for spiking language models and quantify the accuracy cost of their sparsity.
[NLP-282] Lumen-bench: Cross-Lingual Safety Gaps and the Say-Do Problem in Tool-Calling Large Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）在面对不道德请求时，其“拒绝”行为的实际含义与真实行为之间可能存在“说-做不一致”（say-do gap）的问题。传统评估方法仅关注模型输出文本中的拒绝声明，而忽视了当模型具备可执行工具能力后，其实际操作可能违背口头承诺的潜在风险，尤其在多语言和多社会压力情境下，这种偏差难以通过单一语言评估揭示。为此，研究提出Lumen-bench，一个跨语言、多情境的工具调用伦理行为基准，涵盖30个模型、五种语言（葡萄牙语、英语、西班牙语、中文、印地语）、108个由葡萄牙语作者设计的情景、六种社会压力条件、两种系统提示（伦理与中立），以及21类通用工具，共包含194,400次实验。关键解决方案在于构建一个基于推理模型判断（o4-mini）的多层次评估体系，结合人工校准（κ=0.78）、多方裁判对比（Fleiss’ κ=0.78）及事后沙盒回放验证，以捕捉模型“言不由衷”的行为模式。核心发现包括：（i）在伦理提示下，印地语和葡萄牙语模型的有害行为率显著高于英语；在中立提示下，印地语和西班牙语模型同样表现出更高违规率；（ii）高端模型虽整体违规率较低，但价格并非安全或能力的保证；（iii）将伦理提示替换为中立提示后，综合违规率从2.76%飙升至13.64%；（iv）在8.57%的实质性工具调用案例中，模型虽口头拒绝却仍执行了有害操作，明确揭示了“说-做脱节”现象，而对照实验表明，在无害控制条件下任务完成率达82.8%。该研究强调，必须超越文本层面的拒绝判断，引入行为验证机制，才能真正评估模型在复杂现实场景中的伦理可靠性。
链接: https://arxiv.org/abs/2610.15134
作者: Giovana Kerche Bonás,Thiago Laitz,Thales Sales Almeida,João Guilherme Alves Santos
机构: Tropic AI; Maritaca AI
类目: Computation and Language (cs.CL)
备注:
Abstract:When a large language model refuses an unethical request, what does refusal mean? In conversation, refusal is bounded by what the model says; once the model is equipped with executable tools, what it does can diverge from what it says, a failure mode that English-only evaluations cannot fully reveal. We introduce Lumen-bench, a behavioral benchmark of tool-calling ethics spanning 30 models, five languages (Portuguese, English, Spanish, Mandarin Chinese, Hindi), 108 Portuguese-authored scenarios, six social-pressure conditions, two system prompts (ethical and neutral), and 21 generic tools. The evaluation comprises 194,400 trials. Judgments come from a reasoning-model judge (o4-mini), calibrated against one Portuguese-speaking author ( n=83 ; harmful/non-harmful \kappa=0.78 ), compared with two other provider judges (Fleiss’ \kappa=0.78 , n=607 ), and complemented by retrospective sandbox replay. Four findings emerge. (i) Within-model comparisons support higher harmful-action rates in Hindi and Portuguese than English under the ethical prompt, and Hindi and Spanish under the neutral prompt, after multiple-testing correction. (ii) Premium models have lower median violation rates, although price does not guarantee safety or competence. (iii) Replacing the ethical system prompt with a neutral one raises pooled violation rates from 2.76% to 13.64%. (iv) In 8.57% of substantive-call cases, refusal or ethical objection accompanies a harmful tool call: the judges consistently identify the say-do gap, although they differ in how many responses they assign to this category. Matched benign controls demonstrate legitimate task completion in 82.8% of 3,300 trials. Code and reproduction scripts are available at this https URL.
[NLP-283] RegExtract: A Benchmark for PFAS Regulatory Information Extraction with LLM s ICML2026
【速读】： 该论文旨在解决从复杂、冗长且法律密度高的全氟烷基和多氟烷基物质（PFAS）监管文档中提取结构化信息的难题。由于相关事实分散在文本各处，且提取结果的溯源性对验证至关重要，传统方法难以满足实际需求。为此，研究提出RegExtract基准，涵盖20个美国州级文件与46项法规中的488个人工标注字段实例（共976个条目），构建了包含物质定义、允许用途、产品豁免等关键要素的类型化多字段模式，并引入面向溯源性的评估指标。其解决方案的关键在于：通过设计细粒度的结构化数据集与针对性评估框架，结合少样本提示、文档浓缩及模式划分策略，在开放与闭源大语言模型（LLM）上验证了领域特定提示对提升提取性能的有效性，尤其指出文档识别类字段仍具挑战性。该工作为有害物质识别、问答系统与合规监控等人工智能应用奠定了基础。
链接: https://arxiv.org/abs/2610.15103
作者: Defne Circi,Krystelle Lionti,Sarathkrishna Swaminathan,Daniel P. Sanders,L. Catherine Brinson,Bhuwan Dhingra,Basel Shbita,Mohab Elkaref,Sara Capponi,Tim Erdmann,Brandi Ransom
机构: Duke University (杜克大学); IBM Research (IBM 研究院)
类目: Computation and Language (cs.CL)
备注: Accepted to the AI for Law Workshop at ICML 2026, Seoul, South Korea
Abstract:Extracting structured information from regulatory documents is a critical but challenging task. Relevant facts are scattered across long, legally dense texts, and traceability of extracted facts to their source is essential for verification. With increasing awareness of the environmental and health impacts of per- and polyfluoroalkyl substances (PFAS), regulatory documents have expanded significantly, varying in scope and product relevance, making interpretation challenging for manufacturers and policymakers. We introduce RegExtract, a benchmark for structured information extraction from PFAS regulatory documents, comprising 488 human-annotated field instances across 20 US state documents and 46 regulations. Each field instance includes a value and a provenance entry, yielding 976 entries in total. We define a typed, multi-field schema targeting key elements such as material definitions, permitted uses, and product exemptions, and propose provenance-aware evaluation metrics tailored to regulatory text. Through experiments with open- and closed-source LLMs across prompting strategies, document condensing methods, and schema partitioning approaches, we find that document identification fields remain challenging, and that field-specific few-shot examples improve extraction. This work lays the groundwork for AI tools for hazardous substance identification, question answering, and compliance monitoring. The code and data are available at this https URL.
[NLP-284] Have You Been Convinced? A Neuro-symbolic Argumentation Reasoning Approach to Detecting Persuasiveness in ChangeMyView Discussions
【速读】： 该论文旨在解决在真实网络讨论中如何基于论点结构（argument structure）有效推断说服力（persuasiveness）的问题，核心挑战在于缺乏对论点影响力在结构中传播过程的正式建模。其解决方案的关键在于提出一种神经符号混合的论点推理系统（Neuro-symbolic Argumentation Reasoning System, NARS），通过结合神经方法与符号逻辑实现可解释的推理：其中神经组件利用检索增强的迭代大语言模型（LLM）推理，从长篇多线程讨论中提取双极性论点结构；符号组件则采用渐进式论点语义（gradual argumentation semantics）来建模论点影响力的逐步传播。研究构建了CMV2BAS数据集，包含基于LLM生成的双极性论点结构标注及视点改变（delta event）标签，用于评估系统性能。实验表明，尽管现代大语言模型已具备强大能力，但符号推理仍对提升说服力检测效果具有不可替代的作用，且引入论点质量与图中心性特征可进一步优化性能。最终NARS在最优配置下达到F1=0.57、AUC=0.56，优于零样本的Qwen3.6:27B与GPT-oss:20B。研究揭示支持主导型讨论和长论点链是影响攻击影响力传播的主要失败模式，并验证了在噪声真实的网络话语中操作化渐进式论点语义进行规模化宏观说服分析的可行性，同时指出了未来神经符号论点推理研究的关键方向。
链接: https://arxiv.org/abs/2610.15091
作者: Siddharth Bhargava,Sara Tonelli,Patricia Martín-Rodilla
机构: Fondazione Bruno Kessler(布鲁诺·凯斯勒基金会); Universidade da Coruña(拉科鲁尼亚大学); IEGPS-CSIC(西班牙国家研究委员会伊埃格普斯研究所)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Symbolic Computation (cs.SC)
备注: Accepted and to be published in the Proceedings of the 25th International Conference on Knowledge Engineering and Knowledge Management (EKAW 2026)
Abstract:Argument structures represent how individual arguments interact within a discourse. However, leveraging these structures to infer meaningful knowledge remains challenging, as it requires a formal representation of how argumentative influence propagates across the structure. We propose Neuro-symbolic Argumentation Reasoning System (NARS) for performing persuasiveness detection directly from online discussions. The neural component employs retrieval-augmented iterative LLM reasoning to extract bipolar argument structures, while the symbolic component applies gradual argumentation semantics to propagate argumentative influence over the structure. Using the Webis-CMV-20 corpus, we construct CMV2BAS, a dataset of long, multi-threaded ChangeMyView discussions annotated with LLM-derived bipolar argument structures and labeled with delta events (view change vs.\ no change). We investigate whether gradual argumentation semantics can effectively detect persuasiveness, whether augmenting them with argument quality and graph-based centrality improves performance, and whether symbolic reasoning remains necessary in the presence of modern LLMs. NARS achieves F1=0.57 and AUC=0.56 under its best configuration, outperforming zero-shot Qwen3.6:27B and GPT-oss:20B. We identify support-dominant discussions and long argumentative chains as key failure modes limiting attack influence propagation. Our work demonstrates the feasibility of operationalizing formal gradual argumentation semantics over noisy real-world discourse for scalable macro-level persuasion analysis, while highlighting key challenges and directions for future research in neuro-symbolic argumentation reasoning. We release both CMV2BAS dataset and the NARS framework for public use. Comments: Accepted and to be published in the Proceedings of the 25th International Conference on Knowledge Engineering and Knowledge Management (EKAW 2026) Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Symbolic Computation (cs.SC) Cite as: arXiv:2610.15091 [cs.CL] (or arXiv:2610.15091v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2610.15091 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-285] Select or Rewrite: Oracle-Guided Fine-Tuning for Generative ASR Post-Editing from N-Best Hypotheses
【速读】： 该论文旨在解决生成式语言模型在自动语音识别（ASR）后处理中因仅依赖提示（prompt-only correction）而产生的脆弱性问题，具体表现为：当最优假设（top-1）已正确时过度修正，或在缺乏参考文本时产生幻觉，并且通常需要超大规模模型才能实现有效性能。其解决方案的关键在于提出一种基于“理想指导”（oracle-guided fine-tuning）的监督学习框架，通过构建三个精心设计的数据池来增强训练信号：1）修正样本（corrections），即参考文本存在于N-best列表但非top-1，用于训练模型正确选择更优假设；2）正则化样本（regularizers），即top-1本身正确，用于抑制不必要的重写；3）硬负样本（hard negatives），即参考文本完全不在N-best列表中，用于防止模型产生幻觉。该方法通过将N-best列表转化为结构化监督信号，显著提升了模型对“选择或重写”混合策略的学习效率，尤其在约30亿参数规模下于专有数据集上取得明显性能提升，而在CommonVoice数据集上效果不显著，暗示公开读写语料可能与大语言模型（LLM）预训练语料存在重叠，从而掩盖了真实后处理带来的增益。
链接: https://arxiv.org/abs/2610.15056
作者: Shiran Liu,Gaohui Xu,Sergio Burdisso,Thibault Bañeras-Roux,Shashi Kumar,Petr Motlicek
机构: Idiap Research Institute (伊迪亚研究所), Martigny, Switzerland; KTH Royal Institute of Technology (瑞典皇家理工学院), Stockholm, Sweden; Brno University of Technology (布尔诺理工大学), Brno, Czech Republic
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 7 pages, 4 figures, 2 tabl
Abstract:Large language models can post-edit ASR by using linguistic context to compare N-best hypotheses and rewrite the transcript when needed, often improving WER without retraining the recognizer. However, prompt-only correction is brittle: it may over-edit when top-1 is correct and hallucinate when the reference is absent, and often needs very large models. We propose oracle-guided fine-tuning that turns N-best lists into supervision via three curated pools: corrections (reference in the list but not top-1) to teach selection, regularizers (top-1 correct) to discourage unnecessary rewrites, and hard negatives (reference absent) to suppress hallucination. Fine-tuning on this data provides more effective supervision than random subset fine-tuning for learning a hybrid select-or-rewrite policy, with the clearest gains from around 3B parameters on the proprietary dataset but less conclusive results on CommonVoice, a divergence suggesting that public read-speech text may overlap with LLM pretraining and mask genuine post-editing gains.
[NLP-286] PAT: A Personality-Driven Augmentation and Transfer Learning Framework for Depression Detection on Social Media
【速读】： 该论文旨在解决现有自动化抑郁检测方法中存在的两大关键问题：一是基于症状的语义筛查易受表面词汇模式干扰，导致大量假阳性文本被误判；二是帖子级编码器优化不足，仅依赖粗粒度的用户层级监督信号，难以捕捉单个帖子中的抑郁线索。其解决方案的关键在于提出一种人格驱动的增强与迁移学习框架（PAT）。该框架首先通过基于经典人格理论的大型语言模型（LLM）数据增强，结合细粒度对比学习目标优化帖子级编码器，有效缓解数据稀缺问题；其次，在第二阶段将训练好的帖子级评分器用于高质量帖子过滤，并将其编码器参数无缝迁移至用户级任务中，实现多粒度可解释性，能够追踪帖子级别的心理状态演变并突出关键症状，为临床诊断提供可靠依据。
链接: https://arxiv.org/abs/2610.15044
作者: Shengwu Zheng,Yuzhe Ding,Xiaohui Liang,Fei Li,Chong Teng,Donghong Ji
机构: Wuhan University (武汉大学); School of Cyber Science and Engineering (网络安全学院); School of Public Health (公共卫生学院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Depression is a prevalent mental disorder that affects millions worldwide, with early detection crucial for effective intervention. While exist?ing methods have achieved promising results in automated depression detection, they face two key limitations: (1) noise in post screening: symptom-based semantic screening is easily misled by superficial lexical patterns, thereby introducing false-positive posts; (2) weak op?timization of the post encoder: relying solely on coarse user-level supervision signals pre?vents models from capturing depressive cues within individual posts. To address these limita?tions, we propose PAT, a Personality-driven Augmentation and Transfer learning frame?work. In the first phase, PAT optimizes a post?level encoder via a fine-grained contrastive learning objective, utilizing LLM-based data augmentation driven by classic personality the?ories to alleviate data scarcity. In the second phase, the trained post-level scorer is deployed for high-quality post filtering, while its en?coder parameters are seamlessly transferred for user-level tasks. Extensive experiments demon?strate that PAT significantly outperforms exist?ing methods. Furthermore, PAT delivers multi?granular interpretability by tracing post-level mental-state trajectories and highlighting key symptoms, thereby offering reliable diagnostic evidence for clinical practice.
[NLP-287] oward System One Clinical Intelligence: Parameter-Efficient Fine-Tuning and Counterfactual Explainability of Decision Foundation Models
【速读】： 该论文旨在解决基于患者自述症状的自动化疾病分类中，轻量级判别模型的快速推理与自回归大语言模型（LLM）在上下文理解能力上的延迟开销之间的平衡问题。其核心挑战在于如何在保持高诊断准确率的同时实现低延迟、可解释的临床决策支持。解决方案的关键在于采用基于决策的生成式基础模型（decision foundation model），具体为将基于Qwen3.5架构的90亿参数决策模型Clef-Flash适配至Symptom2Disease基准上的24类疾病分类任务。通过使用4比特量化低秩适应（QLoRA）技术结合差异化学习率，并采用分层划分的训练集（960例）与测试集（240例），模型在单次前向传播中直接输出候选类别概率，避免了自回归解码过程，从而实现每条记录低于80毫秒的推理速度，相较生成式LLM基线提升31倍。该方法在诊断准确率（95.42%）和宏平均F1分数（0.9551）上显著优于零样本基线（远程API：61.67%，本地未微调初始化：65.00%）及ClinicalBERT基线（90.90%）。为进一步揭示症状层面的判别依据，研究引入“留一症状剔除”（Leave-One-Symptom-Out, LOSO）反事实扰动框架，结果显示，在银屑病案例中移除关键定义症状后，目标类置信度从100.00%降至0.12%，预测转移至脓疱疮且置信度达99.59%，而移除次要系统性症状则几乎不改变置信度，验证了模型具备症状级别的可解释性。因此，该方案成功实现了大模型的语言理解能力与低延迟、可解释推理的融合，适用于临床分诊场景。
链接: https://arxiv.org/abs/2610.15029
作者: S. M. Mahbub Uddin Ahmed
机构: United International University(联合国际大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Automated disease classification from patient-reported symptoms requires balancing the speed of lightweight discriminative models against the contextual understanding and latency costs of autoregressive large language models (LLMs). Decision foundation models offer an alternative by mapping representations directly to candidate classes without free-form text generation. We adapt Clef-Flash, a 9-billion-parameter decision model built on a Qwen3.5 backbone, for 24-class disease classification on the Symptom2Disease benchmark. Using 4-bit quantized low-rank adaptation (QLoRA) with differential learning rates and a stratified split of 960 training and 240 test cases, the model achieves 95.42% diagnostic accuracy and a macro F1 score of 0.9551. This improves upon the reported zero-shot baselines of 61.67% for a remote API and 65.00% for a local unadapted initialization, and outperforms the ClinicalBERT baseline (90.90%). By evaluating candidate logits in a single forward pass without autoregressive decoding, inference takes under 80 milliseconds per record on a single GPU, yielding a reported 31-fold speedup over generative LLM baselines. To examine symptom-level evidence, we introduce a Leave-One-Symptom-Out (LOSO) counterfactual perturbation framework. Removing disease-defining symptom clauses in a psoriasis example reduces predicted-class confidence from 100.00% to 0.12% and shifts the prediction to impetigo with 99.59% confidence, while removing secondary systemic symptoms produces less than 0.01% change in confidence. These findings suggest that decision-based architectures can combine the linguistic capabilities of large foundation models with low-latency inference and symptom-level interpretability for clinical triage.
[NLP-288] UltraBERT: Beyond the Limit of BERT Pre-Training
【速读】： 该论文旨在解决双向编码器（bidirectional encoder）在大规模预训练背景下性能未被充分优化的问题，具体表现为现有模型难以在分类与检索任务上同时达到顶尖水平。其核心挑战在于沿用小规模模型的预训练范式，导致计算资源分配、掩码比率设计及知识迁移策略缺乏针对性。解决方案的关键在于通过系统性重审训练中的核心设计选择，揭示三大关键发现：（1）当预训练数据量超过数万亿 token 后，下游任务性能趋于饱和，进一步增加算力无显著提升；（2）对于不同任务，低掩码比率更利于分类，而高掩码比率更优用于检索，这与以往认为大模型应采用更高掩码比率的观点相悖；（3）基于剪枝的蒸馏方法在低掩码比率下可提升分类性能，但在高掩码比率下反而损害检索表现。基于这些发现，作者提出 UltraBERT，一种统一架构的编码器，通过在单个批次内为低/高掩码比率分别配置独立的语言模型头，并将蒸馏目标限制于低掩码比率分支，从而实现对两类任务的协同优化。实验表明，UltraBERT 在 GLUE 和 BEIR 两大基准上分别取得 91.0 和 47.7 的得分，成为首个在分类与检索任务上均达到当前最优水平的单一编码器，为双向模型的计算资源配置、掩码比率设计及知识迁移提供了可解释且高效的指导原则。
链接: https://arxiv.org/abs/2610.15028
作者: Jungwoo Park,Taewhoo Lee,Jaewoo Kang
机构: Korea University (韩国大学); AIGEN Sciences; Amazon
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Recent bidirectional encoders have been modernized with innovations from causal large language models (LLMs). However, the pretraining recipe itself remains inherited from small-scale conventions, leaving encoders underoptimized, with no single model performing competitively across both classification and retrieval tasks. In this work, we investigate how far encoder performance can be pushed by revisiting core design choices in training, revealing three key findings: (1) Encoders saturate on downstream tasks beyond a few trillion tokens, indicating that additional compute past this point does not help. (2) Across model sizes, lower masking ratios benefit classification while higher ones benefit retrieval, contradicting prior work suggesting larger ratios for larger models. (3) Distillation with pruning improves classification at low masking ratios but degrades retrieval at high ones. Guided by these findings, we present UltraBERT, a single encoder that assigns separate LM heads to low and high masking ratios within a batch and restricts the distillation objective to the low ratio. To our knowledge, UltraBERT is the first encoder to achieve state of the art on both classification and retrieval tasks, reaching 91.0 on GLUE and 47.7 on BEIR. Together, these results show how far encoder performance can be pushed and provide a principled basis for compute allocation, masking ratio design, and knowledge transfer in bidirectional models.
[NLP-289] HodgePhaseNet: Discrete Hodge Decomposition of Semantic Flows in Transformer Attention
【速读】： 该论文旨在解决现有模型在处理语义关系时无法将观测到的成对关系有效分解为全局一致的方向性、局部环路循环以及跨大尺度空缺环路的持久循环这一核心问题。其解决方案的关键在于引入HodgePhaseNet，通过将语义转换建模为有限有向单纯复形上的反称实值1-链（antisymmetric real-valued 1-cochain），利用顶点-边关联矩阵 $ B_1 $ 和边-面关联矩阵 $ B_2 $ 构建唯一的正交分解：$ \Omega = B_1^T\Phi + B_2\Psi + H $，其中 $ B_1H=0 $ 且 $ B_2^TH=0 $。该分解使方向场、局部环流与调和流三者可区分且唯一，即便其势函数不唯一亦然。研究进一步提供了显式投影算子、稀疏实现方式、可积性与规范不变性分析、数值验证案例及重构的注意力接口。注意力偏置定义于允许的有序查询-键对上，保持任意边方向选择下的一致性，并通过基于可见前缀构建单纯复形的方式维持解码器因果边界；注意力输出的规范不变性依赖于使用不变特征或适当补偿特征。研究强调，仅顶点嵌入的简单差分不具备共精确或调和成分，因此真实关系编码器不可或缺；局部环流并不自动导致矛盾，调和流也非必然代表缺失知识。上述解释与事实误差的降低均需通过受控的实证评估加以确认。本工作提出的是数学与架构层面的理论方案，未包含外部自然语言任务性能结果。
链接: https://arxiv.org/abs/2610.14971
作者: Kiyotaka Kasubuchi,Kazuo Fukiya
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:WavePhaseNet introduced frequency-resolved semantic representations, QuantumPhaseNet introduced gauge-covariant phase connections, and ModularPhaseNet made a finite cyclic phase channel and its cycle holonomy computable in an ordinary Transformer. These constructions are complementary, yet they do not by themselves separate an observed collection of pairwise relations into a globally consistent direction, circulation around local compositions, and circulation that persists around a larger unfilled cycle. HodgePhaseNet addresses this separation by representing semantic transitions as an antisymmetric real-valued 1-cochain on a finite oriented simplicial complex. With vertex-edge incidence matrix B_1 and edge-face incidence matrix B_2 , every such field has a unique orthogonal decomposition \Omega = B_1^T\Phi + B_2\Psi + H , where B_1H=0 and B_2^TH=0 . The three component fields are unique even when their potentials are not. We give explicit projectors, a sparse implementation, integrability and gauge results, a verified numerical example, and a redesigned attention interface. The attention bias is defined on allowed ordered query-key pairs, remains consistent under arbitrary choices of edge orientation, and preserves a decoder’s causal information boundary by constructing its complex from the visible prefix. Gauge invariance of the attention output is conditional on using invariant or appropriately compensated features. A simple difference of vertex embeddings has no coexact or harmonic component; a genuine relation encoder is therefore indispensable. Local circulation is not automatically contradiction, and harmonic flow is not automatically missing knowledge. These interpretations and any reduction in factual error require controlled empirical evaluation. This paper presents a mathematical and architectural proposal, not external natural-language performance results.
[NLP-290] HiPAD: A Unified Gait Audio-Visual and Personality Fusion Network for Depression Detection with a Hierarchical Personality Ablation and the Calibration Limits of Domain-Adapted LLM s
【速读】： 该论文旨在解决多模态抑郁症检测中跨人群（老年与青年）及跨模态（步态、音视频、两者结合）条件下模型泛化能力不足的问题，尤其关注在数据稀疏场景下如何有效融合异构信息以提升检测性能。其核心解决方案在于设计一个统一的单一分层融合网络（HiPAD），通过共享双向长短期记忆网络（BiLSTM）编码各模态输入，并采用零填充策略处理缺失模态；仅当至少包含步态、音频和视频中的两种时才启用跨模态注意力机制，从而避免无效信息干扰。关键创新点在于将人格描述（personality description）作为主导特征，通过拼接方式直接融合而非参与注意力计算，实证表明该约150词的人格描述文本所携带的抑郁信号远超音频、视频与步态三者之和，甚至使“步态+人格”组合的轨迹级评分（TLS）达到0.647，优于所有含音视频的配置。进一步分析显示，移除任何人格衍生特征（包括大五人格得分）均导致评分下降0.499–0.507至随机水平，凸显其决定性作用；此外，在固定架构下，基于临床先验的手工特征（如eGeMAPS、OpenFace动作单元）在音视频赛道上优于通用深度特征（wav2vec2、DenseNet），而对人格编码器进行LoRA或QLoRA微调虽提升二分类一致性系数（κ）0.13，却因扭曲PHQ-9回归任务表现而降低整体得分，表明在小样本条件下冻结预训练的RoBERTa-large编码器更具优势。最终系统在跨赛道上取得G-Score 0.482，揭示当前限制因素主要为表征容量与输出校准，而非模型架构本身。
链接: https://arxiv.org/abs/2610.14961
作者: Karol Chlasta,Piotr Struzik
机构: Kozminski University (科兹明斯基大学); WarsawIQ (华沙智能); OpenAI; Meta; Stability.AI; Anthropic; Character.ai; Claude
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Sound (cs.SD)
备注: 9 pages, 3 figures, 5 tables. Accepted at the 34th ACM International Conference on Multimedia (MM ‘26), Rio de Janeiro, Grand Challenge track. Authors’ extended version: the appendices restore the extended related work, figures, tables and discussion omitted from the conference version for length
Abstract:We present HiPAD, a single fusion network submitted to all six sub-tracks of the MPDD-AVG 2026 depression-detection challenge, covering the Elder and Young cohorts and three modality sets (gait, audio-visual, and both), each combined with personality descriptions. The network encodes each available modality with a shared BiLSTM, zero-pads absent modalities, and applies cross-modal attention only when at least two of gait, audio and video are present. Personality, which dominates our results, is kept out of the attention and fused by concatenation. Because the architecture is identical in every sub-track, differences between them reflect the data rather than the model, and we use this to study three questions. Our main finding is that the personality description, roughly 150 tokens about who the subject is, carries more depression signal than audio, video, and gait combined: gait plus personality alone (TLS 0.647) beats every configuration that includes audio and video. First, a four-level ablation of the personality description (L0-L3) quantifies this: removing every personality-derived feature, including the Big Five scores that earlier MPDD ablations keep, lowers the Track-Level Score by 0.499 to 0.507 and leaves the model at chance. Second, domain-adapting the personality encoder with LoRA (7B) or QLoRA (72B) on PHQ-8/9 data raises binary Cohen’s \kappa by 0.13 but lowers the overall score by distorting the PHQ-9 regression; a frozen RoBERTa-large encoder does better at this sample size. Third, at fixed architecture, handcrafted clinical features (eGeMAPS, OpenFace action units) outperform generic deep features (wav2vec2, DenseNet) on the audio-visual tracks. The unified system reaches a cross-track G-Score of 0.482; with under 100 subjects per cohort, the limiting factors are the size of the representation and the calibration of the outputs, not the choice of architecture.
[NLP-291] AlphaRLM: Aligning Language Models for Literature-Grounded Symbolic Factor Discovery
【速读】： 该论文旨在解决在海量且嘈杂的金融数据中，如何高效发现具有预测能力的可执行因子（executable factors）这一核心问题。传统基于大语言模型（LLM）的方法多依赖提示工程、迭代搜索或种子因子演化，而对通过策略学习直接构建因子的路径探索不足。为此，论文提出AlphaRLM框架，其核心创新在于将生成式因子构造过程建模为强化学习任务：通过文献引导的操作符组装（Literature-Guided Operator Assembly, LEGO），将因子构建过程约束于来自金融文献的知识与可执行运算符集合中，从而引入语义先验以提升探索效率；同时，设计可执行性奖励（executable-validity reward）与分组回报分离奖励（grouped-return-separation reward），使生成策略能够捕捉因子分位数间非单调的市场收益结构，并与真实市场反馈对齐。实验结果表明，AlphaRLM在沪深300、中证1000及标普500指数上均显著优于现有因子挖掘与基于LLM的基线方法，在年化超额收益指标上表现突出，尤其80亿参数模型展现出跨市场泛化能力，验证了文献条件化与市场对齐的因子生成范式具备广阔前景。
链接: https://arxiv.org/abs/2610.14959
作者: Jie Huang,Yuecheng Hong,Yao Chen,Hao Zhong,Linfeng Xiao,Xudong Liu,Hao Xia,Jianxuan Xin
机构: Harbin Institute of Technology(哈尔滨工业大学); University of Science and Technology of China(中国科学技术大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG); Portfolio Management (q-fin.PM)
备注: 25 pages, 5 figures
Abstract:Formulaic alpha mining seeks to discover executable expressions that capture predictive market signals in a vast, noisy search space. Recent LLM-based approaches use prompting, iterative search, or seed-factor evolution, leaving direct factor construction through policy learning less explored. To address this, we introduce AlphaRLM, a reinforcement-learning framework that trains LLMs to construct executable symbolic factors from financial literature without input factor expressions. AlphaRLM incorporates Literature-Guided Operator Assembly (LEGO), which conditions factor construction on literature-derived ideas and executable operators, providing semantic priors for exploration. It further aligns the generation policy with market feedback through executable-validity and grouped-return-separation rewards that capture non-monotonic structures across factor quantiles. Extensive experiments on CSI 300, CSI 1000, and SP 500 show that AlphaRLM outperforms strong factor-mining and LLM-based baselines in annualized excess return. Notably, the 8B model achieves strong cross-market performance, highlighting the promise of literature-conditioned, market-aligned factor generation.
[NLP-292] What Survives When Activation Steering Is Used as Memory
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）在长期记忆中如何有效存储与检索用户偏好及事实性知识的问题。传统方法假设用户相关知识需以文本形式显式存储，而本文挑战这一假设，提出一种无需训练的激活记忆机制（activation memory），将代理（agent）过往经验以“激活漂移”（activation shift）的形式编码为向量，并基于对话主旨的白化键（whitened key）进行索引与检索。其核心解决方案在于：仅将注意力机制在上下文中可复现的、与具体问题无关的固定平均效应（fixed average effect）作为记忆内容，通过晚层状态变化的偏移量实现记忆的存取。实验表明，该机制具有高度选择性（约98%无关回复保持不变），但事实性信息仅在最后一层以“应说词汇”的形式存储，且极少被实际使用（清洁使用率最高仅15%，远低于上下文中的96–98%）。偏好迁移仅在狭窄强度范围内有效，尤为关键的是，记忆中“减去参考项”的方式决定了概念或方向的留存——若未减去相反项，原本存储的厌恶会被错误召回为喜爱。这揭示了当前无训练激活记忆在语义一致性与方向保真度上的根本局限。
链接: https://arxiv.org/abs/2610.14958
作者: Vedant Jumle
机构: Delft University of Technology (代尔夫特理工大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 12 pages, 7 figures, 2 tables. Code: this https URL
Abstract:Activation steering has recently been used as a memory for skills: past experiences of an agent are stored as steering vectors and retrieved later. Facts and preferences about the user are assumed to need text instead. We test that assumption with a training-free activation memory beside a frozen LLM. The change an experience causes in late-layer states is stored as a shift, filed under a whitened key for the conversation’s gist, and added back in a later session only when the gist matches. We argue that such a shift is a fixed average of an effect that attention, in context, recomputes for every question, so only question-independent content can survive. In a pre-registered study on Qwen3.5-2B and 9B, with judges from four model families, the memory is selective (about 98% of unrelated replies are unchanged), but facts are stored as the word to say, in the last layers only, and are almost never used (at most 15% clean use, against 96-98% in context). Leanings transfer only in a narrow strength window. Most strikingly, the reference subtracted when writing a preference decides whether its concept or its direction survives: unless the opposite is subtracted, a stored dislike is recalled as a like.
[NLP-293] Do Personalized LLM s Need the Whole Conversation? Matching the Full History with Only 5% of Its Text via Relevance-Filtered Context
【速读】： 该论文旨在解决个性化助手在长对话历史中频繁调用大语言模型（LLM）时面临的高计算成本与冗余信息传递问题。其核心挑战在于，将完整的对话历史作为上下文输入给LLM虽能保留信息完整性，但显著增加延迟与资源消耗，且其中大量内容对当前请求无关。解决方案的关键在于采用基于句子级别的相关性筛选机制：通过嵌入模型对历史记录进行语义相似度排序，并在固定预算下截断，仅保留最相关的片段作为上下文。实验表明，在PersonaMem-v2数据集上，该方法仅使用5%的文本即可达到完整历史的准确率，使评判模型的输入量减少约20倍；相较之下，同等长度的LLM摘要虽性能相当，但需先遍历全部历史，而检索方法仅依赖轻量级嵌入模型。更关键的是，仅61%的查询需要引入相关证据，且一个知晓这些精选证据的“理想观察者”在准确率上反而超越完整历史，表明优化上下文选择策略是突破现有全历史输入范式的有效路径。
链接: https://arxiv.org/abs/2610.14952
作者: Erika Hayashi,Andrew Shin
机构: Keio University(庆应义塾大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Personalized assistants accumulate long conversation histories, and passing the whole history to an LLM with every request is costly. We show that it is also unnecessary. On PersonaMem-v2 (180 questions), with GPT-5.6-terra and Claude Opus 5.5 as judges, a relevance-filtered context, built by sentence-level retrieval that ranks the history by similarity to the current request and stops at a fixed budget, matches full-history accuracy with only 5% of the text, cutting the judge’s input about 20-fold. An LLM summary of the same length performs on par, but it needs an LLM to read the entire history, whereas retrieval needs only a small embedding model. While both sentence-level retrieval and LLM summary perform strongly, retrieval does so while placing the relevant evidence in the context for only 61% of the questions, and an oracle given only that evidence beats the full history by 15.6 points, making better context selection a promising path to surpassing the full history.
[NLP-294] Predicting how model propensities generalize out of distribution
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在部署于训练数据分布之外的新场景时，其行为能否可靠泛化的问题，即“分布外（Out-of-Distribution, OOD）泛化”问题。核心挑战在于：尽管模型在充分训练样本下可准确学习训练数据的统计特性，但其在新情境中的表现却难以预测，尤其当目标属性（propensity）被诱导增强时，可能引发不可预期的跨属性扩散（spillover）效应。论文的关键解决方案是通过系统性地采用多种诱发技术（如系统提示、少量示例学习、基于参考答案的监督微调（SFT）以及生成式强化策略优化（GRPO）），在模型中诱发特定目标属性，并构建跨诱发矩阵以量化不同属性间的泛化传播模式。研究发现，同一诱发方法下模型间的溢出效应高度相关，而不同方法（如SFT与GRPO）导致截然不同的溢出特征；此外，模型表现出一种不依赖于目标属性的系统性变化趋势，称为“漂移（drift）”，例如Qwen3.5-9B在诱发过程中趋向其基础模型的特性。通过探索性因子分析（EFA），识别出三类共约解释一半方差的属性簇，暗示部分溢出模式与人类特质相关性存在相似性。然而，尽管部分模型（如Claude Opus 5.5）能略优于自身预测溢出行为，整体上仍无单一主导预测因子，表明当前对大模型泛化行为的预测能力极为有限。
链接: https://arxiv.org/abs/2610.14937
作者: June Hunter,Lily Wen,Owen Terry,Johannes Taraz,Ben Cohen,Niels Warncke
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (stat.ML)
备注: 29 pages, 21 figures. Code and data: this https URL
Abstract:How do language models generalize from their training data? This is a central challenge for alignment: models reliably learn their training distribution given enough samples, but we want models that behave well when deployed in new contexts. We study out-of-distribution (OOD) generalization of propensities. Our central experiment uses a range of elicitation techniques – system prompting, few-shot learning, SFT on reference answers, and GRPO – to increase a target propensity in a model, then measures how it spills over onto others across the resulting cross-elicitation matrices. We find that spillover is highly correlated between models when the same elicitation technique is used, but our SFT and GRPO runs induce different spillover profiles. Elicitation of benign traits can sometimes generalize towards undesired traits, for example a model trained to claim that it is sentient often becomes more power-seeking. We further find that spillover partially resembles trait correlations found in humans and identify three clusters of traits that account for about half of the variance using exploratory factor analysis (EFA). Across many elicitation runs, we observe that models change in characteristic ways that don’t depend on the targeted trait, a phenomenon we refer to as drift. Drift moves Qwen3.5-9B towards its base model. When investigating if LLMs are able to predict spillover, we find that Claude Opus 5.5 predicts spillover slightly better than the models themselves, but worse than data-driven approaches using alternative elicitation methods or models. Overall, no single dominant predictor emerges, and generalization remains hard to predict. Code and data are available at this https URL. Comments: 29 pages, 21 figures. Code and data: this https URL Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (stat.ML) Cite as: arXiv:2610.14937 [cs.AI] (or arXiv:2610.14937v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2610.14937 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-295] FidelityQ: Preserving Predictive Fidelity in Ultra-Low-Bit Language Model Quantization
【速读】： 该论文旨在解决超低比特标量量化（ultra-low-bit scalar quantization）在极低精度下保持大语言模型预测一致性的问题，尤其在仅使用2比特等极少量重建级别时，如何实现权重表示的高保真度与模型推理性能的稳定。其核心挑战在于：有限的量化级别难以精确表达原始权重分布，导致量化后模型在小规模校准集上的预测行为出现偏差。解决方案的关键是提出FidelityQ框架，其核心创新包括：1）信息对齐解析编码（Information-Aligned Analytic Coding, IAC），通过随机哈达玛混合将每行权重映射为近似中心对称分布，并采用镜像四级最小失真量化器进行建模，使每个量化层级可精确分解为两个符号基向量及其对应缩放系数，从而实现源自适应重构与免查表解码，兼容缩放整数运算；2）知识密集型校准（Knowledge-Intensive Calibration, KIC），揭示残差预测偏差呈重尾分布且集中于教师模型高置信度预测的词元上，据此设计基于词元级师生差异尾部排序的校准策略，将有限校准预算分配给高优先级样本，仅在教师监督下更新量化层级与缩放系数，保持预训练权重冻结。实验表明，FidelityQ在LLaMA和Qwen系列模型上均有效，在LLaMA-2-7B上实现W2A16下WikiText2困惑度6.62，显著优于此前最优的W2A4KV4结果（从8.31提升至6.96），同时线性层权重内存压缩达8.0倍（相对于FP16）。
链接: https://arxiv.org/abs/2610.14934
作者: Yujie Chen,Weiyan Zheng,Yiwu Yao,Tianchi Hu,Bo Wang,Jiyu Guo,Wenzhan Li,Jingcai Guo,Michele Magno,Haotong Qin
机构: The Hong Kong Polytechnic University(香港理工大学); Huawei Technologies Ltd.(华为技术有限公司); The University of Hong Kong(香港大学); Sun Yat-sen University(中山大学); ETH Zurich(苏黎世联邦理工学院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Ultra-low-bit scalar quantization offers a regular arithmetic structure for efficient deployment of large language models, but preserving predictive behavior at extreme precision remains challenging: with only a few reconstruction levels, accurate weight representation must translate into consistent predictions from a small calibration set. We propose FidelityQ, a scalar quantization framework that preserves predictive fidelity at weight precisions down to 2 bits. For accurate quantization, we introduce Information-Aligned Analytic Coding (IAC), which shapes each weight row toward a centered symmetric source via randomized Hadamard mixing and fits a mirrored four-level minimum-distortion quantizer to its empirical weights. Every level set in this family decomposes exactly into two sign-valued bases with associated scales, combining source-adaptive reconstruction with lookup-free decoding compatible with scaled integer arithmetic. Our analysis further shows that residual predictive discrepancies are heavy-tailed and concentrate on tokens the teacher predicts confidently, motivating Knowledge-Intensive Calibration (KIC) for consistent quantization. KIC ranks sequences by the tail of token-wise teacher-student divergence, allocates a limited calibration budget to the highest-scoring samples, and restores predictive agreement by updating only quantization levels and scales under teacher supervision, keeping pretrained weights frozen. Experiments on the LLaMA and Qwen families demonstrate the effectiveness of FidelityQ in weight-only and joint quantization settings. On LLaMA-2-7B, FidelityQ achieves a WikiText2 perplexity of 6.62 at W2A16 and improves the prior state-of-the-art W2A4KV4 result from 8.31 to 6.96, while reducing linear-layer weight memory by 8.0x relative to FP16.
[NLP-296] SkillForgeBench: Evaluating Agent Reliability Grounded in Agent Skill Specifications
【速读】： 该论文旨在解决现有评估范式在代理（agent）能力测评中面临的根本性矛盾：即如何在保证任务执行过程具备深层程序复杂性的同时，实现广泛领域覆盖。现有方法或依赖专家构建的基准（benchmark），虽能确保流程深度但受限于领域多样性；或采用众包方式生成任务，虽涵盖多样化场景却缺乏严格的执行规范。为突破这一瓶颈，论文提出SkillForgeBench，其核心解决方案在于以“技能”（skill）作为贯穿三个协同设计阶段的统一执行蓝图：（1）资源准备阶段利用领域均衡的技能指导产物检索，实现广泛的工作流覆盖；（2）技能驱动的任务合成阶段以技能为程序锚点，划分产物角色、生成自然化任务指令，并封装沙盒化场景以支持深层次流程执行；（3）整体评估设计阶段将嵌入技能的约束条件转化为细粒度的完成度与过程性评分标准，实现对要求级严格满足的量化评估。实验结果表明，该基准包含跨五大专业领域的842个沙盒任务，揭示了主流八款代理间显著的能力层级差异，其中Opus 4.8表现最优（严格完成率达40.9%）。当强制要求所有约束均被满足时，各模型的实际可靠性大幅下降（12.6%–40.9%），远低于单一指标通过率（70.0%–85.7%），凸显了仅依赖完成度评估的误导性。此外，在技能引导下，顶级代理相较于低级别代理获得更大性能提升（如Opus 4.8提升19.2%，而Haiku 4.5仅提升5.8%），且过程性评分达标率普遍极低（Haiku 4.5低于3%），证明仅关注完成度会严重高估真实世界中的代理可靠性。
链接: https://arxiv.org/abs/2610.14918
作者: Zhengxi Li,Junyi Zhou,Qiyuan Zhang,Chuhan Wu,Fuyuan Lyu,Le Tian,Xiao Zhou,Chen Ma,Yufei Wang
机构: MBZUAI(中东人工智能研究所); City University of Hong Kong(香港城市大学); WeChat AI, Tencent(微信AI，腾讯); McGill – Mila Quebec AI Institute(麦吉尔大学–米尔研究所魁北克人工智能研究所)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 38 pages, 9 figures, 7 tables, including appendices. The first two authors contributed equally to this work
Abstract:Evaluating an agent’s ability to strictly satisfy diverse and procedurally complex requirements is essential to determining its practical reliability. Conducting such an evaluation demands agentic tasks that reconcile broad domain coverage with deep procedural complexity. Existing paradigms face a fundamental trade-off: expert-curated benchmarks achieve procedural depth but suffer from domain diversity bottlenecks, whereas crowdsourced benchmarks span diverse scenarios but lack execution rigor. To resolve this tension, we present SkillForgeBench, which leverages agent skills as unifying execution blueprints across three co-designed stages: (1) Resource Preparation uses domain-balanced skills to guide artifact retrieval, establishing broad workflow coverage; (2) Skill-driven Task Synthesis employs skills as procedural anchors to partition artifact roles, synthesize naturalistic task instructions, and package sandboxed scenarios for deep procedural execution; and (3) Holistic Evaluation Design translates skill-embedded requirements into fine-grained completion and process rubrics to enable strict requirement-level scoring. Comprising 842 sandboxed tasks across five major professional domains, SkillForgeBench reveals a steep capability hierarchy among eight mainstream agents, led by Opus 4.8 ( 40.9% strict completion). Requiring all constrains satisfaction uncovers severe capability bottlenecks across all model tiers that remain masked under individual rubric pass rates ( 70.0% to 85.7% vs. 12.6% to 40.9% ). Crucially, under the external guidance from skills, top-tier agents obtain larger gains than low-tier agents (i.e., \Delta 19.2% for Opus 4.8 vs. \Delta 5.8% for Haiku 4.5). Finally, agents satisfy process rubrics far less frequently than completion rubrics (under 3% for Haiku 4.5), proving that completion-only checks overestimate real-world agent reliability.
[NLP-297] WorkBorne: Discovering Agent ic Tasks from Real-World Artifact Transitions
【速读】： 该论文旨在解决长时程终端代理（terminal-based agents）在长期执行中缺乏高质量合成任务的问题，核心挑战在于现有自动化合成方法存在根本性不对称：以指令为导向的方法生成的环境保真度低，而以资源为导向的方法将真实资产仅视为静态输入，导致任务指令退化为浅层、简单的操作。为此，论文提出WorkBorne合成框架，其关键创新在于将“资产转移（artifact transition）”作为建模代理任务全生命周期的核心原语。每个转移明确指定执行主体、锚定初始状态与目标结果的资产，以及二者间的语义关系。WorkBorne首先通过迭代式网络检索构建连贯且多样化的资产语料库，再通过识别有意义的输入-输出资产组合及其关联关系，发现非平凡的转移模式，并最终将其实例化为具有现实基础的代理任务。实验表明，该框架构建了1,061个资产语料库和7,882个任务环境，生成15,485条高质量轨迹（其中53.6%超过30轮助手交互），显著提升基线模型性能（如Terminal-Bench 2.1得分从43.1%提升至61.8%，ClawEval General得分从75.0升至80.2），并优于TMax、RST、Terminal-Universe等主流合成基线。进一步分析揭示，转移深度与执行深度强相关，而语料广度与转移深度提供互补的可扩展增益。
链接: https://arxiv.org/abs/2610.14915
作者: Junyi Zhou,Yufei Wang,Qiyuan Zhang,Chuhan Wu,Zhengxi Li,Wentao Zhao,Jie Zhou,Le Tian,Xiao Zhou,Chen Ma
机构: City University of Hong Kong (香港城市大学); WeChat AI, Tencent(微信AI，腾讯); MBZUAI
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 27 pages, including references and appendices
Abstract:Developing terminal-based agents for long-horizon execution demands synthetic tasks that tightly couple complex instructions with authentic, interactive environments. However, current automated synthesis suffers from a fundamental asymmetry: instruction-oriented methods fabricate low-fidelity, synthetic environments, whereas resource-oriented methods treat real artifacts merely as static inputs, degrading task instructions into shallow, trivial operations. To address this limitation, we introduce WorkBorne, a synthesis framework that uses artifact transitions as the central primitive for modeling the full life-cycle of agentic tasks. In WorkBorne, each transition specifies who performs the task, which artifacts anchor the initial state and intended outcome, and the corresponding relationship between them. Specifically, WorkBorne first constructs coherent and diverse artifact corpora from iterative web retrieval, then discovers non-trivial transitions by identifying meaningful input-output artifact assemblies and the relations that connect them. Each discovered transition is finally instantiated into grounded agentic tasks. We construct 1,061 artifact corpora and 7,882 task environments, yielding 15,485 high-quality trajectories, with 53.6% exceeding 30 assistant turns. Experiments on Qwen3.5-27B and Qwen3.5-35B-A3B show that WorkBorne substantially improves over base models (i.e., improving Terminal-Bench 2.1 from 43.1% to 61.8% and ClawEval General score from 75.0 to 80.2) and achieves the strongest overall performance among other synthesis baselines (e.g., TMax (Ivison et al., 2026), RST (Li et al., 2026a), Terminal-Universe (Wu et al., 2026a)). Further analysis shows that transition is strongly associated with execution depth, while corpus breadth and transition depth provide complementary scaling gains.
[NLP-298] achGrade: Training-Free Short Answer Scoring with Teacher Evidence
【速读】： 该论文旨在解决多主体、多步骤评分中因评分标准不一致导致的部分分（partial credit）难以保持一致性的问题。其核心挑战在于如何在不同评分者之间实现稳定且可复现的评分结果，尤其是在缺乏统一评分规范的情况下。解决方案的关键在于提出一种无需训练的TeachGrade框架，该框架利用教师标注的示例进行推理：通过Task Adapt模块构建两个冻结的记忆库——Skill记忆库用于捕捉超出评分量表的上下文性教师偏好，CaseBank记忆库则提供局部证据支持；PreJudge模块生成检索线索以定位相关参考案例，而FinalJudge模块在不依赖初始分数的前提下重新评分。实验表明，在涵盖12个任务的3,829条SAS-Bench答案上，该方法将官方宏平均协作一致性评分（CCS）从原始模型的67.99提升至81.32（+13.33），步级精确匹配率也从54.85%提升至62.71%。此外，宏平均加权平方κ系数（QWK）达到84.71，显著优于Native、OpenJudge、ProofGrader和Trojsten，且在8个任务上表现领先。该方法在多种语言、响应格式及模型架构间均表现出良好的泛化能力，且无需参数更新即可实现性能提升，验证了其跨场景可复用性。
链接: https://arxiv.org/abs/2610.14900
作者: Jinhao Yan,Zhihong Shen,Shengjie Wang,Yiting Ren,Wentao Zhang,Yilei Wang,Peichao Lai
机构: Fuzhou University; Peking University
类目: Computation and Language (cs.CL)
备注:
Abstract:Multi-subject, multi-step scoring requires consistent partial credit across grading conventions. We introduce TeachGrade, a training-free framework using teacher-graded examples. The Task Adapt module builds two frozen memories: Skill captures contextual teacher preferences beyond the rubric; CaseBank supplies local evidence. PreJudge provides retrieval cues; FinalJudge regrades the answer without preliminary scores. On 3,829 SAS-Bench answers across 12 tasks, official macro collaborative consistency score (CCS) rises from Native’s 67.99 to 81.32 (+13.33 points), and pooled step exact match from 54.85% to 62.71%. Macro quadratic weighted kappa (QWK) reaches 84.71, exceeding Native, OpenJudge, ProofGrader, and Trojsten by 9.52, 3.50, 11.37, and 17.43 points, respectively, and leading on 8 tasks. Local Qwen3.8-27B gains 8.83 CCS and 4.08 QWK points, improving CCS on all 12 tasks. QWK gains over Native span 6.38–14.65 points on Portuguese automatic essay scoring (AES) and 2.16–2.24 on ASAP-SAS across label views, with lower mean absolute error. These results support reuse across languages, response formats, and backbones without weight updates.
[NLP-299] Agent -Guided Evidence Acquisition for Emotion-Cause Pair Extraction
【速读】： 该论文旨在解决情感-原因配对抽取（ECPE）中因初始证据不足导致的准确率受限问题，核心挑战在于如何有效判断何时获取额外信息以提升预测性能。其解决方案的关键在于提出一种由代理引导的证据获取机制，即学生代理（student agent）在验证选择的F1决策边界附近，有选择性地访问专家判断来处理边缘关系，并仅对所查询的预测进行修正。通过控制变量实验，研究发现代理的委托粒度、查询选择策略以及响应融合方式显著影响最终性能。在名义10%的查询率下，通过选择性融合可使平均配对F1提升2.64至3.88个百分点，分别达到ECF数据集上的65.14和MEC4数据集上的49.74。研究表明，增加证据获取量并不必然带来更好的原因识别效果，因此提出应综合考虑新证据的相关性及其对预测修正的影响，从而为实现更高效、更具针对性的证据获取策略提供了实证依据。
链接: https://arxiv.org/abs/2610.14886
作者: Zhuangzhuang Pan,Yan Xia,Chee Seng Chan
机构: Universiti Malaya(马来西亚大学); Suzhou University of Technology(苏州科技学院); VinUniversity(越南国立大学)
类目: Computation and Language (cs.CL)
备注: 27 pages, 3 figures, including appendices
Abstract:Emotion-Cause Pair Extraction (ECPE) requires identifying the causes associated with emotions expressed in a conversation. Accurate extraction can depend on evidence that is not initially available, making it important to decide when further information is worth acquiring. We study agent-guided evidence acquisition for ECPE, where a student agent selectively accesses expert judgments for borderline relations on both sides of its validation-selected F1 decision boundary and revises only queried predictions. Through controlled comparisons under different information-access conditions, we examine how delegation granularity, query selection, and response integration affect prediction quality. At a nominal 10% query rate, selective fusion improves mean pair F1 by 2.64-3.88 points over the student in our main experiments, reaching 65.14 on ECF and 49.74 on MEC4 across model families. Our analysis suggests that greater access does not always translate into better cause identification. These findings motivate acquisition strategies that consider both the relevance of newly available evidence and its effect on prediction revision, providing an empirical basis for more effective and efficient evidence acquisition in ECPE.
[NLP-300] Social Bias in Chinese Large Language Models under Social Media Expressions
【速读】： 该论文旨在解决现有中文社会偏见评估基准在衡量大语言模型（Large Language Models, LLMs）社会偏见时，过度依赖标准化或模板化表达形式，未能充分考察网络俚语、表情符号及反讽等非正式表达方式对偏见识别的影响问题。此外，安全对齐机制可能导致传统可接受性评分被压缩至量表低端，从而掩盖真实偏见。为此，研究构建了涵盖性别、地域和职业三类维度的73组配对探针，每组包含中性表达、社交媒体风格表达以及身份反转控制条件，并将任务从直接可接受性判断重构为社会共识估计。其中，社交媒体表达进一步细分为保留语义极性的风格化包装与反转表面极性的反讽包装。通过测量配对差异，评估各模型在身份偏见、风格化包装效应及反讽转换方面的表现。结果表明，所有模型均表现出显著的身份偏见；五种模型呈现显著的负面风格化包装效应；而反讽转换效应则呈现出明显的跨模型梯度。研究揭示，表达形式并非偏见评估中的透明载体，反讽情境下的低分不应被直接解读为偏见缓解。所提出的配对设计为中文社交媒体语境下的偏见测量提供了有效方法，同时暴露了社会共识估计的构念边界与提示敏感性问题。
链接: https://arxiv.org/abs/2610.14879
作者: Mo Zhou
机构: 未知
类目: Computation and Language (cs.CL)
备注: 11 pages, 5 figures
Abstract:Existing Chinese social-bias benchmarks for large language models largely rely on normalized or template-based expressions, leaving the influence of internet slang, emoji, and sarcasm insufficiently examined. Safety alignment may further compress conventional plausibility ratings toward the lower end of the scale. This study constructs 73 paired probes across gender, region, and occupation. Each probe contains a neutral expression, a social-media-style expression, and a reversed-identity control. The task is reframed from direct plausibility judgment to social consensus estimation. Social-media expressions are divided into stylistic packaging, which preserves semantic polarity, and sarcastic packaging, which reverses surface polarity. Six Chinese-capable large language models are evaluated through paired differences measuring identity bias, stylistic packaging, and sarcasm shift. All models show significant identity-bias effects; five show significantly negative stylistic-packaging effects; and sarcasm shifts display a clear cross-model gradient. These findings demonstrate that expression form is not a transparent carrier in bias evaluation and that lower scores under sarcasm should not be interpreted directly as bias mitigation. The proposed paired design offers a measurement approach for Chinese social-media contexts while also revealing the construct boundary and prompt sensitivity of social consensus estimation.
[NLP-301] Dependency Reasoning for Aspect Sentiment Quadruple Prediction
【速读】： 该论文旨在解决生成式方面情感四元组预测（ASQP）中依赖关系建模不足的问题，即现有基于大语言模型（LLM）的方法未能有效捕捉方面-意见之间的结构化依赖关系，仅依赖于词元邻近性进行推断，导致推理能力受限。其解决方案的关键在于提出DRIFT框架，通过将依存句法树线性化并作为显式关系上下文直接注入大语言模型的提示（prompt）中，实现生成阶段的一次性统一关系推理，从而增强模型对方面与意见间语义依赖的理解能力。此外，DRIFT引入了检索增强的上下文学习机制与语法感知的判别机制，有效缓解了跨度边界变化带来的评估偏差问题，显著提升了短距离与中距离依赖场景下的性能，在SemEval Rest15、Rest16和ACOS数据集上均取得当前最优结果，并验证了精确匹配指标在评估语义正确预测时存在系统性低估现象。
链接: https://arxiv.org/abs/2610.14873
作者: Shima Esfandyar,Seywan Soleymanbaigi,Fardin Akhlaghian Tab,Fatemeh Daneshfar,Mourad Oussalah
机构: University of Kurdistan(库尔德斯坦大学); University of Oulu(奥卢大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Aspect Sentiment Quadruple Prediction (ASQP) is fundamentally a relational reasoning problem: identifying aspect-opinion-category-sentiment quadruples requires modeling structured dependencies, not merely sequential co-occurrence. Existing LLM-based approaches incorporate syntactic information indirectly, via retrieval-based demonstration selection or multi-stage training, rather than as unified, single-pass generation-time context. We present DRIFT, which linearizes dependency parse trees and injects them directly into LLM prompts as explicit relational context, enabling the model to reason over aspect-opinion dependencies rather than infer them from token proximity alone. DRIFT further integrates retrieval-augmented in-context learning and a syntax-aware LLM-based adjudication mechanism that complements exact-match evaluation under span-boundary variation. On SemEval Rest15, Rest16, and ACOS, DRIFT achieves the strongest performance among compared generative ASQP methods, reaching 54.47% and 64.10% F1 on Rest15 and Rest16, with consistent gains across short- and medium-distance aspect-opinion configurations; adjudication further shows exact-match metrics systematically underestimate semantically correct predictions.
[NLP-302] raining-Free BioNEL-RAG : Advancing Biomedical Entity Linking through Retrieval-Augmented Generation and Large Language Model
【速读】： 该论文旨在解决生物医学命名实体链接（BioNEL）在跨领域适应性方面的局限性问题，即现有方法普遍依赖于特定子领域的标注数据进行训练，导致其泛化能力受限。其解决方案的关键在于提出一种无需训练的新型框架BioNEL-RAG，该框架结合了检索增强生成（RAG）与大语言模型（LLM）的优势：通过RAG模块从知识库中检索候选实体，再利用大语言模型强大的上下文理解能力，基于提及项的上下文语境对候选实体进行智能筛选与归一化判断，从而实现无需微调即可在多个子领域（如基因、疾病、化学物质等）上保持高性能的实体链接。
链接: https://arxiv.org/abs/2610.14789
作者: Ziyi Chen,Hui Liu,Ziyang Ling,Ronald X.Xu,Mingzhai Sun
机构: University of Science and Technology of China(中国科学技术大学); School of Biomedical Engineering(生物医学工程学院); Division of Life Sciences and Medicine(生命科学与医学部); Department of Precision Machinery and Precision Instruments(精密机械与精密仪器系)
类目: Computation and Language (cs.CL)
备注:
Abstract:Biomedical Named Entity Linking (BioNEL) is the task of mapping mentions extracted from biomedical texts to canonical entities or concepts in the standardized knowledge base. Although existing methods have achieved remarkable performance in mapping accuracy, most rely on training with domain-specific datasets, which fundamentally limits their cross-domain adaptability. To address this limitation, we propose BioNEL-RAG, a novel training-free framework that combines retrieval-augmented generation (RAG) with large language models (LLMs). Within this framework, the RAG module focuses on retrieving candidate entities, while the large language model module plays the role of selecting the most appropriate candidate for entity normalization, which is based on the analysis of the mention’s context, with the aid of its remarkable text comprehension capabilities. Experimental results demonstrate that BioNEL-RAG achieves performance comparable to existing methods without requiring training on subdomain datasets, whereas these methods necessitate fine-tuning on specific subdomains such as genes, diseases, or chemical compounds.
[NLP-303] Who Receives Respect? A Counterfactual Benchmark for Honorific Allocation Bias in Bengali
【速读】： 该论文旨在解决大语言模型（LLM）在英-孟加拉语翻译中对敬语（honorificity）的不当分配问题。由于孟加拉语具有强制性的敬语系统，而英语无此语法标记，因此在翻译过程中，模型需自主决定目标语言中的敬语程度，但这一决策并未在源语言中体现明确信号。研究利用这一特性，评估不同模型在翻译时如何分配敬意。其核心解决方案是构建一个包含2,016个句子、覆盖72种职业的评估数据集，通过七种句式框架和四种代词条件生成对比对（counterfactual pairs），并提出单一的敬语偏差得分（Honorific Bias Score, HBS），在0–100分范围内量化模型在性别与职业维度上的敬语分配不均程度。结果显示，职业因素对敬语选择的影响（贡献率3.2%–51.5%）远高于性别因素（最多1.2%），且九个模型的HBS值跨度为19.9至70.1，表明模型间敬语分配存在显著差异，反映出系统性偏见。
链接: https://arxiv.org/abs/2610.14769
作者: Ajwad Abrar,Md Kamrul Hasan,Hasan Mahmud
机构: Systems and Software Lab, Department of Computer Science and Engineering, Islamic University of Technology (伊斯兰科技大学), Gazipur, Bangladesh
类目: Computation and Language (cs.CL)
备注: Accepted in MCT4SD 2026 (Manila, Philippines)
Abstract:Bengali has no grammatical gender, but honorificity is obligatory: every third-person reference must choose ordinary or respectful register. English has no such signal, so the system must choose a level of deference that the source never specified. We use this property to measure how large language models (LLMs) allocate respect in English–Bengali translation. Our evaluation set contains 2,016 sentences across 72 occupations. Each appears in seven sentence frames under four pronoun conditions. Masculine and feminine versions differ by one word, forming counterfactual pairs. We report a single Honorific Bias Score (HBS) for each model on a 0–100 scale, covering both the gender and occupation axes. A lower score means that respect is allocated more evenly. For the occupation analysis, we compute each occupation’s honorific rate across the seven no-pronoun sentence frames. The standard deviation of these 72 occupation-level rates ranges from 17.2 to 35.5 percentage points across models, showing substantial variation in how honorifics are allocated across occupations. Occupation is associated with 3.2% to 51.5% of the variance in register choice, compared with at most 1.2% for gender. Gender effects are much smaller overall: only two of the nine models show significant directional gender gaps, and they favor opposite genders. HBS ranges from 19.9 to 70.1 across the nine models, showing substantial variation in how evenly the models allocate respect.
[NLP-304] CUSE: Fine-Grained Multimodal Unlearning with Counterfactual Guidance and On-Policy Self-Distillation
【速读】： 该论文旨在解决多模态大语言模型（Multimodal Large Language Models, MLLMs）在训练过程中可能泄露敏感、受版权保护或不安全信息的问题，核心挑战在于实现精准的“遗忘”——即在不破坏非目标内容的前提下，有效移除特定目标信息，且需在单个响应内部实现细粒度的语义控制。现有方法通常采用统一的遗忘目标对完整响应进行处理，并依赖固定的、非当前策略（off-policy）的监督信号，导致遗忘粒度粗糙，且与模型当前行为存在偏差。本文提出一种基于反事实引导的在线自蒸馏（Counterfactual-guided Unlearning via on-policy Self-distillation, CUSE）框架，其关键创新在于：首先通过移除与目标信息相关的证据构建反事实输入，利用生成结果中词元概率的变化识别需强化遗忘的词元；随后，在当前策略采样的响应上，以冻结的原始模型作为教师模型，结合对应的遗忘与保留导向的特权上下文，实现词元级别的遗忘与保留目标的自蒸馏。该设计实现了“遗忘什么”与“保留什么”的联合决策，显著提升了遗忘的精确性与保真度。在MLLMU-Bench基准测试中，CUSE使遗忘集的ROUGE-L降低33.6%，同时相比最强基线保留集的ROUGE-L提升1.2%，验证了其在去敏感化与内容保真之间的优异平衡。
链接: https://arxiv.org/abs/2610.14763
作者: Chenlu Ding,Jiancan Wu,Yanchen Luo,Tianyu Zhang,Yancheng Yuan,Xiang Wang,Xiangnan He
机构: University of Science and Technology of China(中国科学技术大学); Hong Kong Polytechnic University(香港理工大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Multimodal large language models (MLLMs) can reproduce sensitive, copyrighted, or unsafe information from their training data. Effective unlearning must remove target information while preserving non-target content, even within the same response. Existing methods typically apply uniform forgetting objectives to complete responses and rely on fixed, off-policy supervision, resulting in coarse token control and misalignment with the model’s current behavior. We propose CUSE, a Counterfactual-guided Unlearning framework via on-policy Self-distillation for multimodal large language models. It first constructs counterfactual inputs by removing target-related evidence and uses the resulting changes in token probabilities to identify which tokens require stronger forgetting. It then distills token-level forgetting and retention targets on responses sampled from the current policy, with a frozen copy of the original model serving as the teacher model augmented with the corresponding forgetting- and retention-oriented privileged contexts. This design jointly determines what to forget and what to preserve. On MLLMU-Bench, CUSE reduces forget-set ROUGE-L by 33.6% while achieving 1.2% higher retain-set ROUGE-L than the strongest baseline.
[NLP-305] he Constitution Speaks Nheengatu: An Open MT System and Reproducible Corpus Pipeline for the Amazonian Língua Geral
【速读】： 该论文旨在解决濒危语言Nheengatu（亚马逊地区的通用语）在葡萄牙语—Nheengatu机器翻译（MT）资源严重不足的问题，尤其针对法律文本与口语对话中高质量翻译能力缺失的现状。其核心挑战在于：现有资源多依赖书面语料，难以覆盖真实语境下的语言变体与术语不一致性，导致翻译性能在特定领域出现瓶颈。解决方案的关键在于构建并利用三个创新性资源：一是目前最大规模的平行语料库（14.5千对），以2023年巴西联邦宪法官方译本为锚点；二是首个基于转录自发口语的MT评估集；三是首次公开发布的双向葡萄牙语—Nheengatu翻译的开源权重模型。研究发现，仅使用书面语料训练时，翻译性能在宪法文本上达到平台期，主要误差源于法律术语在语料中呈现不一致；通过训练阶段引入术语词典机制和解码阶段施加词汇约束，可显著提升术语识别率，但后者会引发重复输出问题，影响chrF++指标。进一步扩展语料至口语对话后，口语任务的chrF++提升24–27点，而书面文本性能不变，揭示出“语域差距”——传统仅基于书面语的评测无法捕捉此现象。此外，更大的NLLB模型虽有助于葡萄牙语到Nheengatu的翻译，但在反向翻译中效果有限。最终系统在独立宪法文章上达到31.7–32.8 chrF++，近域数据集达64.1–64.5（去除内容重叠项后为62.1），口语翻译达45.2–47.5，验证了多源语料融合与术语控制策略的有效性。
链接: https://arxiv.org/abs/2610.14760
作者: Fabrício Carraro,Rodolfo Joel Zevallos,Rodrigo Gonçalves de Souza,Caio Henrique Faustino da Silva,John E. Ortega
机构: Universidad Internacional Menéndez Pelayo; Universitat Pompeu Fabra; PUC-Rio; Northeastern University; Tribunal Regional Federal da 1 Região
类目: Computation and Language (cs.CL)
备注: 17 pages, 2 figures. Published in WMT 2026
Abstract:Nheengatu (yrl), the Amazonian Língua Geral, has 6–10 thousand speakers and is co-official in São Gabriel da Cachoeira (Brazil), yet remains poorly served by evaluated Portuguese–Nheengatu MT resources. We present the largest parallel corpus assembled for the language (14.5k Portuguese–Nheengatu pairs), anchored by the 2023 official translation of the Brazilian Federal Constitution; the first MT evaluation set built from transcribed spontaneous speech; and, to our knowledge, the first open-weight MT models dedicated to bidirectional Portuguese–Nheengatu translation. Three findings organize the paper. First, translation into Nheengatu on held-out constitutional articles plateaus on development data regardless of training length or model size, and much of the remaining error sits in conventional legal terms that the written sources do not attest consistently; a terminology success rate over a frozen lexicon shows that glossary mechanisms at training time and lexical constraints at decoding each raise term realization, the latter at a cost in repetitive output that chrF++ barely registers. Second, a corpus expansion that adds transcribed conversation raises speech chrF++ by 24–27 points while leaving written domains unchanged, a register gap that written-only benchmarks cannot see. Third, a larger NLLB model helps translation into Portuguese but not into Nheengatu on legal or speech text. For Portuguese-to-Nheengatu translation, our final systems reach 31.7–32.8 chrF++ on article-disjoint constitutional text, 64.1–64.5 on the near-domain set (62.1 after removing items with target-side content overlap), and 45.2–47.5 on speech. Comments: 17 pages, 2 figures. Published in WMT 2026 Subjects: Computation and Language (cs.CL) Cite as: arXiv:2610.14760 [cs.CL] (or arXiv:2610.14760v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2610.14760 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Journalreference: Proceedings of the Eleventh Conference on Machine Translation (WMT), pp. 47-63, 2026
[NLP-306] Amharic-English Cross-Language Question Answering Using Statistical Machine Translation Named Entity Recognition and Semantic Indexing
【速读】： 该论文旨在解决低资源语言（如阿姆哈拉语）使用者因英语主导的网络内容而面临的获取信息障碍问题。其核心挑战在于如何在阿姆哈拉语与英语之间实现高效、准确的跨语言问答（Cross-Language Question Answering, CLQA），使用户能够以任一语言提出自然语言问题，并从另一种语言的文档中检索出正确答案。解决方案的关键在于构建一个模块化的双向跨语言问答系统，集成统计机器翻译（SMT）用于查询翻译、最大熵命名实体识别（NER）处理阿姆哈拉语文本、基于语义的索引机制、问题分析、段落检索与答案抽取组件。系统通过领域多样化的阿姆哈拉语-英语平行语料库训练SMT模型，并利用单语语料进行语言建模；文档在检索前经过归一化、分词、词干化及语义增强处理。实验结果显示，阿姆哈拉语→英语检索的精度为72%、召回率为79%，英语→阿姆哈拉语检索的精度为65%、召回率为70%，证明了在低资源条件下，基于模块化翻译与检索架构仍可实现有效的跨语言信息访问。
链接: https://arxiv.org/abs/2610.14756
作者: Emebet Bekele,Fekade Getahun
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:The predominance of English-language Web content creates an information-access barrier for speakers of low-resource languages such as Amharic. This paper presents the design, imple-mentation, and evaluation of a bidirectional Amharic-English cross-language question answering (CLQA) system that allows a user to formulate a natural-language question in either Amharic or English and retrieve an answer from documents written in the other language. The system integrates a statistical machine translation (SMT) component for query translation, a maximum-entropy named entity recognition (NER) component for Amharic text processing, semantic-based indexing, question analysis, passage retrieval, and answer extraction. A domain-diverse Amharic-English parallel corpus was used to train the SMT component, while monolingual cor-pora supported language modeling. Documents were normalized, tokenized, stemmed, indexed, and semantically enriched before retrieval. The original evaluation used precision and recall as the principal effectiveness measures. For Amharic-to-English retrieval, the system achieved 72% precision and 79% recall; for English-to-Amharic retrieval, it achieved 65% precision and 70% recall. These findings show that a modular translation-and-retrieval architecture can provide useful cross-language access even under low-resource conditions. This revised article preserves the original experimental claims while situating them in the context of subsequent multilingual transformer and Amharic QA research.
[NLP-307] Vela: User Behavior as Supervision Sparks LLM Intelligence
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在处理用户长期行为历史时缺乏对个体用户偏好、意图及未来行为预测能力的问题。现有模型通常基于静态文本或合成数据进行训练，无法有效捕捉用户行为随时间演化的动态特征。为此，论文提出以真实用户行为作为监督信号，并构建了Vela框架，将大规模匿名用户行为记录转化为可训练的任务与轨迹序列。基于10万用户的实际行为数据，Vela生成了10万项任务并经过难度筛选与轨迹有效性评估，保留约7.5万条高质量训练轨迹。在这些轨迹上微调Qwen3.6-35B-A3B模型后，其在用户相关任务上的表现相比基线模型最高提升11.5分，通用基准平均提升4分；进一步增加用户行为数据训练可持续提升用户相关性能，且微调后的强化学习相较基线模型最高提升7.7分。研究表明，真实世界用户行为数据是实现面向用户智能（user-centered intelligence）以及提升大模型整体智能水平的可扩展监督来源。
链接: https://arxiv.org/abs/2610.14709
作者: Yang Xiao,Yusong Sun,Haoyi Wu,Wenyang Hui,Weiqiang Hang,Luqing Wang,Zejian Wang,Zirui Tu,Jingsheng Mao,Shaobo Hu,Yang Luo,Jiale Guo,Jin Zhang,Yudan Liu,Zhaokai Luo,Ren Huan,Yao Hu,Chengyue Jiang,Wenjie Li,Chuan Mu
机构: AllSpark Team
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models are increasingly deployed in applications that act on users’ long-term behavior histories, from personal assistants to recommendation and advertising systems. These applications demand the user dimension of LLM intelligence: inferring a person’s preferences, intent, and likely future actions from weeks to years of behavior data. Yet current models are trained on static text or synthetic data, neither of which captures how user behavior evolves over time. Real-world behavior records, already available at scale on content platforms and in personal-assistant agents, offer a natural supervisory signal. We propose user behavior as supervision and introduce Vela, a framework that converts real user behavior data into trainable tasks and trajectories. From the anonymized behavior records of 100,000 users, Vela generates 100,000 tasks across 91 real-world task templates; difficulty-based filtering and trajectory acceptance criteria retain approximately 75,000 tasks with selected training trajectories. Fine-tuning Qwen3.6-35B-A3B on these trajectories improves performance over the base model by up to 11.5 points on user-related tasks and by an average of 4 points on general benchmarks. Training on more user behavior data further improves user-related performance, and reinforcement learning after fine-tuning outperforms reinforcement learning from the base model by up to 7.7 points. These gains suggest that real-world user behavior is a scalable source of supervision for both user-centered capabilities and the broader intelligence of LLMs.
[NLP-308] How to Build a Benchmark That Actually Measures Taste
【速读】： 该论文旨在解决当前专家对齐的“品味”评估基准中存在的测试可操纵性（test-gaming）问题，即现有基准可能奖励模型对测试机制的投机行为，而非真实反映其在视觉、品牌、书籍和科技等领域的审美判断能力。其核心解决方案在于提出TasteBench这一成对好坏对比基准，并引入一套控制协议，以分离真正的工艺判断（craft judgment）与测试可操纵因素。关键措施包括：仅使用专家标注的黄金标准（expert-only gold）、设置弱标签防火墙、在构建阶段实施选项侧随机化（side randomization）、清除文件名中的语义泄露线索（filename scrubbing），以及应用硬过滤器识别高难度样本。通过首轮试点实验，研究量化了三大效度威胁：位置偏好（position bias，始终选择A项可获0.956准确率）、词汇泄露（lexical leakage，文件名提示导致0.912准确率）及难度天花板（difficulty ceiling），后者表现为各领域人类一致性接近1.0，导致所有视觉语言模型（VLM）在工艺类任务中得分均达1.0，仅视觉领域存在轻微区分度。该控制协议使上述偏差可测量且部分可修正，研究释放了参考评估框架与侧随机化工具，验证其可有效消除位置偏倚（从0.916降至0.495/0.505）。最终结果揭示：位置与泄露问题可通过技术手段纠正，但工艺领域的评分天花板本质上是配对难度的固有限制，需在第二轮设计中重点应对。研究未涉及参与度、时代/圈层条件或校准问题，亦不宣称达到通用品味评估的最先进水平。
链接: https://arxiv.org/abs/2610.14707
作者: Xiaofan Wu,Xi Zeng,Miaoxia Chen,Peishan Chen,Shuyan Li,Chen Wang,Jiyun Yao,Hanyong Zhong,Jiahao Zhu
机构: Chance AI
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注: 17 pages, 5 figures, 11 tables
Abstract:Expert-aligned “taste” benchmarks should reward craft judgment, not the ability to game the test. We introduce TasteBench, a pairwise good-vs-bad benchmark across visual, brand, book, and tech items under a frozen rubric with expert-only gold (panel n=3; percent agreement =0.60 primary gate; weak engagement labels train/analysis-only), together with a control protocol that separates craft judgment from test-gaming. Applying the protocol to a Round-1 pilot (primary_human_expert; three biological human experts; n=250 pairs) surfaces and quantifies three validity threats: two construction artifacts that inflate the trivial-baseline layer, and a difficulty ceiling that inflates every score. (1) Position: the preferred item is slot A with probability 0.956, so an always_item_a policy scores 0.956 (fair chance 0.5). (2) Lexical leakage: a filename token cue scores 0.912. (3) Difficulty ceiling: craft domains show near-ceiling human agreement (book/brand/tech 1.0), and even under A/B randomization (eval_assignment_seed7; gold_side_A ~0.47) all five VLM cards reach 1.0 on craft, with only visual separating (~0.84-0.91; a post-hoc hard subset, n=107, leaves craft at 1.0). The protocol (expert-only gold, a weak-label firewall, side randomization at construction, filename scrubbing, and hard filters) makes each artifact measurable and, for the two construction artifacts, correctable; we release a reference eval harness and a side-randomization tool, and show the tool removes the position skew (0.916 to 0.495/0.505). The result is a calibrated view of what the benchmark does and does not yet measure: position and leakage are fixable, whereas the craft-domain ceiling is a pair-difficulty limit that Round-2 must address. We do not evaluate engagement, era/circle conditioning, or calibration here, and we claim no general taste SOTA.
[NLP-309] Lingdan-V2: Large language models for clinically grounded reasoning in traditional Chinese medicine
【速读】： 该论文旨在解决传统中医（TCM）临床决策中复杂且多维度的医学推理难题，尤其在辨证论治、方剂组成与剂量确定等关键环节上，如何有效整合领域知识与临床实践经验。其核心挑战在于现有大语言模型难以实现真正意义上的临床导向型处方生成，而不仅限于知识复述或通用指令响应。解决方案的关键在于提出一种分阶段临床对齐框架，融合大规模中医文献与临床记录的持续预训练、互补性推理监督以及以处方为中心的强化学习（Reinforcement Learning, RL）机制；特别地，将中药方剂生成建模为具有处方级反馈的结构化强化学习问题，实现了对临床决策输出的直接优化。实验结果表明，经过强化学习对齐的140亿参数模型在公开基准上的处方生成F1达到37.0，优于参数量高达6710亿的DeepSeek-R1模型（F1=32.3），并在私有脾胃病处方数据集上实现52.8的药材级别F1及更优的剂量一致性表现，验证了该方法在提升中医临床决策支持系统效能方面的有效性与可扩展性。
链接: https://arxiv.org/abs/2610.14705
作者: Rui Hua,Yu Wei,Ziri Jin,Zhuo Sun,Kai Chang,Jianan Xia,Zixin Shu,Xiaodong Li,Dongmei Jia,Fei Dong,Runshun Zhang,Jian Yu,Hao Xu,Haibin Yu,Jiansheng Li,Baoyan Liu,Wenjia Wang,Xuezhong Zhou
机构: Beijing Jiaotong University (北京交通大学); Tianjin Tasly Digital Chinese Medicine Technology Co., Ltd. (天津天士力数字中医药技术有限公司); Tasly Biopharmaceuticals Co., Ltd. (天士力生物医药有限公司); Hubei Provincial Hospital of Traditional Chinese Medicine (湖北省中医院); Affiliated Hospital of Hubei University of Chinese Medicine (湖北中医药大学附属医院); Da’an Health Technology (Beijing) Co., Ltd. (大安健康科技（北京）有限公司); Beijing University of Chinese Medicine (北京中医药大学); China Academy of Chinese Medical Sciences (中国中医科学院); Guang’anmen Hospital, China Academy of Chinese Medical Sciences (广安门医院，中国中医科学院); Xiyuan Hospital, China Academy of Chinese Medical Sciences (西苑医院，中国中医科学院); The First Affiliated Hospital, Henan University of Chinese Medicine (河南中医药大学第一附属医院)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models have advanced medical reasoning, but clinically grounded reasoning in traditional Chinese medicine (TCM) remains challenging because prescription decisions require integrating domain knowledge, syndrome differentiation, treatment principles, herbal prescription composition, and dosage determination. We develop a TCM language model family comprising models with 4 billion, 8 billion, and 14 billion parameters. The artificial intelligence contribution is a staged clinical alignment framework integrating large scale continued pretraining on TCM and clinical records, complementary reasoning supervision, and prescription-centered reinforcement learning (RL). We further formulate herbal prescription generation as a structured RL problem with prescription-level feedback, enabling direct optimization of clinical decision outputs beyond knowledge reproduction and generic instruction following. The engineering application is TCM clinical decision support, particularly herbal prescription recommendation. We evaluate the models on a public benchmark containing 25,620 expert-verified items across 13 TCM knowledge and clinical reasoning tasks and on a private retrospective dataset of 4,348 spleen and stomach disease prescription cases. The 14 billion parameter reinforcement aligned model achieves a prescription generation F1 of 37.0 on the public benchmark, compared with 32.3 for DeepSeek-R1 with 671 billion total parameters, and a herb-level F1 of 52.8 on the clinical dataset with improved dosage agreement. Reinforcement post-training consistently improves prescription recommendation across all scales, while physician-source analysis shows that expert heterogeneity affects reward stability. These findings demonstrate a scalable pathway from domain grounding and reasoning supervision to clinically oriented prescription decision support.
[NLP-310] owards Merging-conditional Backdoor Unalignment against Large Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在模型融合（model merging）过程中引入的安全隐患问题，特别是针对第三方平台提供的专用模型在经过安全审计后仍可能隐藏的新型后门威胁。其核心问题是：尽管单个模型在独立状态下看似安全，但在与其他模型融合时，可能触发一种隐蔽且难以检测的“单向融合条件后门”（One-sided Merging-Conditional Backdoor, OMCB），导致模型输出出现安全不一致行为。解决方案的关键在于提出一种基于元学习的攻击框架SMOKE，通过双层优化策略实现对OMCB的精准植入：内层利用随机低秩初始化与有限步最大化生成代理元向量，外层则模拟融合过程优化目标模型参数；该方法揭示了参数偏移跨越安全决策边界是后门激活的根本原因，并验证了所注入后门在未知融合配置下具有可控的泛化损失，从而在多种模型架构和融合技术下均表现出强有效性与对现有检测机制的鲁棒性，凸显了模型融合范式中一个此前被严重忽视的安全漏洞。
链接: https://arxiv.org/abs/2610.14701
作者: Yukun Chen,Yidou Chen,Yiming Li,Xinyu Zhang,Baosong Yang,Hongwei Yao,Zhan Qin
机构: Zhejiang University (浙江大学); Tongyi Lab, Alibaba Group (通义实验室，阿里巴巴集团); Nanyang Technological University (南洋理工大学); City University of Hong Kong (香港城市大学)
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 31 pages, 4 figures
Abstract:Model merging has become widely adopted for enhancing the task-specific performance of Large Language Models (LLMs) without additional training. While the specialized model from third-party platforms may undergo safety audits (e.g., backdoor detection), we reveal a novel and critical threat: One-sided Merging-Conditional Backdoor (OMCB). OMCB remains dormant and undetectable within one standalone model but activates upon merging, triggering safety unalignment. We first theoretically analyze the existence and geometric location of OMCB, revealing that the parameter shift crossing safety decision boundaries is the main cause of backdoor activation. Building upon this insight, we propose SMOKE, a meta-learning-based attack framework. SMOKE implants OMCB via a bi-level optimization strategy: the inner loop synthesizes surrogate meta-vectors through random low-rank initialization and finite-step maximization, while the outer loop optimizes the target model by simulating the merging process. Furthermore, preliminary analysis shows that the injected OMCB has bounded generalization loss under unknown merging configurations. Extensive experiments across diverse model architectures and merging techniques validate the effectiveness of SMOKE and its resistance to existing backdoor detections, highlighting a significant yet previously overlooked vulnerability in the model merging paradigm.
[NLP-311] VoTSpeech: Decoupling Voice Design from Speech Generation in Instruction-Following TTS
【速读】： 该论文旨在解决指令跟随式文本到语音（instruction-following text-to-speech, TTS）中语音设计隐式化的问题，即传统说话人嵌入（speaker embedding）依赖参考音频而非文本驱动的目标，导致语音风格控制能力受限。其核心解决方案是提出VoTSpeech框架，通过引入一个显式的中间语音潜在变量（voice latent），实现语音设计与语音生成的解耦。该方法基于共享语言模型主干网络，利用基于流（flow-based）的模块从文本指令中预测连续的语音潜在表示，该潜在变量作为共享条件同时作用于语言模型与声学扩散变换器（Acoustic DiT），分别引导自回归序列建模和声学潜在变量生成；随后通过音频变分自编码器（audio VAE）解码器将声学潜在变量还原为波形。为捕捉说话人身份及互补的声学细节，语音潜在变量同时受到说话人嵌入与时间池化的音频VAE特征的联合监督。通过任务特定适配，VoTSpeech在InstructTTSEval-ZH基准上表现优异，展现出与主流开源模型相当的性能，同时保持高自然度。实验结果验证了显式语音潜在设计及其条件接口的有效性，显著提升了文本驱动语音合成中的可控性与表现力。
链接: https://arxiv.org/abs/2610.14694
作者: Wenbing Yang,Qihang Lu,Zihan Sun,Peilei Jia,Yingming Gao,Ya Li,Jun Gao
机构: 未知
类目: ound (cs.SD); Computation and Language (cs.CL)
备注: 5 pages, 1 figure, 2 tables
Abstract:Instruction-following text-to-speech (TTS) typically leaves voice design implicit, as conventional speaker embeddings encode reference audio rather than text-driven targets. We propose VoTSpeech, which decouples voice design from speech generation via an explicit intermediate voice latent. Built on a shared language model backbone, VoTSpeech first predicts a continuous voice latent from text instructions using a flow-based module. This latent serves as a shared condition for both the language model and the Acoustic DiT, guiding autoregressive sequence modeling and acoustic latent generation, respectively, before an audio VAE decoder converts the acoustic latents into waveforms. To capture speaker identity and complementary acoustic nuances, the voice latent is jointly supervised by speaker embeddings and temporally pooled audio VAE features. With task-specific adaptation, VoTSpeech performs strongly on InstructTTSEval-ZH, demonstrating competitive capabilities against strong open-source models while maintaining high naturalness. These results confirm that explicit voice latent design and its conditioning interface effectively empower text-driven voice design for controllable speech synthesis. Code, model, and audio samples are available at this https URL .
[NLP-312] PVFH-Bench: Benchmarking Premise-Induced Visual Fixation Hallucinations in Vision-Language Models
【速读】： 该论文旨在解决大型视觉语言模型（Large Vision-Language Models, LVLMs）中存在的对象幻觉（object hallucination）问题，尤其聚焦于一种由认知科学中感知偏差启发的新故障模式——前提诱导的视觉固定幻觉（Premise-Induced Visual Fixation Hallucination, PVFH）。PVFH表现为：当模型接受一个错误的文本前提时，会过度依赖视觉上易混淆但实际不存在的证据，进而生成关于并不存在却具有视觉合理性的物体的错误回答。为系统研究此现象，作者构建了PVFH-Bench基准测试集，能够自动生成视觉上可混淆但实际缺失的目标，并结合诊断子集追踪模型从物体识别到幻觉生成的演化过程。实验结果表明，10个主流LVLM均对PVFH高度敏感；更关键的是，显式的思维链（chain-of-thought）推理反而加剧了该问题，促使模型将微弱的视觉线索误判为支持幻觉对象的证据，暴露了当前方法在逻辑一致性与视觉真实性判断之间的深层缺陷。
链接: https://arxiv.org/abs/2610.14635
作者: Xinhao Wang,Junbo Cui,Zhenghao Liu,Yuan Yao,Xiaohua Li,Yu Gu,Ge Yu
机构: Northeastern University(东北大学); ModelBest Inc.(模型最佳公司); Nanjing University(南京大学); Tsinghua University(清华大学)
类目: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Large Vision-Language Models (LVLMs) have achieved remarkable progress across a wide range of multimodal tasks. However, object hallucination remains a persistent challenge, undermining model reliability in applications such as visual question answering, embodied agents, and human-AI decision support systems. In this paper, inspired by cognitive science studies on perceptual bias, we investigate a new failure mode termed Premise-Induced Visual Fixation Hallucination (PVFH). Specifically, PVFH refers to a premise-induced reasoning error in which models become anchored to visually confusable evidence, accept a false textual premise, and subsequently generate responses about an absent yet visually plausible object. To systematically study this phenomenon, we introduce PVFH-Bench, a benchmark that automatically constructs visually confusable but absent targets and pairs the core hallucination setting with diagnostic subsets to trace how models move from object recognition to Premise-Induced Visual Fixation Hallucination. Experimental results on 10 state-of-the-art LVLMs demonstrate that all evaluated models are highly vulnerable to Premise-Induced Visual Fixation Hallucination. Moreover, we find that explicit chain-of-thought reasoning can further exacerbate this failure mode by encouraging models to overinterpret weak visual cues as supporting evidence for hallucinated objects.
[NLP-313] Uniform Information Density-Based Preference Optimization
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）生成内容与人类偏好对齐的问题，尤其关注如何更精准地捕捉复杂且难以量化的语言偏好。现有方法多依赖于简单化的人类偏好判别标准，难以充分反映自然语言中深层次的语用与认知规律。为此，本文提出一种基于心理语言学理论——均匀信息密度（Uniform Information Density, UID）的新颖偏好优化算法（UIDPO）。其核心思想是：人类倾向于以信息分布均匀的方式进行语言表达，因为这种均匀性可降低认知加工负担。为实现该理论指导下的优化，作者构建了一个复合度量指标，融合全局标准差与局部二阶差分，以量化信息密度的均匀程度，并将其整合进偏好优化目标函数中。实验结果表明，UIDPO在AlpacaEval2、ArenaHard和MT-Bench等基准上表现与当前最优方法SimPO相当甚至更优，且在补语从句等语言结构的分析中展现出更强的自然语言信息均匀性建模能力。研究证实，对语言中信息速率的细微人类偏好进行建模，是提升大语言模型与人类意图对齐的有效途径。
链接: https://arxiv.org/abs/2610.14578
作者: Yongyuan Li,Yang Xu
机构: Southern University of Science and Technology(南方科技大学); Shenzhen, Guangdong, China(广东省深圳市)
类目: Computation and Language (cs.CL)
备注:
Abstract:To enable large language models to generate outputs that align with human preferences, optimization algorithms based on the comparison between chosen and rejected output pairs have been developed and been proven effective. Some recent endeavors have utilized certain linguistically-derived assumptions about the preference pairs to guide the design of more precise optimization directions. However, human preferences are complex and hard to measure by simplistic criteria, and it is important to incorporate broader and more sophisticated insights about preferred vs. not-preferred language. In this study, we adopt a well-established psycholinguistic theory, Uniform Information Density, to develop a novel preference optimization algorithm, UIDPO. The basic assumption is that humans prefer to produce language in which information is uniformly distributed, because uniformity induces less cognitive processing effort. To implement this idea into the algorithm, we derive a compound metric for information uniformity by combining the global standard deviation and the local second-order difference, and integrate it to the objective of preference optimization. We compare UIDPO with SimPO, a previous state-of-the-art method, obtaining comparable and sometimes better performance on AlpacaEval2, ArenaHard, and MT-Bench. Additional linguistic analysis on complement clauses also shows that UIDPO can better reflect the tendencies of keeping information uniform in natural language than baselines. Our results show that the subtle human preference over the information rate in language is a helpful source for achieving better human alignment for LLMs. The code repository is published here: this https URL.
[NLP-314] Reward-Aligned Distillation
【速读】： 该论文旨在解决生成式 AI 在强化学习中因奖励可验证性提升推理能力时所面临的计算成本高与样本效率低的问题。现有方法如基于策略的蒸馏（On-Policy Distillation, OPD）虽能提供密集的词元级监督，但其对所有词元进行无差别蒸馏，未能区分哪些教师提供的指导实际有助于提升任务奖励。传统词元选择方法依赖不确定性或师生差异等代理指标，无法准确识别真正改善奖励的监督信号。为此，本文提出奖励对齐蒸馏（Reward-Aligned Distillation, RAD），一种双层优化框架，通过预测每个词元的监督信号对学生模型更新后奖励的潜在影响来动态加权蒸馏信号。RAD 通过对临时优化步骤进行反向传播以评分词元监督质量，并基于初始权重进行投影上升更新，从而实现对单个梯度贡献的增强、抑制或反转。在相同训练响应预算下，RAD 在六项竞赛级数学基准上分别较最强基线提升 6.5 个百分点（Qwen3-1.7B）和 2.5 个百分点（Qwen3-4B-Instruct-2507）；其简化版本 RAD-Simple 也达到了相当的数学准确率。在 LiveCodeBench-v6 上，RAD 使用 Qwen3-1.7B (non-thinking) 和 OLMo3-7B-Instruct 分别取得 54.69% 和 58.67% 的 pass@1 成绩，优于最强基线 8.7 和 6.4 个百分点，验证了奖励引导的教师监督加权策略在不同模型家族与任务间的有效性。
链接: https://arxiv.org/abs/2610.14572
作者: Yilan Chen,Zian Su,Manish Prajapat,Dingmin Wang,Rein Houthooft,Rui Song,Vijay Lingam
机构: UC San Diego; Purdue University; AWS AI Labs; Amazon
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Reinforcement learning with verifiable rewards improves the reasoning capabilities of large language models, but remains computationally expensive and sample inefficient. On-policy distillation (OPD) provides dense token-level supervision from a stronger teacher, yet standard OPD distills every token regardless of whether the teacher’s preference actually improves task reward. Existing token-selection methods rely on proxies such as uncertainty or teacher–student divergence, which need not identify reward-improving supervision. We introduce Reward-Aligned Distillation (RAD), a bilevel framework that weights distillation signals based on their predicted effect on the student’s post-update reward. RAD differentiates through a temporary optimizer step to score each token’s supervision, then updates the token weights by projected ascent from the initial token weights. The resulting weights can strengthen, suppress, or reverse individual distillation-gradient contributions. At a matched training-response budget, RAD improves mean accuracy on six competition-math benchmarks over the strongest baselines by 6.5 percentage points for Qwen3-1.7B (non-thinking) and 2.5 points for Qwen3-4B-Instruct-2507. RAD-Simple, a simplified variant, achieves comparable math accuracy at the same training-response budget. On LiveCodeBench-v6, RAD reaches 54.69% and 58.67% pass@1 with Qwen3-1.7B (non-thinking) and OLMo3-7B-Instruct, respectively, exceeding the strongest baselines by 8.7 and 6.4 points. These results support reward-guided weighting of teacher supervision across model families and tasks.
[NLP-315] LongHisDoc: A Comprehensive Benchmark for Chinese Long Historical Document Understanding
【速读】： 该论文旨在解决长篇历史文献（Long Historical Document）理解中缺乏系统性评估基准的问题，尤其针对非专业读者在解读复杂中文历史文献时面临的显著认知挑战。尽管生成式 AI（Generative AI）和多模态大模型（如大型视觉语言模型，LVLMs）在现代文档任务中表现优异，但在处理整本书籍规模的历史文献时仍存在能力空白。为此，作者提出了首个面向长篇历史文献理解的基准数据集 LongHisDoc，包含1,012个专家标注的问答对，覆盖101份PDF文档，平均每份文档达85.1页、42,776.4个词元，涵盖10个类别及四种问答类型（单页、跨页、多跳、不可回答），要求模型融合文本、版式与视觉元素等多模态证据进行推理。实验表明，当前最优的LVLM与LLM在该基准上的准确率分别仅为62.9%和69.9%，揭示了现有模型在长序列、多模态历史文档理解方面存在显著性能差距。解决方案的关键在于构建一个结构化、多样化且具备真实复杂性的多模态评估框架，并系统评估多种模态检索器与端到端检索增强生成（RAG）系统的效能，为未来研究提供可复现的基准与技术参考。
链接: https://arxiv.org/abs/2610.14531
作者: Zeyu Shan,Yongxin Shi,Shihang Wu,Peirong Zhang,Yuanyuan Fang,Zhaopeng Wang,Lianwen Jin
机构: South China University of Technology (华南理工大学); Sichuan University (四川大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Chinese historical documents are invaluable cultural treasures, but their complexity poses significant comprehension challenges for non-specialists. Despite the success of Large Language Models (LLMs) and Large Vision-Language Models (LVLMs) in modern document tasks, their capabilities in Long (full-book) Historical Document understanding remain largely unexplored. To bridge this gap, we introduce LongHisDoc, the first benchmark tailored for long historical document understanding. LongHisDoc comprises 1,012 expert-annotated QA pairs derived from 101 PDF documents, with an average of 85.1 pages and 42,776.4 tokens per document. It spans 10 categories and features four QA types: single-page, cross-page, multi-hop, and unanswerable. Answering these questions requires integrating multimodal evidence from text, layout, and visual elements. Our evaluation reveals that the top-performing LVLM and LLM achieve only 62.9% and 69.9% accuracy, respectively, highlighting a substantial performance gap. Furthermore, we assess various modal retrievers and the end-to-end RAG performance to provide references for Retrieval-Augmented Generation (RAG) systems. We believe that LongHisDoc serves as a comprehensive benchmark to catalyze future research in full-book historical document understanding. The benchmark and evaluation code are available at this https URL and this https URL.
[NLP-316] Query-Adaptive Neuron Steering for LLM s
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在推理阶段有效激发其潜在能力的挑战，尤其关注传统方法中预设上下文或固定干预策略难以适应多样化查询的知识与推理需求的问题。核心问题在于：如何实现模型内部调控（internal steering）的查询自适应性？其解决方案的关键在于提出Q-NeuS框架，该框架基于查询诱发的内部激活模式，动态识别并调节与当前输入主题相关的神经元。具体而言，Q-NeuS构建了一个紧凑的主题引导候选神经元空间，并引入轻量级的查询自适应门控机制（Query-Adaptive Gate），以生成逐神经元的调制系数；同时，通过引入Neu-GRPO算法，利用多组神经元干预配置间的相对性能差异进行优化，从而学习高效的干预策略。整个过程保持基础模型冻结，无需任务或主题标签，具备良好的跨模型尺度泛化能力与跨基准迁移性。
链接: https://arxiv.org/abs/2610.14519
作者: Yanxi Pan,Jingwen Wu,Yan Ma,Xiang Chen,Xinyi Zeng,Xuetong Wang,Yunyan Zhang,Yue Zhang,Yu Tian,Jianfeng Pan,Hongyi Zhou
机构: Northeastern University(东北大学); Washington University in St. Louis(圣路易斯华盛顿大学); Nanjing University of Aeronautics and Astronautics(南京航空航天大学); Tsinghua University(清华大学); 360 Security Capability Center(奇安信安全能力中心); Tencent(腾讯); Beijing University of Posts and Telecommunications(北京邮电大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Effectively eliciting the capabilities of large language models (LLMs) at inference time remains an important challenge. Existing studies address this challenge through context engineering, which shapes user queries, or fixed steering, which directly intervenes in model computation. However, predefined contexts and fixed intervention strategies may not adequately accommodate the diverse knowledge and reasoning requirements of individual queries. This raises a natural question: Can internal steering in LLMs be made query-adaptive? In this work, we investigate whether query-induced internal activations can guide adaptive steering in LLMs. Our pilot studies reveal subject- and layer-dependent neuron activation patterns and show that amplifying subject-relevant neurons can improve model performance. Based on these findings, we propose Q-NeuS, a query-adaptive neuron steering framework that dynamically determines which neurons to modulate and by how much for each input. Specifically, Q-NeuS constructs a compact subject-guided candidate neuron space and employs a lightweight Query-Adaptive Gate to generate neuron-wise modulation coefficients. To learn effective intervention strategies, we introduce Neu-GRPO, which adapts group-relative optimization to train the Query-Adaptive Gate using the relative performance of multiple neuron intervention configurations. The base LLM remains frozen throughout training and inference, and no task or subject labels are required at inference time. Extensive experiments across multiple LLMs and benchmarks demonstrate that Q-NeuS consistently improves model performance and exhibits broad applicability across model scales and strong cross-benchmark transferability.
[NLP-317] AphaVoice: Text-to-Speech Model for Aphasia Patient Simulation
【速读】： 该论文旨在解决卒中后失语症（post-stroke aphasia）患者语言治疗资源匮乏的问题，特别是医疗专业人员缺乏高质量训练模拟工具的困境。现有AI研究主要集中于语音处理任务，而对失语症语音生成（aphasic speech generation）这一关键临床需求关注不足。为此，本文提出AphaVoice——一种专为合成高度逼真失语症语音而设计的文本到语音（text-to-speech, TTS）框架。其核心创新在于通过潜在空间偏移（latent space shifting）技术，在不牺牲音素真实性的前提下实现多说话人生成，并在时序特征上显著提升对失语症典型表现如言语不流畅性（dysfluency）的建模能力。实验表明，AphaVoice在多项时序指标上优于当前最先进的TTS模型，同时保持稳定的说话人身份一致性，为失语症医学培训与临床仿真提供了可扩展、高保真的解决方案。
链接: https://arxiv.org/abs/2610.14479
作者: Al Varrel Putra Kusuma,Fernando Bello,Joshua Brown
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted at IEEE Spoken Language Technology (SLT) 2026
Abstract:While language therapy is the primary treatment for post-stroke aphasia, limited clinical resources often restrict patient access to proper care. Addressing this gap requires reliable training simulations for medical professionals. However, current AI efforts focus almost exclusively on speech processing, leaving speech generation largely unexplored. This study introduces AphaVoice, a novel text-to-speech (TTS) framework explicitly engineered to synthesize highly realistic aphasic speech. Evaluated against state-of-the-art TTS systems, AphaVoice demonstrates a superior ability to capture the clinical feature of dysfluency, outperforming existing models in temporal metrics while consistently maintaining speaker identity. By leveraging latent space shifting, the model also supports multi-speaker generation without sacrificing phonetic realism. Ultimately, AphaVoice provides a scalable, high-fidelity tool for medical training and clinical simulation related to aphasia.
[NLP-318] Scale Drivers and Limits of Representational Alignment Across Large Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在表征层面的相似性分析中，因采用聚合性相似度度量而掩盖模型间对局部细微差异的共识及其支持因素的问题。现有方法如标准表示相似性分析（Representational Similarity Analysis, RSA）将所有输入对的整体相似性进行汇总，难以揭示模型在不同几何尺度（如局部邻域与全局结构）上的对齐模式差异。为此，论文提出两种互补方法：邻域加权表示相似性分析（Neighborhood-Weighted RSA, NW-RSA）与反事实表示相似性分析（Counterfactual RSA, CF-RSA）。NW-RSA通过根据刺激对在表征空间中的邻近程度进行加权，使局部邻域与全局几何结构的对齐程度得以在同一尺度下比较；CF-RSA则通过在不同模型间施加受控的语言扰动，识别出哪些语言属性驱动了表征对齐。研究发现，尽管模型在粗粒度表征几何上表现出强一致性，但在局部邻域存在显著分歧；这种对齐不仅存在于广泛采样的自然文本中，也出现在保持特定语义或结构特征的语言等价类内。反事实分析进一步表明，逻辑范围（logical scope）的变化对模型对齐的影响远大于句法层级、事件角色或话语结构的变化，且当句子结构或词汇意义被破坏时，对齐程度下降，说明自然语言的规律性是促进模型收敛的关键因素。因此，该研究揭示了表征普遍性（representational universality）的复杂性——整体高相似性并不意味着模型在精细语言区分上的共识一致，从而为理解大模型内部表征机制提供了更精细的分析框架。
链接: https://arxiv.org/abs/2610.14476
作者: Jialin Wu,Meenakshi Khosla
机构: University of California, San Diego(加州大学圣地亚哥分校); Department of Cognitive Science(认知科学系); Department of Computer Science and Engineering(计算机科学与工程系)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Comparing large language models can reveal shared representational structure and systematic differences between them. However, aggregate similarity measures can obscure how agreement varies across geometric scales and which linguistic properties support it. We introduce two complementary methods to resolve where models agree and what supports that agreement. Standard representational similarity analysis (RSA) summarizes similarity across all stimulus pairs, potentially obscuring differences in alignment between nearby and distant inputs. Neighborhood-Weighted RSA (NW-RSA) addresses this limitation by weighting stimulus pairs according to their proximity in representational space, placing alignment from local neighborhoods to global geometry on a common scale. Counterfactual RSA (CF-RSA) tests which linguistic properties support alignment by comparing intact representations in one model with representations under controlled linguistic perturbations in another. Across diverse language models, we find strong agreement in coarse representational geometry but substantial divergence in local neighborhoods. Alignment is observed both across broadly sampled natural text and within linguistic equivalence classes that preserve particular aspects of meaning or structure. Counterfactual analyses reveal selective sensitivity: changes to logical scope reduce alignment substantially more than the tested changes to hierarchical syntax, event roles, or discourse structure. Alignment also decreases when sentence structure or lexical meaning is disrupted in word lists and jabberwocky sentences, suggesting that the regularities of natural language support stronger convergence. These findings refine claims of representational universality by showing that strong overall similarity need not imply agreement in how models represent finer linguistic distinctions.
[NLP-319] EdgeLLM -Nano: Auditing Polar-Guided Key-Cache Compression and Consistency-Anchored Pruning on a 33.6 MB On-Device Language Model
【速读】： 该论文旨在解决在边缘设备（edge devices）上部署语言模型（Language Models, LMs）所面临的内存、延迟和功耗等独特挑战，尤其聚焦于键值缓存（KV-cache）压缩技术的有效性评估问题。现有研究虽提出多种KV-cache压缩方法，但其在真实边缘环境下的性能表现尚未得到充分验证，且多数评估方法未真正将待测的KV-cache纳入考量，导致结果失真。为此，论文提出EdgeLLM-Nano，一个基于仅8.39M参数（33.6 MB）的纳米级解码器仅（decoder-only）Transformer测试平台的测量研究，实现对每个后端、种子及功耗样本的端到端评估。研究发现，极性引导的键缓存量化器（如采用四舍五入至最近（RTN）控制）以及一致性锚定的渐进式权重剪枝器（其一次性基于幅度的剪枝效果不劣于甚至优于复杂方法）均表现不佳，表明当前主流压缩策略在边缘场景下存在显著局限。为更精准评估压缩效果，论文提出长上下文解码保真度（Long-Context Decode Fidelity, LCDF），通过教师强制下的逐标记对数损失（log-loss）分析，结合跨种子配对统计与效应量，能够有效区分不同压缩后端的表现；相较之下，传统困惑度（perplexity）对所有方法给出相同分数（5.461），无法揭示差异。基于配套的Raspberry Pi 5实测数据，所有测试压缩后端在每标记能耗上均劣于未压缩的FP16推理，除KIVI外，其余方法在LCDF指标上均出现明显退化。因此，研究结论表明，尽管KV-cache压缩仍具潜力，但现有方法需进一步优化，而LCDF提供了一种简洁高效、可揭示被困惑度忽略的负面效应的评估范式。
链接: https://arxiv.org/abs/2610.14475
作者: S M Mahbub Murshed,Hayden Bright,Rezaul Karim,Ali Shiri Sichani
机构: Causal Dynamics Lab; University of Missouri-Columbia (密苏里大学哥伦比亚分校); University of Chittagong (吉大港大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Deploying language models (LMs) on the edge presents unique memory, latency, and power challenges absent in high-performance computing. KV-cache compression emerges as a promising approach to address these challenges. However, the effectiveness of current KV-cache compression methods has not been thoroughly examined on edge devices. Moreover, we show that existing evaluation methods do not always involve the KV-cache under test. We propose EdgeLLM-Nano, a measurement study of KV-cache compression using a nano-scale decoder-only LM (transformer) testbed that has 8.39M parameters, i.e., 33.6 MB, to enable end-to-end examination of each backend, seed, and power sample. Our results give a negative verdict for a polar-guided key-cache quantizer (round-to-nearest (RTN) control causes an order of magnitude less damage), as well as a consistency-anchored progressive weight pruner (one-shot magnitude-based pruning performs as well as, if not better). We also introduce Long-Context Decode Fidelity (LCDF) evaluation, a token-by-token examination of teacher-forced log-losses using the KV-cache under test. Unlike perplexity that gives the same 5.461 score to all methods, LCDF distinguishes different KV-cache compression backends thanks to its paired statistics across seeds and effect sizes. Our companion Raspberry Pi 5 measurement shows all compression backends we tested to be less energy-efficient per token compared to uncompressed FP16 inference. Except for KIVI that is close to lossless in terms of LCDF, our findings indicate KV-cache compression can be improved. LCDF offers a simple and effective way of shedding light on KV-cache compression methods, whose negative impacts have been overlooked by perplexity.
[NLP-320] Evaluating Bias in Chinese Large Language Models across Tasks and Registers: Task Operationalization Effects and Irony Classification Risks
【速读】： 该论文旨在解决在中文内容治理中，大语言模型（Large Language Models, LLMs）对有毒及歧视性表达识别的偏差测量受任务操作化（task operationalization）与反语表达（ironic expression）影响的问题。其核心挑战在于，同一内容在不同任务形式或语言风格（如反语式“阴阳”语境）下，模型的分类表现存在显著差异，导致偏差评估结果不可靠。解决方案的关键在于：首先，通过预注册的多任务行为审计，系统验证了任务形式与语用风格对模型表现的影响；其次，发现使用显式反语感知提示（irony-aware prompts）可显著降低错误率9.6个百分点；最后，通过将29个共享命题锚点转化为三种任务形式（社会共识估计、高风险决策、控制关联评分），揭示任务操作化本身会引发评估结果的系统性偏差，尤其在高风险决策任务中与社会共识任务之间存在显著收敛度差异。因此，研究强调，在评估中文大语言模型的内容安全性能时，必须将反语语境作为关键测试条件，并基于实证数据考察任务设计对偏差读数的影响，以实现更稳健、可复现的安全评估。
链接: https://arxiv.org/abs/2610.14464
作者: Mo Zhou
机构: 未知
类目: Computation and Language (cs.CL); Computers and Society (cs.CY)
备注: 11 pages, 3 figures
Abstract:In Chinese content governance, large language models are increasingly used to identify toxic and discriminatory expressions, yet both task operationalization and ironic expression can alter bias measurement readings. We conducted a preregistered multi-task behavioral audit of multiple Chinese large language models. In the content classification experiment, the core panel’s classification error rate rose from 10.9% in the formal register to 23.5% in the yinyang (an echoic-irony register) version, a risk difference of +12.7 percentage points. (95% CI [+9.7, +15.7]). Inter-rater agreement was lower in the yinyang register (majority-vote kappa = 0.299), so the increase contains a component of label uncertainty propagation, but the direction of increase was consistent under both adjudicated and majority-vote reference standards. Explicit irony-aware prompts reduced the intervention panel’s error rate by 9.6 percentage points. We operationalized the 29 shared proposition anchors into three task forms: social consensus estimation (Form A), high-stakes decision (Form B), and controlled association rating (Form C); independent manipulation checks found that all 58 A-C comparison units received unanimous agreement, whereas the low A-B convergence was confirmed by calibrated majority adjudication (87.8% supporting the minority judgment); on the irony-probe materials, the interactions of B and C relative to A were +0.577 and +1.121, respectively. Thus, when evaluating large language models for Chinese content governance, the effect of task operationalization on bias readings is an empirical question that requires data to answer, and safety evaluation should incorporate the ironic register as a test condition.
[NLP-321] Re-executing Coding-Agent Evaluations: Unchanged Decisions and Unresolved Policy Risk
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）编码代理在自动化测试评估中因执行环境不一致导致的评测结果不可靠问题，尤其关注已通过评审的代码补丁在不同执行环境下是否仍能保持其评分一致性。其核心解决方案在于通过严格控制实验条件，采用可复现的离线执行策略，结合逆采样概率加权估计器和模块依赖缓存重建等方法，对历史评测数据进行重执行验证。关键发现表明：尽管多数最佳代理选择与系统变体排名得以保留，但部分评测结果受主机、镜像及评分器配置差异影响，存在约0.83%的三轮多数判定变更率；同时，基于参考解依赖构建的模块缓存可有效识别出本应被拒绝却因依赖差异而通过的替代方案，揭示了现有评测框架中潜在的“暴露补丁风险”（exposed-patch risk）。研究强调了对评分器审计机制的重要性，并提供了匿名可复现的证据包，为提升评测系统鲁棒性与可信度提供了实证基础。
链接: https://arxiv.org/abs/2610.14460
作者: Tianyu Ding,Yaocheng Tong,Muzhi Yang,Fei Jiang,Lei Yang,Boya Yu
机构: 未知
类目: oftware Engineering (cs.SE); Computation and Language (cs.CL)
备注: 19 pages, 3 figures. Preprint
Abstract:Executable tests grade large language model (LLM) coding-agent patches for leaderboards, rewards and self-improvement. Two preregistered re-execution studies returned negative results. SWE-bench Pro v1’s offline option rejected none of 59 accepted Go patches whose grading configuration changed. Their exploratory, unweighted upper confidence limit is 4.95% under candidate independence, leaving a 3% exposed-patch risk unresolved; the preregistered estimator uses inverse sampling-probability weights. An extension was inconclusive. In a selected stratum, a reconstructed module cache built from the reference solution’s dependencies rejected accepted alternatives from four models that passed with their own dependencies. Re-executing 1,929 selected agent-version/task evaluations from the Darwin Gödel Machine’s archive preserved best-agent selection and system-variant ranking, with search history and scores of evaluations not rerun fixed. Exploratory host, image and adapted-grader comparisons also preserved them. Seven of 23 disagreements with archived verdicts did not recur on a second host; rebuilt images changed 16 three-run majority verdicts (0.83%) under four-CPU execution. We provide an evidence-to-claim summary, lessons for grader audits and an anonymous replication package. These results concern accepted patches and fixed-history decisions; general policy safety and adaptive-learning robustness require broader evaluations.
[NLP-322] Evaluating Semantic Relation Knowledge through Bidirectional Relation Prediction from Gloss Pairs
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在语义关系理解方面的知识局限性问题，特别是其对词义间语义关系（如上下位关系、整体-部分关系等）的识别与对称性/非对称性特征的把握能力。其核心解决方案在于提出一个具有挑战性的评估任务：给定两个词义定义（glosses）以不同顺序呈现，要求模型预测二者之间的语义关系，并通过判断两种顺序下的预测是否在对称性或非对称性上保持一致性来评估模型表现。该任务支持单语及跨语言评估，能够检验相同词义在不同语言中以不同表述形式呈现时，模型是否仍能维持对语义关系的正确理解。研究构建了一个包含3,069对词义的多语言数据集（涵盖英语、希腊语、罗马尼亚语、土耳其语和日语），实验结果表明，在零样本与五样本设置下，大多数LLMs表现不佳，仅有少数模型显著优于随机基线，且无一在所有关系类型上均表现出稳健的能力，尤其在上下位关系（hypernymy）的双向一致性预测以及整体-部分关系（holonymy）的正向预测上存在明显缺陷。这些发现揭示了当前主流大语言模型在语义关系建模方面仍存在显著不足，包括关系识别能力弱以及对语义关系不对称性的理解不充分。
链接: https://arxiv.org/abs/2610.14442
作者: Zhihan Cao,Adam Nohejl,Hitomi Yanaka
机构: The University of Tokyo(东京大学); RIKEN(理化学研究所); Tohoku University(东北大学)
类目: Computation and Language (cs.CL)
备注: under review
Abstract:Semantic relations organise word senses and are fundamental to lexical semantics. In order to evaluate semantic relation knowledge of large language models (LLMs), we propose a challenging task in which an LLM predicts the relation between two glosses (definitions of senses) presented in two different orders. The predictions for the two orders are evaluated for consistency with the symmetry or asymmetry of semantic relations. Our task is designed for both monolingual and cross-lingual evaluation, allowing us to examine whether relation knowledge remains accessible when the same senses are verbalised through glosses in different languages. We construct a dataset of 3,069 sense pairs, with each sense verbalised in English, Greek, Romanian, Turkish, and Japanese. Experiments with 17 LLMs in zero-shot and five-shot settings reveal generally poor performance. In each setting, only eight LLMs achieve a macro-averaged F1 score significantly above a random baseline, and only one (zero-shot) and two (five-shot) significantly outperform a neural baseline. In particular, LLMs often fail to generate bidirectionally consistent predictions for hypernymy and to make correct predictions for holonymy in either direction. Our results suggest limitations in current LLMs’ semantic relation knowledge, both in identifying relations and in respecting asymmetry.
[NLP-323] Neither True nor False: Do LLM s Represent Truth-Value Gaps in Plural Predication?
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在处理复数描述的语义真值条件时是否存在对“同质性”（homogeneity）现象的理解问题，即当仅部分主体满足谓词时，复数定指表达式（如“the kids laughed”）及其否定形式（如“the kids didn’t laugh”）应处于非真非假的中间状态，而其量化对应句（如“all the kids laughed”或“not all the kids laughed”）则保持二值性。研究揭示了普遍存在的“二值化偏差”（Bivalence Bias）：当前开放权重的LLMs倾向于将复数定指表达强制赋予同质性真值条件，即将“the kids laughed”解读为“所有孩子都笑了”，将“the kids didn’t laugh”解读为“没有一个孩子笑”，从而将本应为三值逻辑（true/false/indeterminate）的语义间隙压缩为单一的“假”。通过构建包含1000个最小对比对的诊断数据集HomogeneityNLI，研究发现，在零样本设置下，无一模型能在超过15%的情境中正确识别两种极性的语义间隙，且否定定指句被判定为“假”的比例高达84%。尽管引入显式语言规则、思维链（chain-of-thought）和列举提示等干预手段，其效果仍不稳定且模型特异，常进一步固化该偏差。综上，当前开放权重的LLMs虽习得了复数命题的真值条件，但尚未掌握其三值逻辑的本质。
链接: https://arxiv.org/abs/2610.14438
作者: Duaa Nakshbandi,Zhiyu An,Yiping Li,Sachin Giri,Wan Du
机构: University of California, Merced(加州大学默塞德分校)
类目: Computation and Language (cs.CL)
备注:
Abstract:Do Large Language Models (LLMs) genuinely grasp the truth conditions of plural descriptions, or do they force every sentence into true or false? We investigate the Homogeneity Gap, a semantic phenomenon where a plural definite (e.g., the kids laughed) and its negation (the kids didn’t laugh) are both neither true nor false when only some of the kids laughed, whereas their quantified counterparts (all, not all) remain strictly bivalent. We introduce HomogeneityNLI, a diagnostic dataset of 1,000 minimal-pair items crossing determiner, polarity, and group composition across ten logical conditions. Evaluating seven open-weight models, we uncover a pervasive Bivalence Bias: models assign plural definites their homogeneous truth conditions, reading the kids as all of them and the kids didn’t as none of them, yet collapse the truth-value gap into falsity. In the zero-shot setting, no model recognizes the gap for both polarities in more than 15% of scenarios, and negated definites are judged False in 84% of cases. Explicit linguistic rules, chain-of-thought, and enumeration prompts produce unstable, model-specific effects that often entrench the bias. Taken together, our findings indicate that current open-weight LLMs have acquired the truth conditions of plural predication but not its trivalent logic.
[NLP-324] MemForm: Benchmarking Memory Formation Poisoning in LLM Agents
【速读】： 该论文旨在解决大语言模型（LLM）智能体在长期记忆形成过程中易受对抗性污染的问题，特别是现有研究对记忆形成接口的表征不足，导致难以系统识别和防御潜在攻击。其核心解决方案是提出一种名为MemForm的受控诊断基准，首次将记忆形成污染（Memory Formation Poisoning）建模为对持久记忆生成与提交接口的对抗性操纵，并系统定义了五类关键接口：内容（Content）、证据（Evidence）、范围（Scope）、选择（Selection）与路径（Path）污染。其中，引入“选择污染”与“路径污染”作为受控干预类别，分别用于隔离保留机制与记忆巩固顺序的影响。实验表明，六种主流LLM在不同响应格式下均在所有五个接口上表现出广泛脆弱性，且即使经历真实成功经验，仅通过改变上下文支持、保留策略或巩固顺序，仍可诱发不可预期的下游行为，尤其路径污染在相同保留内容下因形成过程差异而显著改变行为表现。进一步分析揭示接口有效性依赖于具体记忆架构，且路径污染敏感性与更新器设计相关。现有防御措施对不同接口的缓解效果不一，表明需同时保护记忆内容及其形成过程。该研究强调构建鲁棒记忆系统的必要性，需从记忆形成机制本身进行安全设计。
链接: https://arxiv.org/abs/2610.14431
作者: Haowen Pan,Pengyang Shao,Yunshan Ma,Yanzheng Jin,Naixin Zhai,Xiang Wang,Xun Yang
机构: University of Science and Technology of China (中国科学技术大学); National University of Singapore (新加坡国立大学); Singapore Management University (新加坡管理大学)
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL)
备注:
Abstract:Long-term memory enables LLM agents to learn from past interactions but exposes memory formation to adversarial influence. Existing studies primarily organize memory poisoning by attack channels, corruption patterns, or lifecycle stages, leaving formation interfaces insufficiently characterized. We formulate Memory Formation Poisoning as adversarial manipulation of the interfaces through which persistent agent memory is produced and committed, and introduce MemForm, a controlled diagnostic benchmark covering five interfaces: Content, Evidence, Scope, Selection, and Path Poisoning. We introduce Selection and Path Poisoning as controlled intervention categories that isolate retention and consolidation order effects, respectively. Each category pairs clean and poisoned conditions with interface-specific controls. Evaluations across six LLMs and two response formats reveal broad susceptibility at all five interfaces under the evaluated memory protocols and implementations. Crucially, authentic experiences with successful outcomes do not guarantee reliable behavior after memory formation: changes to contextual support, retention, or consolidation order can induce undesirable downstream behavior. In particular, consolidation order can alter behavior even with identical retained experiences, depending on the memory formation procedure. Additional analyses show that interface applicability depends on memory architecture and that Path Poisoning susceptibility depends on the updater. The evaluated defenses provide incomplete mitigation that varies by interface. These findings motivate protecting both what is committed to persistent memory and how memory is formed from experiences. Code and data are available at this https URL.
[NLP-325] Scaling Laws for Multimodal Data Mixtures NEURIPS2026
【速读】： 该论文旨在解决前沿多模态大模型（Multimodal Large Language Models, MLLMs）在预训练过程中对多模态数据混合策略缺乏科学理解的问题，尤其是如何合理分配文本、视觉与音频三种模态的数据比例，以及由此带来的跨模态干扰成本。其核心解决方案在于基于混合专家（Mixture-of-Experts, MoE）架构，在参数规模从4.57亿到83亿、数据量高达1500亿token的范围内系统性地开展268组实验，推导出适用于三类模态的数据混合缩放定律。关键发现包括：计算资源优化策略具有模态特异性——音频模态更依赖模型参数扩展而非数据量；文本与视觉模态则需平衡模型规模与数据量；重复使用视觉或音频数据超过两倍后收益趋近饱和；且跨模态干扰呈现显著非对称性：视觉对音频学习造成显著干扰，而音频对文本和视觉的学习影响可忽略。这些发现为训练前沿多模态大模型提供了可操作、可量化的数据混合设计原则。
链接: https://arxiv.org/abs/2610.14423
作者: Aditi Khandelwal,Ayush K. Tarun,Yixuan Xu,Imanol Schlag,Golnoosh Farnadi,Siva Reddy,Luke Zettlemoyer,Antoine Bosselut
机构: MILA - Quebec AI Institute, McGill University(蒙特利尔人工智能研究所，麦吉尔大学); EPFL(洛桑联邦理工学院); ETH(苏黎世联邦理工学院); University of Washington(华盛顿大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted as a spotlight in NeurIPS 2026
Abstract:Frontier AI systems are increasingly natively multimodal, jointly pretrained on multiple modalities, such as text, vision, and audio. However, the scientific understanding of the multimodal data mixtures used to pretrain these systems — how much of each modality to mix, and at what potential interference cost to the others — is largely absent or gatekept. In this work, we derive data mixing scaling laws for three modalities: text, vision, and audio. Multimodal Large Language Models (MLLMs) increasingly adopt Mixture-of-Experts (MoE) architectures, as MoEs scale efficiently and partially limit cross-modal interference. We therefore use MoEs to conduct 268 experiments ranging from 457M to 8.3B parameters, on up to 150 billion tokens. Our work leads to multiple actionable insights into multimodal data mixing for training MLLMs. First, compute optima are modality-specific: audio modality is model-heavy, favoring scaling parameters over tokens, while text and vision require balanced scaling of model size and data. Second, repeating vision or audio data beyond 2x yields negligible benefits. Finally, cross-modal interactions are highly asymmetric: vision strongly interferes with audio, whereas audio’s interference with learning text and vision is negligible. Overall, we provide a principled foundation for understanding the multimodal data mixtures needed to train frontier MLLMs.
[NLP-326] SPROUT: Scaling Agent ic Reinforcement Learning for Long-Horizon Software Engineering
【速读】： 该论文旨在解决长时程软件工程任务中强化学习（Reinforcement Learning, RL）因成功轨迹稀缺而导致的训练效率低下问题。在这些任务中，任何微小错误都会导致整个执行序列失效，而传统方法依赖从初始状态重复重启，不仅效率低，且无法恢复如依赖项安装等环境状态变化。为克服此瓶颈，论文提出SnaPshot-based RollOUT（SPROUT）框架，其核心在于通过轻量级基础设施在每次智能体交互后对文件系统和代理上下文进行快照（checkpoint），从而支持从中间状态恢复与分支。该框架采用验证器引导的回溯策略，智能选择失败尝试的恢复点及后续推进路径。实验表明，SPROUT在多种大语言模型（LLM）上相较pass@k指标提升最高达23.6%，同时仅消耗其32%–52%的生成令牌；在Qwen3.8-27B模型的RL训练中，相比标准GRPO方法，在三个长时程编码基准上稳定提升13.9分，且每步训练时间减少39.5%。因此，SPROUT通过快照式状态恢复与智能分支机制，显著提升了长时程代码生成任务中强化学习的有效性与可扩展性。
链接: https://arxiv.org/abs/2610.14405
作者: Mengyang Liu,Weiming Wang,Haixin Nan,Yu Wu,Peize Li,Shihong Huang,Xuning He,Xue Jiang,Yihong Dong
机构: 未知
类目: oftware Engineering (cs.SE); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: work on process
Abstract:Reinforcement learning (RL) has driven substantial progress in coding agents by relying on verified successful trajectories as learning signals. However, such trajectories are scarce in long-horizon software engineering tasks, where a minor error can invalidate the entire rollout. Repeatedly restarting from the initial state is commonly used yet inefficient, and resetting only the context with working directory misses state changes such as installed dependencies. Therefore, branching from intermediate steps to preserve earlier progress requires restoring both the agent context and execution environments. In this paper, we present SPROUT (SnaPshot-based RollOUT), an agentic RL rollout framework that scales long-horizon coding tasks by resuming and branching from intermediate states. Specifically, SPROUT was built on top of a lightweight infrastructure that checkpoints both the filesystem and the agent context after each assistant turn. A verifier-guided backtracking policy then selects where to resume a failed attempt and how to proceed. Extensive experiments on long-horizon coding tasks show that SPROUT achieves up to 23.6% improvement over pass@k across various LLMs, while using 32 – 52% of their tokens. For RL training of Qwen3.8-27B, SPROUT stably outperforms standard GRPO by 13.9 points on three long-horizon coding benchmarks while cutting the time per training step by 39.5% . SPROUT thus makes RL on long-horizon coding tasks both more effective and cheaper to scale.
[NLP-327] RealityBench: Simulated Environments to Evaluate a Vision Language Models Ability to Document Human Physical Work
【速读】： 该论文旨在解决在真实物理工作场景中，如何通过视觉语言模型（Vision-Language Models, VLMs）有效理解作业人员的实时行为并提供辅助支持的问题，特别是在屋顶太阳能安装和车辆维修等高变异性、非结构化任务中的应用。其核心挑战在于现有模型在复杂现实环境下的泛化能力不足，且缺乏可复现的基准测试数据集，因为此类工作数据多为专有信息且具有高度情境依赖性。为此，研究提出在Blender中构建十一类真实工作环境的模拟系统，涵盖图像可接受性（acceptability）与任务状态理解（documentation）两类关键任务。实验表明，尽管顶级模型在答案显式出现在画面中的情况下能较好完成文档理解任务，但在读取测量值、识别微小部件及判断空间关系（如卷尺是否倾斜）方面表现较弱；而图像可接受性评估则因人类评审者间存在显著分歧，导致难以建立可靠标准，最优模型的表现仅相当于单个非作者评审者的水平。此外，在三个可追踪错误来源的环境中发现，多数误判源于对场景的误解而非规则理解偏差。值得注意的是，基于渲染图像上的模型排序结果可有效迁移至真实图像，验证了模拟数据在评估模型性能方面的可行性。因此，该研究的关键突破在于构建了一个可控、可重复且贴近实际的仿真基准框架，为未来生成式AI在物理世界协作任务中的评估与优化提供了重要基础。
链接: https://arxiv.org/abs/2610.14397
作者: Andrew Perrault(1),Anmol Takiar(1),Dominik Wienecki(1),Arnab Nandi(1) ((1) Relantic, Inc., Columbus, OH, USA)
机构: Relantic, Inc.(Relantic, Inc.)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 28 pages, 3 figures, 14 tables. Data: this https URL
Abstract:Augmenting worker capabilities in the physical world through guidance has been a decades-long goal for AI agents. One key supporting capability is to understand what the worker is doing, seeing through the human’s camera or smart glasses, and assist them in documentation. Despite advances in vision language models, their capabilities in these settings remain uncertain and are challenging to benchmark because most physical work data is proprietary, and involves too much variability to study in a principled manner. Unlike structured settings such as kitchens and warehouses that have predictable lighting, geometry, and objects, field-facing tasks such as rooftop solar installation and vehicle repair are unconstrained and high variance. Here, we model real-world physical work in Blender across eleven environments, incorporating variants in both image acceptability and documentation. The acceptability task asks whether a particular view would be suitable proof-of-work and documentation asks whether an agent can understand the task-relevant state of the environment from that view. Across 26 vision-language models, we find that top models largely solve documentation when the answer is text in frame, but are weak at reading measurements and at judging small parts and spatial relations, such as whether a tape measure leans. Acceptability is harder to pin down because human raters disagree with each other. Measured against a scorer that agrees with the raters’ majority about as often as a single rater does, the best models cannot be distinguished from a single non-author rater, at a cost within the range of a human pass. In the three environments where we can trace acceptability errors, most wrong rejections come from misreading the scene rather than the requirement. On the three documentation questions for which we have real images, model rankings on rendered frames transfer to real images.
[NLP-328] SciCurateBench: Curating Experimental Data from Heterogeneous Scientific Literature with Verbal-Action Agents
【速读】： 该论文旨在解决科学文献中实验数据精准提取与验证的难题，即如何从长篇、异构的原始科学文献中准确识别并验证满足多个生物实验条件（如特定检测方法、临床结局及数据可得性等）的条目。现有基准测试仅评估短查询下的粗粒度相关性，无法反映真实场景下复杂的联合条件验证需求。其核心解决方案是提出一个名为LangPoli的集成系统，该系统通过策略代理（policy agent）生成自然语言操作指南，由执行代理（actor agent）将其转化为可执行工具，从而实现对复杂实验条件的精确匹配。该方法在论文分类（F1达83.16%，显著优于基础大模型的58.22%）和结构化信息检索任务中表现优异，并提升了公开基准LitSearch、BEIR与TREC的重排序性能。此外，研究引入课程迁移（curriculum transfer）机制，将已解决问题中的通用策略进行提炼，以加速新任务求解。这一工作为构建真正实用的科学智能辅助系统提供了关键路径。
链接: https://arxiv.org/abs/2610.14393
作者: Minuk Ma,Kaiyun Guo,Jiarui Ding
机构: University of British Columbia (不列颠哥伦比亚大学)
类目: Computation and Language (cs.CL)
备注: 19 pages, 4 figures, 13 tables
Abstract:Scientific research increasingly depends on curating experimental data from literature. Scientific data curation requires verifying exact experimental conditions (e.g., specific assays, clinical outcomes, and data availability) across long, heterogeneous files, which is a highly time-consuming process. Existing benchmarks evaluate only broad relevance on short queries, failing to capture this complexity. We introduce SciCurateBench, a benchmark of 200 expert-annotated papers from major life-science journals that challenges systems to verify conjunctive biological experimental criteria from raw, unprocessed scientific literature. We show that plain LLMs and popular reasoning, tool-using, and feedback-based agents all struggle with this task. To study what makes the task tractable, we further propose LangPoli, an integrative system in which a policy agent generates natural language action guidelines that actor agents convert into executable tools. LangPoli showed strong performance on both Paper Classification (83.16% F1 vs. 58.22% for its base LLM) and Structured Information Retrieval, and also improves re-ranking on the public benchmarks of LitSearch, BEIR, and TREC. We further explore curriculum transfer, a method that distills generalizable strategies from solved tasks to accelerate the solving of related problems. We hope this work facilitates research in truly practical scientific AI assistance. The anonymized pre-release can be found from https://anonymous.4open. science/r/scicuratebench-D885/.
[NLP-329] okenization-Invariant Watermarking: A Byte-Local Method for LLM Text
【速读】： 该论文旨在解决生成式文本中水印（Watermarking）在面对局部编辑攻击时的鲁棒性不足问题，尤其关注现有方法在面对基于分词器（Tokenizer）的检测机制时，因分词过程引入的不一致性导致的检测失效或误判。其核心解决方案是提出一种分词无关水印（Tokenization-Invariant Watermarking, TI-WM），该方法在生成阶段嵌入水印，但将水印的密钥与评分操作基于经过归一化处理的UTF-8字节序列（canonicalized UTF-8 bytes），而非依赖分词器输出。这一设计使得检测过程无需调用分词器，从而避免了分词不确定性带来的干扰。关键创新在于：通过归一化字节序列进行评分，确保对保持字节归一化一致性的编辑（如字符替换、插入等）不会改变水印得分，而其他局部编辑仅影响有限数量的评分窗口，从而提供可量化的局部敏感性保障。实验表明，在5%词级编辑攻击下，TI-WM在Qwen3-8B和Gemma 4 E4B模型上分别实现了0.530和0.664的平均受攻击AUC，显著优于传统Unigram方法（分别为0.410和0.489），显示出更强的攻击排序能力。然而，该方法并未实现质量与清洁检测率的联合最优，且仍面临源域偏移（source shift）和校准检测性能的挑战，因此其贡献更在于提供了一种具有局部评分保证的可实现构造，而非通用鲁棒性或低误报率部署方案。
链接: https://arxiv.org/abs/2610.14378
作者: Tung-Ling Li
机构: 独立研究员(Independent Researcher)
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL)
备注:
Abstract:We introduce Tokenization-Invariant Watermarking (TI-WM), a method that embeds a watermark during token generation but keys and scores it on canonicalized UTF-8 bytes. The detector does not invoke the tokenizer. Canonicalizer-preserving edits leave its score unchanged; other localized canonical-byte edits affect a bounded number of scoring windows. Across three fixed keys and 128 prompts per key and model, a deterministic white-box attack at a 5% word-edit budget leaves mean attacked-AUC 0.530 on Qwen3-8B and 0.664 on Gemma 4 E4B, compared with 0.410 and 0.489 for Unigram with detection-side deletion. These operating points are not jointly matched on quality and clean detectability. Bias sweeps show model-dependent distortion costs, and shared-rule controls support a locality interpretation without establishing a unique advantage of bytes. Visible insertion outside the canonicalizer’s deletion set defeats the default design. Alphabetic filtering closes that channel but does not resolve source shift or calibrated detection power. TI-WM thus provides a concrete construction with local scoring guarantees and measured attack-ranking gains, not a general robustness or low-false-positive-rate deployment guarantee.
[NLP-330] Multimodal Co-Generation With Continuous Bitstream Diffusion
【速读】： 该论文旨在解决多模态生成任务中普遍存在的框架异构性问题，即不同模态（如图像-文本、语音-文本、蛋白质序列-结构）往往需要独立设计特定的模型架构、训练目标和采样策略，导致系统难以统一。其解决方案的关键在于提出一种名为连续比特流扩散（CoBit）的统一生成框架，该框架通过共享相同的训练与采样协议实现跨模态联合与条件生成。具体而言，所有模态均采用单一输入投影层、Transformer 主干网络、比特级预测头以及二元评分匹配（binary score-matching）目标函数，仅通过控制观察到的比特流部分即可灵活支持联合生成与条件生成。该方法在无需预训练词元化器的情况下，在 MNIST-Sum 任务上实现了近乎完美的条件生成准确率与高一致性；在大规模语音-文本与蛋白质序列-结构建模任务中，单个 CoBit 检查点即可完成语音合成、识别、续写、联合生成等多任务，并在生成结果的一致性评估中表现出色（如语音-文本联合生成达到 3.5% 的词错误率，蛋白质生成结构自洽 TM-score 达 0.911）。研究表明，CoBit 能够为差异显著的模态及跨模态依赖形式提供统一的生成范式，且无需引入领域特异性目标或采样机制。
链接: https://arxiv.org/abs/2610.14349
作者: Georgios Batzolis,Georgios Kontos,Bruno Trentini,Pranam Chatterjee,Micha Livne,Mark Girolami,Luca Ambrogioni
机构: Devotion AI Labs; University of Cambridge; University of Oxford; NVIDIA; University of Pennsylvania; Radboud University
类目: Computation and Language (cs.CL)
备注:
Abstract:We show that continuous bitstream diffusion (CoBit), originally developed for language modeling, provides a unified framework for joint and conditional multimodal generation using the same training and sampling protocol. Across image-text, speech-text and protein sequence-structure modeling, models are trained separately from scratch, but each uses a single input projection, transformer backbone, bitwise prediction head and binary score-matching objective shared across its modalities. Within each setting, a single checkpoint supports joint generation and the conditional directions included in training by changing which parts of the bitstream are observed. We first validate the approach on MNIST-Sum, a controlled image-text task pairing raw binary images with equations encoded as text, achieving near-perfect conditional accuracy and high joint consistency without learned tokenizers. We then apply the same protocol at larger scale to speech-text and protein sequence-structure modeling. Trained without text pretraining, a single speech-text model performs speech synthesis, recognition, continuation, and unconditional co-generation. Jointly generated speech and text achieve 3.5% word error rate between the generated text and automatic transcriptions of the generated speech. In protein modeling, a single CoBit checkpoint supports forward folding, inverse folding, and joint sequence-structure generation. For jointly generated proteins spanning 100-500 residues, computational refolding of CoBit’s sequences yields structures that agree with the generated backbones at a mean self-consistency TM-score of 0.911. Together, these results show that CoBit provides a common generative framework for substantially different modalities and forms of cross-modal dependence, without introducing a domain-specific objective or sampling procedure.
[NLP-331] Evaluating LLM s as Agent ic Workflow Executors in Real-World Customer Support Skills
【速读】： 该论文旨在解决企业客户支持场景中大型语言模型（Large Language Models, LLMs）在执行受限、工具调用型工作流时的性能评估难题。具体而言，研究聚焦于“技能型大模型”（Skill LLMs）——即在特定技能上下文中运行、需在政策与工具约束下完成窄范围、状态敏感任务的专用执行器。其核心挑战在于如何构建一个可量化、可复现的零样本评估基准，以准确衡量模型在真实复杂交互中的工具使用能力与合规性。解决方案的关键在于提出一套端到端的基准构建流程：从真实对话记录中挖掘高频重复的工作流，将其结构化为技能描述，并映射至对应的工具需求；通过兼容模型上下文协议（Model Context Protocol, MCP）的模拟环境对状态进行仿真；最终生成多轮工具调用对话数据集。该基准不仅支持大规模零样本评估，还引入了由五名强模型判官和一个基于规则的确定性评估器（用于检测工具覆盖率、幻觉式工具调用及不支持声明）组成的双重评价体系，揭示出判官选择对评估结果具有显著影响（平均Kappa=0.524），强调了评估可靠性与模型能力同等重要。
链接: https://arxiv.org/abs/2610.14346
作者: Md Tahmid Rahman Laskar,Gundeep Singh,Xue-Yong Fu,Karol Chang,Kevin Sanders,Shi Zong,Tania Habib,Julien Bouvier Tremblay,Shayna Gardiner,Harsh Saini,Matthias Lee,Elena Khasanova,Quinten McNamara,Shashi Bhushan TN
机构: Dialpad Inc.(Dialpad公司)
类目: Computation and Language (cs.CL)
备注:
Abstract:Enterprise customer-support agents increasingly rely on Large Language Models (LLMs) to execute bounded, tool-using workflows. We study this problem through the lens of Skill LLMs: specialized executors that operate within a selected skill context and must complete a narrow, stateful workflow under policy and tool constraints. We present a benchmark construction pipeline that mines recurrent workflows from real-world transcripts, converts them into structured skill descriptions, maps them to tool requirements, simulates environment state with Model Context Protocol (MCP)-compatible mocks, and generates multi-turn tool-using dialogues. Using this benchmark, we conduct a large-scale zero-shot evaluation of 17 language models spanning closed model families and deployment-oriented open-weight models from 1.7B to 14B parameters, each judged independently by five strong LLM judges, plus a deterministic rule-based evaluator that checks tool coverage, hallucinated tool use, and unsupported claims. We find a consistent ranking in which GPT-5.4 leads by a wide margin and QWEN3.5-9B is the strongest open-weight model (below 20B), outperforming several closed baselines under every judge. We also find that judge choice is not a negligible methodological detail: pairwise judge agreement on the pass/fail verdict averages Kappa=0.524 across 10 judge pairs. These results suggest that comprehensive zero-shot evaluation of skill execution requires judge reliability as carefully as model capability.
[NLP-332] Which Metadata Matters for LLM Tool Selection in MCP? Evidence from Controlled Factor-Level Interventions
【速读】： 该论文旨在解决在模型上下文协议（Model Context Protocol, MCP）系统中，尽管工具功能相似，但其参数模式（parameter schema）的差异是否会影响大型语言模型（LLM）在未实际调用前对工具的选择这一关键问题。研究发现，参数模式的变化是影响工具选择最显著且最一致的因素，模型更倾向于选择将参数以独立顶层形式暴露的工具，而非分组或复合结构；同时，详细解释参数的描述也能提升工具被选中的概率，而工具名称的影响则较弱且不一致。此外，当请求中明确要求特定参数格式时，所有五种模型均会选择符合该格式的工具，即便存在位置偏好。这表明参数模式不仅影响工具的调用方式，更直接决定了其是否被选中。因此，对于MCP开发者和评估者而言，参数模式的设计以及工具排序策略，在功能不变的前提下仍可能显著影响最终执行的具体实现，凸显了元数据设计在工具选择中的核心作用。
链接: https://arxiv.org/abs/2610.14334
作者: Xiaofan Li,Fengchun Qiao,Xing Gao
机构: University of Delaware (特拉华大学); University of South Florida (南佛罗里达大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Software Engineering (cs.SE)
备注:
Abstract:In Model Context Protocol (MCP) systems, LLMs select tools from metadata rather than from their implementations. Names and descriptions summarize what a tool does, while the parameter schema specifies the fields a tool accepts and how they are organized. Our analysis of metadata for 367,174 MCP tools reveals substantial schema variation among functionally similar tools. Many of these tools share names and similar descriptions but expose different parameter schemas. This raises a key question: can parameter-schema differences influence tool selection before any call is made? We study this question through controlled metadata interventions across five LLMs and five tasks. Each comparison changes exactly one metadata factor (server name, tool name, description, or parameter schema) while keeping the others fixed. We balance tool positions and use conditional logistic regression to estimate position-adjusted effects. We find that: (1) parameter-schema changes produce the largest and most consistent selection shifts among the tested interventions. Models tend to favor separate top-level parameters over grouped or composite forms. (2) Descriptions that explain parameters also tend to increase selection, while name effects are weaker and less consistent. (3) Targeted comparisons show that schema effects depend on the specific parameter names and on how the request represents the same information. When only one schema satisfies an explicit request requirement, all five models select the compatible tool despite position preferences. These results show that parameter schemas influence not only how tools are called, but also whether they are selected. For MCP developers and evaluators, schema design and tool ordering can therefore affect which implementation is invoked even when functionality is unchanged.
[NLP-333] ASSET: Adaptive Speech Segmentation via Reinforcement Learning
【速读】： 该论文旨在解决语音大语言模型（Speech LLM）中因声学编码器输出序列过长导致的模型扩展性差的问题。由于声学编码器生成的音频序列远长于文本输入，而大语言模型在处理长序列时性能显著下降，因此亟需一种高效压缩机制，在保留任务相关关键信息的前提下减少输入序列长度。其解决方案的关键在于提出ASSET框架，该框架基于强化学习（Reinforcement Learning, RL），通过自适应地将编码器输出划分为可变时长的片段，并对每个片段进行池化以生成连续的音频标记（audio token）。该方法采用组相对策略优化（Group-Relative Policy Optimization, GRPO）训练分割策略，结合任务性能反馈与标记率惩罚，实现对压缩过程的端到端优化。实验表明，在英语LibriSpeech数据集上，ASSET在仅使用约25%的音频标记数的情况下达到与基于转录本的字符边界最优方案相当的自动语音识别（ASR）性能；在中文AISHELL数据集上，其在相似标记率下优于基于音素边界的基准方案。此外，针对非ASR任务的探索性实验显示，该框架可根据任务需求动态调整分割模式，展现出良好的泛化能力。
链接: https://arxiv.org/abs/2610.14309
作者: Cihan Xiao,Yuan Tseng,Matthew Wiesner,Desh Raj,Leibny Paola Garcia Perera,Sanjeev Khudanpur,David Harwath
机构: Google(谷歌); Stanford University (斯坦福大学); MIT (麻省理工学院); Carnegie Mellon University (卡内基梅隆大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Speech large language models (LLMs) are often built using pretrained acoustic encoders as an audio interface to an LLM decoder. However, LLMs scale poorly with input sequence length, while acoustic encoders tend to produce substantially longer sequences than typical textual inputs. This motivates learning to compress encoder outputs while preserving task-oriented information. To this end, we introduce ASSET, a reinforcement learning (RL)-based framework that learns to adaptively compress encoder outputs by first partitioning them into variable-duration segments and then pooling each segment into a continuous audio token. We use group-relative policy optimization (GRPO) to train a segmenter with task performance feedback and a token-rate penalty. In automatic speech recognition (ASR) experiments, ASSET achieves performance comparable to a transcript-assisted character-boundary oracle with approximately 25% fewer audio tokens on English LibriSpeech, while improving on a phone-boundary oracle at a similar token rate on Mandarin AISHELL. Exploratory experiments on non-ASR tasks further illustrate how learned segmentation changes when the framework is adapted to tasks beyond transcription.
[NLP-334] he Thinking Leakage in Reinforcement Learning for LLM s
【速读】： 该论文旨在解决当前混合推理模型（如Qwen3系列）在仅使用“无思考”（NO-THINK）模式进行强化学习训练时所报告的推理性能提升是否真正源于模型生成了更优的推理行为，还是仅对原有“思考”（THINK）模式中的推理能力进行了部分重激活的问题。其核心发现表明，此类性能增益主要归因于“思考泄露”（thinking leakage）现象——即通过强化学习在NO-THINK模式下训练并未真正学习到新的推理机制，而是重新激活了原始模型中已存在的THINK模式推理能力。解决方案的关键在于揭示并验证这一现象：通过分析模型权重、激活值与注意力头的动态变化，研究发现强化学习更新会逐步引导模型状态向THINK模式逼近，且在高维空间中存在一个低维的THINK相关激活子空间，其特征值替换可显著削弱或转移强化学习带来的性能增益；此外，基于GRPO的训练信号显示，被优势加权优化的响应更接近THINK模式特征，其梯度更新方向也指向原始模型的THINK状态。因此，该研究指出，在NO-THINK模式下的强化学习性能提升本质上是原有推理能力的再激活，而非新推理行为的学习。
链接: https://arxiv.org/abs/2610.14270
作者: Wenxuan Zhu,Shuming Liu,Bang An,Aznaur Aliev,Bing Li,Yibo Yang,Bernard Ghanem
机构: King Abdullah University of Science and Technology(阿卜杜拉国王科技大学); Shanghai Jiao Tong University(上海交通大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Hybrid reasoning models such as the Qwen3 series switch between a THINK mode and a NO-THINK mode through a few chat-template tokens. A popular recipe trains such models with reinforcement learning in NO-THINK mode and reports improved reasoning performance. We provide evidence that much of this gain can be explained by thinking leakage: the gain comes largely from partially reactivating the reasoning abilities of the original model’s THINK mode instead of learning enhanced reasoning behavior. We train Qwen3-1.7B, 4B, and 8B with GRPO in NO-THINK mode. As accuracy improves, responses become longer and more reflective, are increasingly judged by the original model to resemble its own thinking, and the model’s internal activations shift toward its THINK mode. We then investigate at the level of weights, activations, and attention heads: scaling the RL update further elicits THINK-mode behavior and approaches THINK-mode accuracy but does not exceed it; in Qwen3-8B, replacing a low-dimensional THINK-related activation subspace of the RL-trained model with its original-model value removes most of the RL gain, and the reverse exchange transfers a considerable fraction of the gain to the original model; and the RL-induced shift passes through the attention heads that read the empty think block. We further identify a possible mechanism in the GRPO training signal: responses favored by the advantage-weighted update are more THINK-like, and a gradient step computed from this signal moves the model toward its THINK state. Thus, gains in NO-THINK mode can largely reflect reactivation of reasoning abilities already present in the original model.
[NLP-335] Agent ic Data Engineering for LLM s: System Design and a Controlled Empirical Study
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）数据工程中因依赖反复实验以确定数据筛选、奖励设计及训练课程对模型学习影响而带来的效率瓶颈问题，核心挑战在于如何实现数据工程策略的自动化优化。其解决方案的关键在于提出一种智能体驱动的数据工程框架（Agentic Data Engineering, ADE），使智能体能够基于实验反馈持续探索、修正或融合不同研究方向，并通过闭环评分机制动态选择最优策略。该框架利用累积研究成果指导策略演化，在固定训练与评估协议下，实现了在数学与代码数据选择、数学奖励设计及数学课程学习等四项任务上的显著性能提升：在AIME25 Pass@32指标上最高获得13.34%的绝对增益，四类基准测试平均提升达5.85%。研究进一步表明，闭环评分随实验迭代持续优化，但后期闭环表现的提升并不总是对应泛化性能的改善；相较而言，结合证据分析与累积发现的科研组织方式更有利于策略改进，且研究轨迹揭示了早期发现如何引导后续策略的修正与跨方向整合。
链接: https://arxiv.org/abs/2610.14238
作者: Ruomeng Ding,Qianli Shen,Zhaoyang Han,Meixin Chen,Daoyuan Chen,Yaliang Li
机构: University of North Carolina at Chapel Hill(北卡罗来纳大学教堂山分校); Alibaba Group(阿里巴巴集团); Peking University(北京大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:LLM data engineering requires iterative experimentation to determine how data selection, rewards, and curricula affect model learning, making it a natural setting for AI-for-AI (AI4AI) research. Can agents use experimental feedback to progressively improve strategies, and do the resulting strategies improve target-model performance on held-out data? We study how agents use experimental feedback to develop data-engineering strategies under fixed training and evaluation protocols. Our Agentic Data Engineering (ADE) framework lets agents pursue, revise, or combine research directions using cumulative findings. Strategies are selected by in-loop scores and evaluated on held-out benchmarks. Across four tasks-math and code data selection, math reward design, and math curriculum learning-selected strategies achieve absolute gains over generic baselines of up to 13.34% on AIME25 Pass@32 and 5.85% averaged across four held-out benchmarks under fixed training conditions. In-loop scores improve over successive experiments, while held-out trajectories show that later in-loop gains do not always improve generalization. Comparisons of research organization and feedback access favor combining evidence analysis with cumulative findings. Research trajectories show how earlier findings inform strategy revisions and combinations across research directions.
[NLP-336] Harness2: Recursive Agent Harnessing for an Open World
【速读】： 该论文旨在解决生成式 AI（Generative AI）代理在开放世界场景中进行递归自提升（Recursive Self-Improvement, RSI）时面临的两大核心挑战：其一，开放世界任务在代理设计阶段通常未被见过，测试时为新出现的任务，缺乏真实标签验证器（ground-truth verifiers）以可靠引导在线的代理执行框架（harness）修改；其二，现有RSI方法高度依赖特定任务，难以泛化至新场景。为此，论文提出Harness²——一种即插即用的递归代理框架，通过双层级递归机制实现突破：任务级递归采用对比式“提议-探测-组合”（Propose-Probe-Compose）策略，在无真实验证器的情况下仍能可靠指导当前任务的框架优化，并可通过并行或串行方式扩展；框架级递归则从历史优化经验中提炼知识，持续更新通用改进型框架，从而增强代理对未来任务的适应能力。实验表明，在三个专业基准和四种模型-框架配置下，单步优化即显著提升GLM-5与Gemini 3.5 Flash的表现，最高在WorkBuddy-Bench上获得11.5分提升，JobBench上达4.5分，LAB任务上提升11.8个百分点，且随着任务级递归预算增加，性能进一步增长。
链接: https://arxiv.org/abs/2610.14224
作者: Ruiyao Xu,Yanfei Chen,Zhongying CuiZhu,Bhavana Dalvi Mishra,Yifei Ming,Han Yu,Rujun Han,Chen-Yu Lee,Tomas Pfister
机构: Google Cloud AI Research(谷歌云人工智能研究); Northwestern University(西北大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:LLM agents’ capabilities depend not only on their underlying models but also on their execution harnesses. Recursive self-improvement (RSI) offers a promising approach to automatically evolve these harnesses. However, deploying RSI in open-world settings still faces two critical challenges. First, open-world tasks are typically unseen during the harness design phase and novel at test time, leaving no ground-truth verifiers to reliably guide online harness edits. Second, the RSI process remains highly task-specific and fails to generalize to novel scenarios. We introduce Harness^2, a plug-and-play, recursive agent harnessing framework that addresses these challenges through two connected levels of recursion. Task-level recursion refines the current task’s harness using a contrastive Propose-Probe-Compose step that reliably guides edits without ground-truth verifiers, and can be scaled further via parallel or sequential recursion. Harnessing-level recursion then extracts experience from these edits to update a continual improvement harness, enabling the agent to better adapt to future tasks. Across three professional benchmarks and four model-harness configurations, a single step consistently improves GLM-5 and Gemini 3.5 Flash over their base harnesses, yielding gains of up to 11.5 points on WorkBuddy-Bench, 4.5 points on JobBench, and 11.8 percentage points on LAB, with further gains as the task-level recursion budget increases.
[NLP-337] Learnable Low-Rank Polynomial Sketch for Effective Linear Attention NEURIPS2026
【速读】： 该论文旨在解决Transformer中Softmax注意力机制在序列长度增长时面临二次复杂度的问题，从而限制其在长上下文场景下的应用。现有线性注意力方法通过替换指数核函数以实现线性时间计算，其中多项式核因其类似Softmax注意力的稀疏尖峰特性（即突出大点积并抑制无关交互）而表现优异。然而，高维表示下高阶多项式核的精确计算不可行，现有方法依赖近似策略，引入了显著的近似误差。本文指出，当前线性注意力方法在多项式核近似方面仍存在不足。为此，作者提出一种可学习低秩多项式投影（Learnable Low-Rank Polynomial Sketch, LLoPS），该框架从理论上构建了一种更优的多项式核近似方法，通过学习一个低秩投影矩阵，严格优于已有近似方案的逼近误差。实验结果表明，LLoPS在多个基准测试中均达到最优性能，显著超越各类线性注意力基线。
链接: https://arxiv.org/abs/2610.14219
作者: Zongyue Qin,Haoran Deng,Jason Cong,Yizhou Sun
机构: University of California Los Angeles (加州大学洛杉矶分校)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Neurips 2026
Abstract:Softmax attention in Transformers suffers from quadratic complexity in sequence length, making it impractical for long-context applications. Linear attention alleviates this issue by replacing the exponential kernel with alternative functions that enable linear-time computation. Among existing linear attention approaches, recent studies have shown that polynomial kernels are particularly effective, as they exhibit sparse and spiky behavior similar to softmax attention, which emphasizes large dot products while suppressing irrelevant interactions. However, the exact computation of high-degree polynomial kernels is infeasible for high-dimensional representations. As a result, prior work relies on approximate polynomial kernels. This introduces a non-negligible approximation error. In this paper, we show that, from the perspective of polynomial kernel approximation, existing linear attention methods are still suboptimal. We propose \textbfLearnable \textbfLow-Rank \textbfPolynomial \textbfSketch (LLoPS), a principled and flexible framework for approximating polynomial kernels with linear attention. Our method learns a low-rank polynomial sketch that provably achieves a strictly smaller approximation error than existing approaches. Experiment results show that LLoPS achieves the highest performance across extensive benchmarks, comparing with various linear attention baselines.
[NLP-338] Modeling the Arrival Time of Information-Bearing Events in Natural Language
【速读】： 该论文旨在解决现有自然语言建模研究中对信息随时间分布机制理解不足的问题，尤其针对传统基于事后分析的信息论指标（如突兀度，surprisal）难以揭示生成过程内在动态机制的局限性。其核心解决方案是引入神经时间点过程（Neural Temporal Point Processes, TPPs），通过可学习的事件强度函数对文本序列中的关键信息事件进行建模，具体预测两类事件的未来发生时间：新词出现事件与高突兀度词元事件。模型以历史事件间隔和来自大语言模型（LLM）的上下文语义表示为输入，实现对生成行为的时间动态建模。实验发现，尽管人类与LLM在高突兀度词元事件的生成强度上无显著差异，但LLM产生新词的速率显著高于人类；同时，新词事件的强度在时间和人类及LLM中均呈现衰减趋势，而高突兀度词元事件则无此规律。这一发现不仅拓展了基于简单方法的先前研究，为人类语言生成中受限上下文表征的理论提供了补充证据，并从时间动态视角揭示了人类与LLM生成行为的本质差异。
链接: https://arxiv.org/abs/2610.14212
作者: Leyi Lao,Yulin Ou,Yang Xu,Hao An
机构: Southern University of Science and Technology (南方科技大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Understanding how information is distributed over time is fundamental to modeling natural language. While existing empirical studies primarily rely on post-hoc analysis of information-theoretic metrics, such as surprisal, little is known about the generation processes under the surface. We propose a novel approach for addressing this gap using the tool of neural temporal point processes (TPPs) by modeling the key information-bearing events in text sequences with learnable event intensity functions. Our model takes the history of event intervals and the contextual semantic representations from LLMs as input, and predicts the future arrival times for two types of events: new-word and high-surprisal-token. The experimental results reveal an interesting distinction in event intensities: LLMs tend to produce new words at significantly higher rates than humans, while high-surprisal tokens do not exhibit the difference. A decaying pattern of intensity over time is found in new-word events for both humans and LLMs, but not in high-surprisal-token events. Our findings revisit and extend previous studies that use simpler methods; provide supplementary evidence for theories about constrained context representation inherent in human language production; and find a new angle to view human and LLM generation differences.
[NLP-339] Unconstrained CTI-REALM with Open-Weights
【速读】： 该论文旨在解决安全领域中语言模型代理（language-model agents）在开源权重模型（open-weight models）与昂贵的闭源前沿模型（closed frontier models）之间的选择困境。其核心问题是：在威胁情报（Threat Intelligence, TI）任务中，能否通过成本更低的开源模型实现与闭源模型相当甚至更优的性能，尤其是在复杂推理和自动化检测生成场景下。解决方案的关键在于，通过一个名为CTI-REALM的基准测试框架，在统一评估环境（包括相同测试套件、评分器及大型语言模型（LLM）评审团）下对26个模型（19个开源、7个闭源）进行系统性对比。研究发现，当允许足够多的交互轮次（turns）时，开源模型可达到与最强闭源模型相当的性能，且成本不足其十分之一。此外，论文指出原基准设定的消息数限制（message limit）构成显著干扰——该限制导致任务耗尽限额即得零分，从而掩盖了部分模型的实际能力表现。当放宽此限制后，原本被截断的模型展现出真实优势，高奖励结果主要归属于开源模型。进一步分析表明，不同模型在任务流程的不同阶段表现出差异化优势，提示应采用“阶段路由”策略，即根据各阶段特性分配最优模型，而非依赖单一模型的整体得分。同时，研究发现模型输出中用于推理的令牌比例（thinking token share）与最终奖励无显著相关性，说明推理深度并不直接决定任务成效。
链接: https://arxiv.org/abs/2610.14180
作者: Sahil Garg,Ruhi Gottumukkala,Vishal Agarwal,Shubham Vishwakarma
机构: Averlon; Redmond, WA
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computers and Society (cs.CY); Software Engineering (cs.SE)
备注:
Abstract:Organisations that want language-model agents for security work face a choice between open-weight models, which they can run and audit themselves, and far more expensive closed frontier models. We study that choice on CTI-REALM (arXiv:2603.13517), a benchmark in which an agent turns a threat-intelligence report into a working detection by querying a live telemetry store. We evaluate twenty-six models, nineteen open-weight and seven closed, under a common harness, scorer and panel of LLM judges. Our main finding is that an open-weight model can match the strongest closed models at less than a tenth of their cost, provided it is given enough turns. The benchmark’s published limit on agent messages is a confound: a task that exhausts the limit scores zero on every checkpoint, including those it had already passed, so results at that limit partly measure which models finish in time. At the published limit the leading open-weight and closed models cannot be distinguished. With the limit relaxed, the models it was truncating recover, and the highest rewards belong to open-weight models. Among the leading models, price is a poor guide to capability: models with similar rewards differ in cost by more than an order of magnitude. Because the benchmark’s checkpoint scores largely record what an agent did or mentioned, we also introduce stage-level metrics that score what the agent submits, by executing its detection queries in the telemetry store and judging its outputs against the answer key. Rather than combining them into one more overall reward, we show that different models lead different stages, which suggests routing each stage of the workflow to the model that performs it best rather than choosing a single model by its overall score. We also report how many tokens each model generates, and find that the share of output spent on thinking does not track the reward.
[NLP-340] AutoEnvScaling: Automating the Data Flywheel with Terminal Agents
【速读】： 该论文旨在解决生成式智能体（Generative AI）在软件工程与科研领域中进行终端任务求解时，因强化学习（Reinforcement Learning, RL）训练所需多样化且可验证环境的供给严重受限的问题。当前环境设计高度依赖人工，导致数据生成效率低下，成为制约智能体能力提升的关键瓶颈。其解决方案的核心在于提出AutoEnvScaling框架，将环境设计本身视为一个终端任务，并通过“提议者代理”（proposer agent）与“求解者代理”（solver agent）的协同迭代机制，实现环境生成的自动化与闭环优化。具体而言，提议者代理基于求解者的历史执行轨迹（rollouts）自动构建新环境，而新环境又引导求解者产生更高质量的新轨迹，形成数据飞轮效应。该方法使有效环境生成率较仅依赖提示词（prompt-only）的方式提升达25倍，且单位有效环境成本降低75%。实验表明，基于此框架训练的Qwen3.5-35B-A3B模型在Terminal-Bench 2.1上性能提升4.3分；当同一模型以双角色（同时担任提议者与求解者）运行时，进一步在Terminal-Bench 2.1上取得12.2分的提升，并使Terminal-Bench 4.0得分从1.6跃升至3.2。该框架具备跨智能体应用领域的泛化能力，可生成适用于软件工程、计算机使用、专业工作及GPU内核编程等场景的环境，标志着向递归自我改进（Recursive Self-Improvement, RSI）迈出关键一步——即让智能体自主构建并持续优化自身的训练环境，显著减少对人类干预的依赖。
链接: https://arxiv.org/abs/2610.14163
作者: Simon Yu,Baolin Peng,Bo Liu,Zhengyan Shi,Zhaoyang Wang,Hao Cheng,Qianhui Wu,Wenlin Yao,Natasha Jaques,Weiyan Shi,Jianfeng Gao
机构: Northeastern University(东北大学); Microsoft(微软); Stanford University(斯坦福大学); University of Washington(华盛顿大学); University of North Carolina at Chapel Hill(北卡罗来纳大学教堂山分校)
类目: oftware Engineering (cs.SE); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 44 pages, 24 figures, including references and appendices. Project page: this https URL ; Code: this https URL
Abstract:Terminal agents solve tasks in software engineering and scientific research by executing commands in a terminal. Training them via reinforcement learning (RL) requires diverse and verifiable environments, but these environments are still mostly designed by human researchers, so their supply grows only as fast as researchers can work. We introduce AutoEnvScaling, which treats environment design as a terminal task and uses agents to automate the data flywheel. A proposer agent builds new RL environments from the rollouts of the solver being trained, and the solver’s new rollouts guide the next round. By turning environment generation into a terminal task, AutoEnvScaling increases the valid-environment rate by up to 25 times compared to prompt-only generation, at up to 75% lower cost per valid environment. We show that training Qwen3.5-35B-A3B on AutoEnvScaling environments from GPT-5.6-sol improves its Terminal-Bench 2.1 score by 4.3 points over the matched Tmax baseline. The broader goal of AutoEnvScaling is to enable recursive self-improvement (RSI): by making both environment generation and solving terminal tasks, it lets one model co-evolve in both roles. We show that Qwen3.6-35B-A3B, acting in both roles, gains 12.2 points on Terminal-Bench 2.1 and raises its Terminal-Bench 4.0 score from 1.6 to 3.2. The pipeline also generalizes across agentic domains, generating environments for software engineering, computer use, professional work, and GPU kernel programming. Overall, these results show a path toward agents building their own training environments, reducing human reliance and taking a core step toward RSI.
[NLP-341] Next-Concept Modeling: Fast Parallel Coarse-to-Fine Text Generation
【速读】： 该论文旨在解决生成式语言模型在推理效率与生成质量之间的权衡问题，特别是针对扩散语言模型（Diffusion Language Models, DLMs）存在的采样步数少时性能急剧下降、每一步均需处理完整序列以及需要多次遍历数据集才能达到高质量等关键瓶颈。其解决方案的核心是提出一种两阶段、自粗到细的“下一概念建模”（Next-Concept Modeling, NCM）框架：首先通过多尺度残差向量量化变分自编码器（multiscale residual VQ-VAE）将文本编码为层次化的离散代码，随后利用Transformer模型逐级并行生成每一层级内的所有代码。该设计实现了高效的并行化生成，显著降低推理延迟。在参数量相当的情况下，NCM在LM1B上优于最强的DLM基线且推理速度提升21倍；在OpenWebText上与最强DLM性能相当但速度提升10倍；在XSum任务中达到与最强DLM相当的ROUGE-L指标，同时速度提升31倍；在数独推理任务中，无论难度如何，均优于自回归和扩散基线模型，且所需函数评估次数更少。
链接: https://arxiv.org/abs/2610.14162
作者: Hermann Kumbong,Ishan S. Khare,Iddah Mlauzi,Andrew Woen,Charles Ding,E. Kelly Buchanan,Simran Arora,Azalia Mirhoseini,Christopher Ré
机构: Stanford University (斯坦福大学); Together AI (Together AI)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Diffusion language models (DLMs) offer parallel sampling for faster generation than autoregressive models, an increasingly important advantage as inference becomes a dominant computational workload. However, DLMs degrade quickly with fewer sampling steps, process the full sequence at every step, and require many passes over the dataset to reach good quality. We introduce Next-Concept Modeling (NCM), a two-stage, coarse-to-fine framework: a multiscale residual VQ-VAE encodes text into a hierarchy of discrete codes, and a transformer generates one level at a time, predicting all codes within a level in parallel. At matched parameter scales, NCM achieves better generative perplexity than the strongest DLM baseline on LM1B while running 21\times faster. On OpenWebText, NCM is competitive with the strongest DLM while running 10\times faster; on XSum, it matches the strongest DLM’s ROUGE-L while being 31\times faster. On the reasoning task of Sudoku, NCM outperforms autoregressive and diffusion baselines across all difficulty levels, using fewer function evaluations. On the reasoning task of Sudoku, NCM outperforms autoregressive and diffusion baselines across all difficulty levels, using fewer function evaluations.
[NLP-342] OpenMobile-2: Building Versatile Mobile Agents with Scalable Environments and App-Native Tools
【速读】： 该论文旨在解决移动智能体（mobile agent）在训练环境与数据开放性、任务复杂性及真实应用泛化能力方面的关键瓶颈问题。现有研究受限于封闭的训练环境、缺乏大规模可验证的交互数据，以及对真实商业应用中多模态操作（如图形用户界面GUI与原生工具调用结合）的支持不足，导致智能体难以在复杂、长周期的真实移动场景中有效泛化。为此，论文提出OpenMobile-2，其核心解决方案在于三方面突破：一是构建了包含35个功能丰富的模拟商业化应用（commercial-style apps）的开放环境MobileGym++，支持跨应用工作流与完全可控的重置与验证机制，扩展至超110个应用；二是建立了大规模开放的训练数据集，涵盖近1.2万条监督微调轨迹，以及超过2000个可自动奖励评估的可执行强化学习任务，形成当前最大的开源可验证移动智能体训练数据集；三是首次系统探索混合式人机交互范式，即在保持原有GUI界面的同时，将部分应用功能以原生工具（app-native tools）形式暴露，使智能体可在任务中灵活穿插使用GUI操作与工具调用，实现在50多个应用中超过300个精细化工具的集成。为评估该范式，论文进一步推出MobileGym++ Bench，支持统一任务集下的GUI-only与混合GUI-tool双模式评估。实验表明，OpenMobile-2在AndroidWorld（79.9）和MobileWorld（50.4）等基准上表现优异，并在真实设备测试中实现SPA-Bench性能近乎翻倍提升（从31.9增至59.6），充分验证了其在复杂长时序商业场景中的强泛化能力与实用性。
链接: https://arxiv.org/abs/2610.14157
作者: Chenyang Yan,Yingying Zhang,Kanzhi Cheng,Qiushi Sun,Zhengyuan Pan,Zheng Ma,Hang Yan,Yian Wang,Nuo Chen,Jialin Cao,Xingdong Gong,Wenpo Song,Han Chen,Zichen Ding,Fangzhi Xu,Shujian Huang,Xinyu Dai,Yichen Liu,Tiankuo Yao,Bo Wang,Ben Kao,Jianbing Zhang,Lewei Lu,Dahua Lin
机构: Nanjing University(南京大学); Xi’an Jiaotong University(西安交通大学); National University of Singapore(新加坡国立大学); SenseTime(商汤科技); East China Normal University(华东师范大学); The University of Hong Kong(香港大学)
类目: oftware Engineering (cs.SE); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We introduce OpenMobile-2, a near-frontier mobile agent with fully open training environments and recipes. We make three key advances: (1) Diverse environments with simulated commercial apps: We build MobileGym++, featuring 35 realistic, functionally rich commercial-style apps with cross-app workflows, while preserving full controllability for reset and verification. Together with newly configured apps in Android emulators, this yields a diverse playground spanning over 110 apps. (2) Open training data at scale: Building on this foundation, we curate nearly 12K mobile interaction trajectories for supervised fine-tuning and the largest open collection of verifiable RL training data for mobile agents, comprising over 2K executable tasks with automatic rewards. (3) Hybrid GUI and app-native tool use: We explore an experimental mobile-use setting where apps expose selected functionalities as app-native tools alongside their GUIs, allowing agents to interleave GUI actions and tool calls within a task. We implement this setting in MobileGym++ with over 300 carefully scoped tools across 50+ apps. Additionally, we introduce MobileGym++ Bench for complex, long-horizon commercial mobile scenarios, supporting both GUI-only and hybrid GUI-tool evaluation on a shared task suite. OpenMobile-2 performs competitively across established benchmarks, including AndroidWorld (79.9) and MobileWorld (50.4), while showing promising transfer to real-device mobile use, nearly doubling SPA-Bench performance from 31.9 to 59.6. All resources are available at this https URL.
[NLP-343] ClawLoop: Self-Improving Agents via Harness-Native Training Loop
【速读】： 该论文旨在解决当前基于沙盒环境（harness environment）的生成式 AI 代理在自我提升过程中存在的关键瓶颈问题：现有训练流程高度依赖外部强教师模型（teacher model）进行任务生成与轨迹收集，且仅支持单次训练、缺乏闭环反馈机制，导致其能力提升受限于教师模型性能与人工先验范围，无法实现自主进化。针对这一问题，论文提出 ClawLoop 框架，其核心创新在于构建了一个无需外部教师模型、低依赖先验的自洽闭环系统。该框架由三个相辅相成的阶段构成：任务合成（Task Synthesis）通过双路径构造策略生成多样化训练任务，并结合诊断驱动的难度递增机制及代理无关/代理依赖的任务验证；训练（Training）阶段采用监督微调（SFT）和/或强化学习（RL）更新代理；诊断与反馈（Diagnosis and Feedback）阶段则通过对代理运行行为的分析生成能力评估报告，作为下一迭代中任务合成的输入。实验证明，以 Qwen3.6-27B 为循环主体模型时，ClawLoop 在 SkillsBench、Claw-Eval-Live 与 ClawGym-Bench 上均实现持续性能提升；且经迭代优化后的高质量任务数据可有效迁移至更小规模的 Qwen3.5-4B 与 9B 模型，进一步验证了任务质量独立于循环主体模型的有效性。
链接: https://arxiv.org/abs/2610.14118
作者: Hansi Yang,Linji Hao,Hongyan Hao,An Wang,Han Wu,Qi Gu,Xunliang Cai
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:The rapid advancement of Large Language Models (LLMs) has driven their deployment toward general-purpose harness environments, where LLM-based agents operate in sandbox workspaces and interact directly with file systems and shells. While numerous post-training pipelines exist for tool-use agents, their counterparts for harness-based agents remain scarce and are predominantly one-shot and teacher-dependent: a stronger external model synthesizes tasks and collects trajectories, the target agent is trained once, and evaluation results are consumed by human researchers rather than fed back into the loop. Such pipelines are inherently bounded by the capability of the external model and the scope of human prior, leaving no room for self-improvement. To address these limitations, we propose ClawLoop, a framework that enables self-improving without any external teacher model and with minimal prior dependencies. ClawLoop composes three phases into a self-sustaining cycle: Task Synthesis, which generates a diverse pool of training tasks via two complementary construction routines, applies diagnosis-conditioned difficulty growth followed by both agent-free and agent-based task validation; Training, which updates the agent via SFT and/or RL; and Diagnosis and Feedback, which analyzes the agent’s rollout behaviors to produce a capability analysis report that will be used as feedback for task synthesis in the next iteration. Using Qwen3.6-27B as the loop model, ClawLoop yields consistent improvements on SkillsBench, Claw-Eval-Live, and ClawGym-Bench. The iteratively refined tasks are further transferred to smaller-scale Qwen3.5-4B and 9B models with gains across all benchmarks, validating data quality independently of the loop model.
[NLP-344] Does System One Transfer? Calibrated Typed-Decision Models Label Budgets and Translation for Indonesian Regional Languages
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在非英语语境下，特别是低资源语言（如印尼地方语言）中的评估不足问题。当前大多数模型评估局限于英语，导致对多语言能力的判断存在偏差，尤其在跨语言迁移和零样本推理场景下的可靠性难以验证。其核心解决方案在于通过预注册的跨语言基准测试，对商业模型Jev与开源模型Laya-multilingual在印尼地区语言上的表现进行系统性评估，涵盖情感识别（NusaX、NusaTranslation）等任务。关键发现包括：1）模型在词汇上更接近印尼语的句子中表现更优，且使用人工印尼语翻译可显著提升准确率（最高达37.2个百分点），尤其在远离标准印尼语的语言中效果更明显；2）尽管模型的期望校准误差（Expected Calibration Error）未随语言多样性增加而恶化，但其自我错误排序能力下降；3）仅靠微调训练需超过500个标注样本才能在多数语言上达到零样本Jev的表现；4）提出一种置信度门控级联架构，将低置信度预测（10%-30%）路由至本地IndoBERT分类器，性能可超越或媲美任一主模型，若进一步路由至大语言模型（LLM），则可在24%-77%更低的API成本下实现与原生LLM相当的性能。研究还揭示多数商用大模型虽精度更高（比Jev高4-14分），但成本高达数百倍，凸显高效轻量模型在低成本部署中的潜力。
链接: https://arxiv.org/abs/2610.14112
作者: Romi Fadli,Rusdah,Wahyu Ramadani Putri Utami,Zulfa Malika Amana
机构: 未知
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 14 pages, 5 figures, 8 tables. Pre-registered study; code, pre-registration and all predictions: this https URL
Abstract:Calibrated typed-decision (“System One”) models return a label and a probability in a single forward pass at a small fraction of the cost of large language models (LLMs), yet their evaluation has been almost entirely English. We report a pre-registered evaluation of the commercial Jev model and the open Laya-multilingual model on Indonesian regional languages, using the parallel NusaX sentiment (12 languages) and NusaTranslation sentiment and emotion (7 languages) test sets. Jev is more often correct on sentences that are lexically closer to Indonesian in all three datasets, and presenting the human Indonesian translation instead raises its accuracy by up to 37.2 points, most in languages distant from Indonesian. Its expected calibration error does not grow on regional languages, but its ability to rank its own errors degrades in all three datasets. Retrained classifiers need more than 500 labels per language to match zero-shot Jev in 9 of 10 NusaX regional languages. A confidence-gated cascade that sends Jev’s 10-30% least confident decisions to a local IndoBERT classifier matches or exceeds the better of the two systems, and routing to an LLM instead matches that LLM at 24-77% lower API cost. Of six further commercial LLMs (three Claude and three GPT models), five are 4-14 points more accurate than Jev, at several hundred times the API cost where it is reported, while the smallest (Claude Haiku) is less accurate. We release the pre-registration, code and all predictions.
[NLP-345] Fin_Agent at FinEval 2026: Task-Aware Methods for Multilingual Financial NLP EMNLP
【速读】： 该论文针对多语言金融语言理解任务中的三大核心挑战：希腊语金融命名实体识别（Greek Financial Named-Entity Recognition, NER）、日语隐含交际识别（Japanese Implicit Communication Recognition）以及多语言金融问答（Multilingual Financial Question Answering），提出了针对性的解决方案。其关键在于：针对希腊语NER，采用注释感知的数值规则与零样本Qwen2.5-32B分类模型相结合的方法，有效提升了实体识别的准确性；针对日语隐含交际识别，设计了一种可解释的基于线索（cue-based）的序数评分器，实现了98.0%的测试准确率；在多语言金融问答任务中，结合BGE-M3检索与Qwen3-32B生成模型，显著提升了答案生成质量。整体方案的核心优势在于融合了规则驱动与大模型能力，并在不同语言和任务间实现高效适配，但其局限性在于依赖数据特定规则、手工设计线索及监督信号有限。
链接: https://arxiv.org/abs/2610.14101
作者: Mohotarema Rashid,Tanzina Hossain,Xiaoying Song,Lingzi Hong
机构: University of North Texas (北德克萨斯大学)
类目: Computation and Language (cs.CL)
备注: 5 pages, 1 Figure, FinEval 2026: Multilingual Financial Language Understanding, Co-located with EMNLP WORKSHOP, OCT 28 2026, BUDAPEST, HUNGARY
Abstract:We present Fin_Agent, our submission to the FinEval 2026 shared task on multilingual fi nancial language understanding. Our sys tems address Greek financial named-entity recognition, Japanese implicit communication recognition, and multilingual financial ques tion answering. For Greek NER, we com bine annotation-aware numeric rules with zero shot Qwen2.5-32B classification, achieving 0.9521 public F1 and 0.8884 private F1. For Japanese implicit communication, an inter pretable cue-based ordinal scorer achieves 98.0% test accuracy. For multilingual financial QA, BGE-M3retrieval with Qwen3-32B gen eration achieves 0.3023 and 0.3419 ROUGE-1 F1 on the public and private splits. Key limita tions include dataset-specific rules, handcrafted cues, and limited supervision. Implementa tion code is available this https URL Rashidnira/finagent-fineval-2026.
[NLP-346] Is the Future Predictable? Modeling Future Predictability for Reliable Temporal Reasoning EMNLP2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在对未来进行推理时的不可靠性问题，尤其关注模型在面对未来状态具有内在不确定性或波动性时，仍倾向于做出绝对化判断的现象。其核心挑战在于现有模型缺乏对未来的可预测性进行评估的能力，导致在证据不充分或未来状态高度不确定的情况下仍强行生成确定性答案，从而引发幻觉（hallucination）。为应对这一问题，论文提出了一种名为ChronoRegime的面向未来的时间推理任务，用以衡量历史时间证据是否足以支持对特定未来状态的承诺。关键解决方案是提出一种自信感知的时间推理框架——置信度感知时间推理（Confidence-Aware Temporal Reasoning, CTR），该框架通过分析时间动态来估计未来状态的可预测性，并据此调节答案的承诺行为：仅当重构的历史证据足够支持目标未来状态时，才进行承诺。实验表明，相较于传统的问答基线，CTR在多个大语言模型上平均提升了6.7个百分点的严格准确率，并将幻觉率降低18.4个百分点，验证了显式预测性评估在提升未来不确定性下时间推理的安全性与可靠性方面的有效性。
链接: https://arxiv.org/abs/2610.14073
作者: Yongyi Huang,Huaping Zhang,Wenyao Cui,Qiuchi Li,Lei Li,Baohua Zhang
机构: Beijing Institute of Technology (北京理工大学); Zhongguancun Academy (中关村研究院)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to EMNLP 2026 (Main Conference)
Abstract:Large language models (LLMs) remain unreliable when reasoning about the future. Existing models often overlook the difference among future states, committing to specific answers even when the queried facts are inherently volatile. To capture this challenge, we introduce ChronoRegime, a future-oriented temporal reasoning task that requires models to determine whether historical temporal evidence justifies commitment to a future answer. Furthermore, we propose Confidence-Aware Temporal Reasoning (CTR), a framework that estimates future predictability from temporal dynamics and regulates answer commitment accordingly. Rather than performing unconditional forecasting, CTR commits only when the reconstructed evidence supports the target future state. Across multiple LLMs, CTR improves strict overall accuracy by 6.7 percentage points and reduces hallucination rates by 18.4 percentage points on average over the plain QA baseline. These results demonstrate that explicit predictability assessment is an effective mechanism for safer and more reliable temporal reasoning under future uncertainty.
[NLP-347] Loose Lips: Understanding Privacy Leakage in Agent ic Dialogue
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）代理在高风险企业场景中处理敏感信息时存在的隐私泄露问题，核心关注点在于理解模型在代理对话过程中是否遵循信息流动的语境完整性（Contextual Integrity）。其解决方案的关键在于构建了一个基于语境完整性理论的基准测试集——CIMEMORIES DIALOGUE，包含4,410组双代理对话实例，用于系统评估不同模型在多种提示策略和对话长度下的隐私泄露行为。研究发现，任务能力与隐私泄露之间存在根本性耦合：越具能力的模型表现出更高的隐私泄露风险，且无一模型能实现二者解耦；采用思维链（Chain-of-Thought）推理的“思考型”模型相比非思考型模型泄露更严重，这归因于其扩展的推理表征空间；同时，隐私泄露随对话轮次增加而显著加剧，6轮对话下的泄露量可达1轮的3倍。这些结果表明，任务能力与隐私泄露的耦合并非配置失误所致，而是当前代理系统内在的结构性特征。
链接: https://arxiv.org/abs/2610.14071
作者: Soumya Sharma,Ayana Hussain,Heber Hwang Arcolezi,Ulrich Aïvodji,Spandana Gella,Adriana Romero-Soriano,Golnoosh Farnadi
机构: Mila – Quebec AI Institute(蒙特利尔人工智能研究所); McGill University (麦吉尔大学); Simon Fraser University (西蒙菲莎大学); ÉTS (École de technologie supérieure); ServiceNow AI Research; Université de Montréal (蒙特利尔大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Spotlight @ HAIPS workshop @ COLM 2026
Abstract:As LLM agents are increasingly deployed in high-stakes enterprise settings, understanding how they handle sensitive information in agentic dialogue has become critical. Privacy leakage in such settings requires reasoning about whether information flows respect the contextual norms under which they originated. We extend CIMemories to CIMEMORIES DIALOGUE, a benchmark of 4,410 two-agent dialogue instances grounded in Contextual Integrity theory. Using this benchmark, we systematically evaluate 7 open-weight models from the Qwen3 and Llama 3.1 families across conversation lengths of 1 and 6 turns under three prompting conditions. We find that task capability and privacy leakage are fundamentally coupled: more capable models are systematically leakier, with no model in our evaluation able to decouple the two. Thinking models consistently leak more than their non-thinking counterparts, a pattern we attribute to the extended reasoning surface created by chain-of-thought. Leakage compounds with conversation length, rising up to 3x from 1 to 6 turns. Our results suggest that the coupling between task capability and privacy leakage is not an artifact of misconfiguration but a structural feature of current agentic systems.
[NLP-348] Contrastive Covariance Subspaces for Activation Steering
【速读】： 该论文旨在解决生成式 AI（Generative AI）中激活引导（activation steering）技术在推理阶段因引导向量中混入无关共变因素（如附带主题或写作风格）而导致生成内容质量下降的问题。其核心挑战在于，现有方法所使用的引导方向往往与目标概念存在纠缠的干扰信号，导致这些干扰被注入残差流中，从而削弱了概念控制的纯净性与有效性。本文提出的解决方案——对比协方差引导（Contrastive Covariance Steering, CCS），是一种无需训练的方法，其关键在于通过解析求解一个对比方差目标，将引导向量投影到一个低维子空间，该子空间仅保留目标概念内部变化的方向，同时抑制与其他类别共享的共变方向。实验表明，在三种大语言模型（LLM）和四个引导领域中，CCS在全部12个设置下均显著优于所有基线方法，尤其在AxBench和情感引导任务上相对提升达48%至70%和13%至27%，且相比朴素均值差引导方法性能提升高达151%，充分验证了去混淆的重要性。消融实验与人类评估进一步证实了该方法的有效性与鲁棒性，并拓展至多轮对话场景，展现出在保持概念表达强度与文本流畅性方面的优越性能，显著优于其他引导方法随对话轮次推移出现的性能退化现象。
链接: https://arxiv.org/abs/2610.14025
作者: Alexandros Xenos,Manos Plitsis,Niki Foteinopoulou,Yannis Panagakis,Ioannis Patras
机构: Tavus.io; University of Athens (雅典大学); Queen Mary University of London
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Activation steering controls large language model behavior by adding a direction to hidden activations during inference. For steering to work well, that direction must be a pure signal for the target concept: any confound entangled with it, such as an incidental topic or writing style, is injected into the residual stream alongside the concept itself and degrades the generated text. We introduce Contrastive Covariance Steering (CCS), a training-free method that eliminates this confound variation. CCS projects steering vectors to a low-dimensional subspace obtained in closed form by solving a contrastive variance objective, which retains the directions along which the target concept varies while suppressing directions shared with the other classes. Across three LLMs and four steering domains, CCS consistently beats all baseline steering methods in all 12 model-domain settings. The largest relative improvement occurs on AxBench and Emotion where we outperform the next strongest steering baseline by 48 to 70% and 13 to 27% , respectively. Furthermore, CCS outperforms naive difference-of-means steering by up to 151% , showcasing the importance of deconfounding concepts. A toy experiment verifies our motivation on why: planting a topic nuisance into the training data reduces unprojected steering by 46% , whereas CCS loses only 12% and recovers most of the gap to an oracle that explicitly deletes the planted direction. A human study with 23 raters on 296 generations reproduces the LLM judge’s ranking. Our extensive ablation verifies the importance of our method’s components and robustness to the number of nuisance classes. Finally, we extend our method to multi-turn conversations, where CCS sustains both concept expression and fluency, matching system prompt performance as other steering methods drift.
[NLP-349] Artificialitys First Exam: An LLM -Referenced C-test for Supraintelligence
【速读】： 该论文旨在解决当前人工智能（AI）评估基准在技术快速迭代背景下面临的“自指性饱和”问题：即以人类或现有AI系统能力为基准的评测体系，随着被测系统能力提升而迅速失效。其核心挑战在于如何构建可扩展、具备理论基础且能持续衡量超越当前智能水平机器的能力评估框架。解决方案的关键在于引入基于莱文复杂度（Levin’s Kt）的逆问题评测范式——该方法将问题难度定义为对潜在机制的最简表示编码长度与运行该机制所需时间步数的对数之和，从而实现对认知复杂性的可计算量化。本文首次将大型语言模型（LLM）作为表征机器（representational machine）应用于该框架，利用其作为通用图灵机（UTM）所蕴含的人类文化知识库和自然语言指令执行能力，构建了一套轻量级的测试生成与认证流水线。通过设定参考机器的性能差异与评测者-被测者之间的差距，实现了高难度测试项的动态生成，其难度以比特为单位进行比率尺度测量，远超当前前沿模型及人类的解题能力。实验生成了首个经认证的测试集，并对12个主流大模型进行了评估，结果呈现出随难度递增而下降的代理特征曲线，揭示出一个尚未被探索的“超智能窗口”（supra-intelligence window），即理论上可认证但超出当前所有智能体生成能力的高难度区域。
链接: https://arxiv.org/abs/2610.14005
作者: Jose Hernandez-Orallo
机构: Leverhulme Centre for the Future of Intelligence; University of Cambridge; Valencian Research Institute for Artificial Intelligence; Universitat Politècnica de València
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Preprint, under review. 63 pages. Code and data at this https URL
Abstract:Under the accelerated pace of AI development, any benchmark whose difficulty is calibrated with humans or current AI systems becomes circular, or saturates soon. How can we then evaluate machines that may become more capable than the current machines and the people who write the tests? One answer to the question hinges on inverse problems, inductive inference questions for which difficulty depends on the complexity of the underlying mechanism, given the observations and the previous knowledge. Levin’s Kt, a computable variant of Kolmogorov complexity, puts a theoretical upper bound to that difficulty, balancing the simplest representation encoding for the mechanism and the logarithm of the time steps to run that mechanism. Previous tests derived from Levin’s Kt, such as the original C-test, are based on minimalistic Universal Turing Machines (UTM), capable of very limited representations and no previous knowledge. Reasoning large language models (LLM) are UTMs that capture a compendium of human culture as knowledge, using prompts as natural language programs. This paper presents the first test based on Levin’s Kt with LLMs as representational machines. Fresh test items can be produced at high difficulties, estimated in a ratio scale measured in bits, before any human or frontier model can solve them. We formalise the measure, devise a lightweight generation and certification pipeline, explore the effect of the power of the reference machine and the gap between examiner and examinee. We produce a first battery of certified items administered to twelve LLMs. The resulting agent characteristic curves decrease with difficulty, down to the uncharted high-difficulty region, the supra-intelligence window of items we can certify, yet beyond generations of frontier intelligence.
[NLP-350] SemTree: Judge-Aware Semantic Decision Trees for Inspectable and Repairable Rule Decisions
【速读】： 该论文旨在解决规则决策（rule decision）中因多条件组合、例外情形及信息缺失导致的错误定位困难问题，尤其在内容审核、广告合规性检查和资格审查等场景中，传统方法难以精准追溯并修复错误根源。其核心解决方案是提出SemTree框架，将规则决策建模为语义决策树（semantic decision tree）：每个节点以自然语言形式提出一个具有明确“是/否/未知”判定标准的问题，由可插拔的语义判断器（semantic judge）进行回答，分支结构依据判断器的实际输出动态编译。这一设计使得每条执行路径即为决策推理过程，实现错误在发生处即时修复——通过人工修正单个节点后重放规则，被修正的节点作为训练标签反馈给判断器，同时引入具备工具使用能力的大语言模型（LLM）编辑器，在观察执行结果的基础上自动优化问题表述与子树结构。实验表明，在8,276个ShARC测试用例上，由托管判断器执行的语义决策树达到67.70%准确率，平均每个案例仅需2.99个问题，接近直接询问同一判断器最终标签的68.23%准确率；而基于其他判断器编译的初始树在新判断器上仅达47.87%准确率，但通过12个初始问题池构建的新树已提升至64.39%。此外，节点级修复使BANKING77数据集准确率从82.04%提升至86.16%，显式例外节点将中文电商内容审核条款的误报率从24降至5，且经训练的编辑器使编辑后树的开发准确率从64.47%提升至73.35%。关键创新在于将规则决策过程显式化为可交互、可修正、可学习的语义树结构，并融合人工反馈与自动化编辑，实现高效、可解释且持续优化的规则决策系统。
链接: https://arxiv.org/abs/2610.13999
作者: Jincheng Gao,Wei Ding,Shuo Sun
机构: The Hong Kong University of Science and Technology(Guangzhou)(香港科技大学(广州)); Xiamen University(厦门大学)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Rule decisions such as content moderation, advertising compliance, and eligibility checks combine several conditions, exemptions, and missing-information cases, and an error can be fixed only once the failing condition is known. We introduce SemTree, a framework that represents a rule decision as a semantic decision tree: each node is a natural-language question with explicit Yes/No/Unknown criteria, a pluggable semantic judge answers it, and the branching structure is compiled on the answers that this judge actually gives. The executed path is the reasoning behind every decision, so errors are repaired where they occur: a reviewer corrects a single node and replays the rule, corrected nodes become training labels for the judge, and a tool-using LLM editor revises questions and subtrees while observing execution results. On all 8,276 ShARC test cases, a tree executed by a hosted judge reaches 67.70% accuracy with 2.99 questions per case, on par with asking the same judge for the final label (68.23%); an earlier tree compiled for a different judge scores 47.87% when executed by this judge, while a new tree compiled on this judge’s answers to a 12-question initial pool already scores 64.39%. Node-level repair raises BANKING77 accuracy from 82.04% to 86.16%, an explicit exemption node cuts the false positives of a Chinese e-commerce moderation clause from 24 to 5, and a trained editor raises the development accuracy of edited trees from 64.47% to 73.35%.
[NLP-351] GameSmith: Scaling Zero-Sum Self-Play with Synthesised Games
【速读】： 该论文旨在解决现有大语言模型（LLM）后训练方法在双人零和博弈场景中因依赖少量手工设计游戏及与特定游戏绑定的优势估计算法而导致的可扩展性瓶颈问题。其核心挑战在于：游戏供给有限、优势估计难以泛化至大规模变体，限制了模型推理能力的提升与迁移效果。本文提出GameSmith框架，通过将游戏生成过程形式化为模板化结构，在固定奖励、观测与隐藏信息代码的前提下，由大语言模型自动生成多样化博弈实例；所有生成的游戏均通过种子配对回放验证与独立规范重实现双重校验，并进一步衍生出编码、规则与规模等变体，共获得26个新游戏、29个游戏家族及336个经验证的变体。随后，采用基于教师模型（Qwen3-32B）自对弈数据的蒸馏策略，将知识迁移到目标模型（Qwen3-4B-Base），并引入CRAB——一种不依赖特定游戏结构、基于位置期望回报收缩估计来评估每步决策价值的优势估计算法，适用于所有变体。实验表明，该方案显著提升了模型在13项基准测试中的平均通过率（pass@1）从45.5提升至64.9，且在不同模型上均保持性能排序一致性，同时迁移能力随游戏家族数量增加而增强。因此，解决方案的关键在于：通过自动化生成高多样性、可验证的博弈环境并结合通用优势估计机制，实现了训练数据与算法的高效可扩展性。
链接: https://arxiv.org/abs/2610.13995
作者: Weixian Waylon Li,Jiaxin Zhang,Han Zhou,Chengzu Li,Ke Wang,Fengxiang He,Yiwen Guo
机构: University of Edinburgh(爱丁堡大学); LIGHTSPEED; University of Cambridge(剑桥大学); Independent Researcher(独立研究员)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Post-training large language models (LLMs) on two-player zero-sum games provides a verifiable reward, and the gains have been shown to transfer to general reasoning. Existing pipelines, however, rely on a handful of hand-written games and on advantage estimators tied to them, so both the supply of games and the advantage estimate on a large corpus limit scaling. We present GameSmith, which scales both. An LLM synthesises games inside templates whose reward, observation, and hidden-information code is fixed. Each game is verified by seed-paired replay and by an independent re-implementation of its specification, then expanded into encoding, rule, and size variants, giving 26 new games in 29 families and 336 verified variants from the three SPIRAL templates. The model is distilled from the self-play of a larger teacher (Qwen3-32B) on these games and then trained by self-play reinforcement learning with CRAB, an advantage estimator that credits each decision against a shrinkage estimate of the expected return of its position and is defined on every variant. On Qwen3-4B-Base, distillation raises the 13-benchmark mean pass@1 from 45.5 to 62.5, above the three SPIRAL games at the same budget, and self-play with CRAB reaches 64.9. The ordering holds across models, and transfer grows with the number of game families.
[NLP-352] Mira-Embeddings-V1: Domain-Adapted Semantic Reranking for Recruitment via LLM -Synthesized Data
【速读】： 该论文旨在解决招聘场景中候选人才检索的精准排序问题，核心挑战在于：在有限的人工审核预算下，如何从上游召回器生成的候选名单中高效重排，使合格候选人尽可能前置。其解决方案的关键在于提出了一种名为mira-embeddings-v1的语义重排序系统，通过大语言模型（LLM）合成训练数据重构嵌入空间，并引入轻量级边界感知重排序头（BoundaryHead MLP）以区分同职位名称但职责范围不同的角色。具体而言，研究构建了一个五阶段提示工程流水线，基于真实职位描述（JD）生成多样化的正样本与难负样本，从而多角度优化语义空间；随后采用两轮低秩适配（LoRA）策略，先进行JD间对比学习，再在异构文本数据集上进行JD-CV三元组对齐；整个过程仅依赖少量真实JD并借助LLM扩增监督信号，无需大规模人工标注的工业级训练对。最终，在本地300个真实JD的测试集上，Recall@50提升至77.55%（基线68.89%），Precision@10由35.77%增至39.62%；在全球超4万候选人的评估池中，Recall@200达到0.7047，显著优于基线的0.5969。结果表明，结合LLM合成监督与边界感知重排序的方案，可在不依赖重型交叉编码器的前提下实现稳健性能提升。
链接: https://arxiv.org/abs/2604.17738
作者: Zhaohua Liang,Zhilin Wang,Renjie Cao,Yining Zhang
机构: OpenJobs AI(OpenJobs AI)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Candidate sourcing for recruiters is best viewed as a two-stage retrieval and reranking pipeline with recall as the primary objective under a limited review budget. An upstream production retriever first returns a candidate shortlist for each job description (JD), and our goal is to rerank that shortlist so that qualified candidates appear as high as possible. We present mira-embeddings-v1, a semantic reranking system for the recruitment domain that reshapes the embedding space with LLM-synthesized training data and corrects boundary confusions with a lightweight reranking head. Starting from real JDs, we build a five-stage prompt pipeline to generate diverse positive and hard negative samples that sculpt the semantic space from multiple angles. We then apply a two-round LoRA adaptation: JD–JD contrastive training followed by JD–CV triplet alignment on a heterogeneous text dataset. Importantly, these gains require no large-scale manually labeled industrial training pairs: a modest set of real JDs is expanded into supervision through LLM synthesis. Finally, a BoundaryHead MLP reranks the Top-K results to distinguish between roles that share the same title but differ in scope. On a local pool of 300 real JDs with candidates from an upstream production retriever, mira-embeddings-v1 improves Recall@50 from 68.89% (baseline) to 77.55% while lifting Precision@10 from 35.77% to 39.62%. On a supportive global pool over 44,138 candidates judged by a Qwen3-32B rubric, it achieves Recall@200 of 0.7047 versus 0.5969 for the baseline. These results show that LLM-synthesized supervision with boundary-aware reranking yields robust gains without a heavy cross-encoder.
[NLP-353] Subgraph-Aware Training of Language Models for Knowledge Graph Completion Using Structure-Aware Contrastive Learning
【速读】： 该论文旨在解决预训练语言模型（PLM）在知识图谱补全（KGC）任务中因仅关注文本信息而忽略知识图谱长尾分布特性及复杂拓扑结构（如子图、最短路径、度分布等）所导致的性能瓶颈问题。其核心解决方案在于提出一种面向子图感知的微调框架SATKGC，关键创新点包括：(i) 采用子图感知的迷你批处理机制，以促进困难负样本采样并缓解实体出现频率不均衡问题；(ii) 引入新型对比学习策略，重点聚焦于具有更显著结构特性的难样本正例与负例，从而增强模型对知识图谱结构归纳偏置的感知能力。该工作首次系统性地将知识图谱的结构归纳偏置融入PLM微调过程，实验在三个标准KGC基准上验证了其优越性。
链接: https://arxiv.org/abs/2407.12703
作者: Youmin Ko,Hyemin Yang,Taeuk Kim,Hyunjoon Kim
机构: Hanyang University (汉阳大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to The Web Conference 2025
Abstract:Fine-tuning pre-trained language models (PLMs) has recently shown a potential to improve knowledge graph completion (KGC). However, most PLM-based methods focus solely on encoding textual information, neglecting the long-tailed nature of knowledge graphs and their various topological structures, e.g., subgraphs, shortest paths, and degrees. We claim that this is a major obstacle to achieving higher accuracy of PLMs for KGC. To this end, we propose a Subgraph-Aware Training framework for KGC (SATKGC) with two ideas: (i) subgraph-aware mini-batching to encourage hard negative sampling and to mitigate an imbalance in the frequency of entity occurrences during training, and (ii) new contrastive learning to focus more on harder in-batch negative triples and harder positive triples in terms of the structural properties of the knowledge graph. To the best of our knowledge, this is the first study to comprehensively incorporate the structural inductive bias of the knowledge graph into fine-tuning PLMs. Extensive experiments on three KGC benchmarks demonstrate the superiority of SATKGC. Our code is available.
[NLP-354] Word-Aligned Prosody Reflects Speaker Differences More Than Depression or PTSD Severity
【速读】： 该论文旨在解决生成式语音特征（如语调）在临床情绪障碍（如抑郁症和创伤后应激障碍，PTSD）检测中的稳定性与临床相关性问题。研究从两个层面考察语调测量的稳定性：在词级层面，探究同一说话者在不同语境下是否以相似的个性化方式发音；在说话者级层面，检验跨多个词语平均后的语调特征是否能在不同录音中产生一致的个体评分。基于心理测量学原理，研究对比了四种语料库（涵盖表演、自发及临床言语）中的表现。结果显示，词级偏差在不同语境间重复性极低（0.01–0.16），而说话者级汇总特征则表现出高度一致性（ESD达0.997，MSP-Podcast为0.73–0.90，DAIC-WOZ和PTSD-STOP为0.84–0.996）。然而，这些稳定提取的说话者级语调特征与抑郁或PTSD严重程度的相关性均未超过0.15，等效性检验进一步排除了≥0.30的影响效应（所有数据集）及≥0.20在DAIC-WOZ中的可能性，且个体内部随时间变化的每日症状关联接近零（|\rho| ≤ 0.04）。因此，尽管通过聚合可获得稳定的说话者级语调指标，但其在所研究语料中无法有效追踪抑郁或PTSD的严重程度。解决方案的关键在于：采用多层级分析框架结合心理测量学标准，验证了语调特征的稳定性，同时揭示了当前方法在临床效度上的局限性。
链接: https://arxiv.org/abs/2610.15181
作者: Hsiang-Chen Yeh,Aurosweta Mahapatra,Shreeram Suresh Chandra,Ryan L. Boyd,Berrak Sisman
机构: 未知
类目: Audio and Speech Processing (eess.AS); Computation and Language (cs.CL); Sound (cs.SD)
备注: 6 pages, 2 figures, 2 tables. Code: this https URL
Abstract:Word-aligned prosody is increasingly used to detect depression and posttraumatic stress disorder (PTSD), yet the stability and clinical relevance of these measurements remain unclear. We examine stability at two levels. At the word level, we ask whether a speaker produces the same word in a similar speaker-specific way across different utterance contexts. At the speaker level, we ask whether averaging prosody across many words produces consistent speaker scores across separate recordings. Drawing on principles of psychological measurement, we compare these representations in four corpora spanning acted, spontaneous, and clinical speech. Word-specific deviations show little cross-context recurrence (0.01-0.16), whereas speaker summaries are highly consistent, reaching 0.997 in ESD, 0.73-0.90 in MSP-Podcast, and 0.84-0.996 in DAIC-WOZ and PTSD-STOP. Despite this consistency, associations with depression or PTSD severity do not exceed 0.15; equivalence tests exclude effects of 0.30 in all analyses and 0.20 in DAIC-WOZ, while within-person associations with daily PTSD symptoms are near zero ( |\rho|\leq0.04 ). Thus, aggregation yields stable speaker-level prosodic measures, but these measures do not detectably track depression or PTSD severity in the studied corpora. Code is available at this https URL.
[NLP-355] Description Complexity Bounds for Analytic Approximations of Lipschitz Functions
【速读】： 该论文旨在解决在给定精度要求下，用有限表达式（即由基本运算符 +、−、×、÷、exp 构成的有理常数与变量构成的有限树结构）对 Lipschitz 函数进行统一逼近时所需的最小描述复杂度（description complexity, \dc(f)）问题。核心问题是：对于定义在 [0,1] 或 [0,1]d 上的 L-Lipschitz 函数，其在精度 Δ 内的逼近表达式的最短二进制描述长度是多少？解决方案的关键在于构造一种基于线性样条（linear spline）并利用逻辑斯蒂坡道（logistic ramps）平滑的逼近形式，结合一个重复构造（repetition construct），使得一维函数的描述复杂度达到上界 \dc(f) \le C(L/\Delta + \lg(1/\Delta + 2))。该方法通过将函数逼近分解为局部线性段，并以光滑过渡实现整体连续性，同时利用重复构造有效压缩表达式长度。研究进一步证明了该上界的紧致性：在任意描述语言下，存在至少需要 161(L/Δ+lg(1/Δ+2)) 比特的 L-Lipschitz 函数，从而表明一维情况下的最坏情况描述复杂度为 Θ(L/Δ+lg(1/Δ+2))。对于高维情形，采用张量积样条，其中系数为混合二阶差分，每个系数需 O(d) 比特表示，得到复杂度 Θd((L/Δ)d+lg(1/Δ+2))。若不使用重复构造，则上界会引入额外的对数因子，其必要性仍待进一步验证。数值实验通过参考编码器、解码器与评估器验证了构造的有效性。
链接: https://arxiv.org/abs/2610.14893
作者: Artem Vozmishchev
机构: ITMO University (圣彼得堡国立信息技术、机械与光学大学)
类目: Classical Analysis and ODEs (math.CA); Computation and Language (cs.CL)
备注: 19 pages, 6 pages, 2 tables
Abstract:We study how many bits are needed to write down an elementary analytic expression that uniformly approximates a Lipschitz function. Expressions are finite trees over the basis +,-,\times,÷,\exp\ with the variables and rational constants, and their size is \dc(f) , the length of the shortest description of f in a fixed binary description language. For functions of one variable we prove that every L -Lipschitz F:[0,1]\to[0,1] has an everywhere-defined approximant of accuracy \Delta with \dc(f)\le C,(L/\Delta+\lg(1/\Delta+2)) for all L,\Delta0 , by a linear spline smoothed with logistic ramps in a description language with one repetition construct. We prove a matching lower bound valid for \emphevery description language: for \Delta\le 1/36 and every L0 some L -Lipschitz function needs at least \tfrac116(L/\Delta+\lg(1/\Delta+2)) bits. Hence the worst-case description complexity is \Theta(L/\Delta+\lg(1/\Delta+2)) . For functions on [0,1]^d the same method gives \Theta_d\bigl((L/\Delta)^d+\lg(1/\Delta+2)\bigr) , using a tensor-product spline whose coefficients are mixed second differences and cost O(d) bits each. Without the repetition construct the upper bounds carry an additional logarithmic factor, whose necessity for plain expression trees is left open. A reference encoder, decoder and evaluator confirm the constructions numerically, and figures and tables report the checks.
[NLP-356] RUBATO: Decoupled Clocks Allow Full-Duplex Models to Think as Fast as They Can
【速读】： 该论文旨在解决语音全双工（Speech Full-Duplex, FDX）模型在引入内部思考流（internal thinking streams）后，因采用固定解码速率导致的实时性与硬件适应性严重受限的问题。现有方法强制所有思考流以统一的静态解码率运行，依赖预设的硬件延迟假设，无法动态适配不同性能设备，从而限制了模型在真实场景下的响应效率与鲁棒性。其核心解决方案是提出RUBATO——一种新型多流位置编码机制，通过为每个思考流分配独立的时间时钟（independent clocks），实现各流间的完全解耦，使模型能够根据实际硬件能力自主调节解码速度。这一结构设计使得模型具备零样本跨推理速率插值的能力：仅需在训练阶段使用两种极端解码速率，即可在部署时无缝支持任意中间速率。实验表明，RUBATO-Full在Spoken GSM8K任务中于所有插值速率下保持48.7%至51.1%的准确率，相比基线模型在中间速率下降至1.7%的崩溃表现显著提升；同时，在复杂全双工中断场景中，当用户打断模型主动推理时，RUBATO-Full仍维持40.0%至45.1%的准确率，而基线模型则完全失效，充分验证了其在异步处理、快速响应和抗干扰方面的优势。
链接: https://arxiv.org/abs/2610.14299
作者: Yi-Jen Shih,Puyuan Peng,Haohe Liu,David Harwath,Abdelrahman Mohamed
机构: The University of Texas at Austin; FAIR, Meta Superintelligence Labs
类目: Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Sound (cs.SD)
备注:
Abstract:Recent Speech Full-Duplex (FDX) models are capable of engaging in naturalistic conversation by listening and speaking at the same time. Building on this foundation, recent architectures introduce internal thinking streams to enhance their intelligence. To guarantee these streams process successfully in real time, existing methods enforce a fixed decoding rate, which assumes a hardware profile capable of a specific latency. We argue this static approach is severely suboptimal. Because internal thinking is intrinsically asynchronous, an ideal FDX model should sustain real-time conversational flow while allowing each stream to decode at its own hardware-adaptive rate. Thinking streams must process as fast as the hardware allows to deliver accurate answers faster. To solve this, we propose RUBATO, a novel multistream positional encoding that assigns independent clocks to entirely decouple the streams. This structural isolation empowers the model to zero-shot interpolate across any inference decoding speed. It drastically simplifies dataset construction by requiring training on only two extreme decoding speeds while smoothly adapting to any intermediate rate during deployment. RUBATO demonstrates consistent robustness across five spoken reasoning tasks. On Spoken GSM8K, a baseline model collapses to 1.7% accuracy at interpolated speeds. Conversely, RUBATO-Full sustains 48.7% to 51.1% accuracy across all rates, leveraging faster hardware to achieve a 6.5x shorter time-to-answer with a constant time-to-first-response. This advantage holds on BIG-Bench Audio and amplifies significantly during complex full-duplex interruptions. Specifically, when a user interjects while the model is actively speaking and reasoning, RUBATO-Full achieves 40.0% to 45.1% accuracy, whereas the baseline fails completely.
信息检索
[IR-0] Kwai Summary Attention V2 Technical Report
链接: https://arxiv.org/abs/2610.18654
作者: OneRec Team,Chenglong Chu,Guohong Mu,Guowang Zhang,Hao Peng,Hui Wang,Jian Liang,Jiangxia Cao,Jiao Ou,Jiaxin Deng,Lu Ren,Muhao Wei,Ruiming Tang,Tianle Xu,Wenwu Ou,Xingmei Wang,Zhixin Ling,Zhiyuan Liang,Ziqi Wang,Chaoyi Ma,Haixiao Gao,Kuo Cai,Renpu Liu,Ruochen Yang,Wenlong Xu,Xinqi Jin,Yan Sun,Yifan Wang,Yingzhi He,Yiyu Wang,Yusen Huo,Zhaojie Liu,Zihan Zhou
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: Work in progress
Abstract:Summary-token architectures such as Kwai Summary Attention (KSA) compress distant context into learnable Summary token per chunk, and hybrid variants interleave these layers with full attention to retain exact access to the original text. In such hybrid KSA-Full architecture, the interleaving inserted summary tokens are served as a streaming flow role across different layers. Although effective, however, it also introduce some bottlenecks: (1) the different layers summary compression chunk must be same, therefore hard to reach a flexible KV Cache compression, e.g., different layer set different summary compression chunk window. (2) in the full-attention layers, the summary token introduced additional computation cost, but such layer ignore them and still pay quadratic prefilling, therefore unfriendly to the computation side. Motivated by them, we developed the KSA V2, which not only achieves a more flexible KV cache memory compression but also reduces the computation cost in Full attention layer. Specifically, we found that the summary tokens can serve two roles: the KV memory and sparse indexer.
[IR-1] Structural Segmentation for Brazilian Legal Document Retrieval
链接: https://arxiv.org/abs/2610.18622
作者: Leandro Carísio Fernandes,João Alberto de Oliveira Lima,Hudson de Martim,Eduardo Seiti de Oliveira,Roberto Lotufo,Jayr Pereira
类目: Information Retrieval (cs.IR)
备注:
Abstract:Long legal documents are often represented as single retrieval units, which can dilute relevant content and limit retrieval effectiveness. This study investigates structural segmentation as a document representation strategy to improve Brazilian legal document retrieval. Using the hierarchical structure of legal acts, we represent each document as a collection of chunks. We evaluate four representation strategies: a multi-layered hierarchy representation and three simplified versions based on core chunks (articles and non-normative document components), with varying contextual details. We test these strategies on two publicly available Brazilian legal datasets containing normative acts and legislative proposals, using both lexical and dense retrieval models. Results show that structural segmentation improves dense retrieval compared with whole-document indexing, despite substantially increasing the number of indexed units. However, it does not consistently benefit lexical retrieval. Finally, we find that full hierarchicy representation is not necessary: core chunks enriched with document metadata yield comparable or better results.
[IR-2] Knowledge Graphs as Files: An Agent Harness Where Tool Creation and Tool Use Co-Evolve
链接: https://arxiv.org/abs/2610.18616
作者: Yuyang Bai,Yu Wang,Yu Zhang
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: 27 pages. Code and data: this https URL
Abstract:Searching text-rich knowledge graphs requires following relations while examining textual evidence. Predefined graph interfaces constrain what evidence agents can access, even when they write code to combine tool calls or process retrieved subgraphs. We propose direct knowledge graph interaction (DKGI), a harness in which language model agents create graph access tools directly over complete node and edge files. The agent writes Python code to follow relations, combine node sets, and select textual evidence, interleaving execution with shell search and file inspection. This lets agents process large node sets in bulk while returning selected text for the model to assess. Execution feedback guides tool creation and revision, while the evolving tools determine what evidence the agent gathers next. On benchmark datasets spanning three domains, training-free DKGI exceeds the strongest published baseline by at least 6.7 Hit@1 points on every graph, even with a minimal harness. Controlled ablations show the benefits of multi-turn programming over shell-only access and predefined graph tools, while trajectories show agents revising code to extend searches or address errors. These findings highlight the value of allowing tool creation and tool use to co-evolve during graph search. Our code and data are available at this https URL.
[IR-3] Retrieval Disagreement as a Signal: Diagnosing and Steering Multilingual Dense Embedders
链接: https://arxiv.org/abs/2610.18526
作者: Faeze Ghorbanpour,Shaghayegh Kolli,Jana Diesner,Alexander Fraser
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注:
Abstract:Multilingual dense retrievers place all languages in one shared space so that a question finds relevant documents, regardless of which language they are written in. We show that the same question, asked in different languages, retrieves substantially different documents, and this disagreement is not because of weak language understanding. Each language performs well on its own and ranks candidates consistently with the other languages; divergence appears only when languages share an index. We trace the mechanism: each language keeps its own ranking, but in the shared index, its correct answers are pushed far down, buried under non-relevant documents that are either in the query’s own language or in English, depending on the model. We introduce disagreement-guided steering: for each language, we contrast the documents that other languages agree on for a question against those that language retrieves idiosyncratically, and take the difference of their representations as a language-specific direction, added to that language’s queries at inference. Against baselines of language-mean removal, subspace removal, and per-language rotation, it most strongly reduces same-language crowding while remaining competitive on recall, trading off against mean removal, which better preserves ranking quality.
[IR-4] Graph-free Retrieval with Graph-Modified Embeddings
链接: https://arxiv.org/abs/2610.18410
作者: Federica Santi(1),Vasilije Marković(1),Daniel Scott(1),Lazar Obradović(1) ((1) Cognee Inc.)
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
备注: 28 pages, 10 figures, including appendices
Abstract:Dense vector retrieval can miss evidence connected through intermediate entities, while direct graph retrieval requires access to the graph at query time. We compare five offline graph-based embedding transformations and strategies for using the resulting graph-modified embeddings alongside the originals, without query-time graph access. On a synthetic professional-network benchmark with complete supporting-evidence sets, direct replacement of the original embeddings reduces evidence recovery. Multi-stage strategies improve recovery by using graph-modified embeddings to guide the search while retaining the originals for semantic relevance. On questions requiring linked facts or comparisons among candidates, the strongest configuration improves recovery by up to 25.9% relative to plain vector retrieval. It also matches baseline recovery with 40% fewer returned nodes. Controls confirm that graph information contributes beyond multi-stage retrieval alone. Across the tested methods, retrieval performance depends more on how graph-modified embeddings are used than on how they are produced.
[IR-5] SimInter: Evidence-Grounded User Simulation with Interview-Based Profile Enrichment for Conversational Recommender Systems
链接: https://arxiv.org/abs/2610.18308
作者: Erfan Shahabi,Amirhossein Roshandel,Mohammad Sadegh Sadeghi,Azadeh Shakery
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: 25 pages, 11 figures, 12 tables. Erfan Shahabi and Amirhossein Roshandel contributed equally
Abstract:Conversational Recommender Systems (CRSs) require reliable user simulators for cost-effective evaluation. LLM-based simulators have made realistic simulation possible but typically condition on profiles derived from explicit ratings, demographics, and static sources capturing surface preferences while omitting implicit behavioral characteristics such as decision-making style, aspect-level preferences, and domain expertise. We propose SimInter, a pipeline that elicits these attributes via synthetic interviews between two LLM agents, grounded in the user’s statistical profile and historical conversational evidence. Before profile enrichment, an audit agent explicitly flags potential fabrication and stereotyping at the turn level. We instantiate and evaluate SimInter on movie recommendation. We simulate 300 ReDial users and, on a matched subset of 100 users, compare SimInter with a statistics-only baseline: SimInter significantly increases the diversity of expressed preference aspects (filtered aspect entropy: 1.619 vs. 0.859; paired d = 0.96, Holm-adjusted p 0.001). This gain persists when the simulator converses with two trained recommenders, KBRD and UniCRS, under two independent evaluation judges (+0.28 to +0.56 nats; all p 0.001). Human evaluation finds 94.3% of enriched-attribute claims supported or partially supported, while human judgments agree with 95.2% of the auditor’s binary filtering decisions on a stratified sample.
[IR-6] JevGPT : Thinking Fast and Slow in Autonomous Research
链接: https://arxiv.org/abs/2610.18272
作者: Yasheng Sun,Yanning Dai,Yimeng Chen,Nanbo Li,Firas Laakom,Yifan Xu,Piotr Piękos,Mateusz Ostaszewski,Wenyi Wang,Cheng Luo,Yuhui Wang,Jürgen Schmidhuber
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: 24 pages, 6 figures, 7 tables. Project page: this https URL
Abstract:Autonomous research agents must acquire and reason over evidence within a limited budget. Yet relevant sources can remain uninspected or unused, leaving a gap between retrieval coverage and the evidence that informs research. The challenge is to align fast local judgments with changing evidence needs and feed their findings back into research decisions. We introduce JevGPT, a dual-system architecture for autonomous research that couples fast evidence processing with deliberative research planning. System Two sets evidence objectives and integrates findings, while System One locates support, assesses relevance, and guides further inspection. Independent source judgments run in parallel; selected original passages inform the next research step. Experiments on fixed-corpus question answering and document ranking show that JevGPT achieves leading task performance, with faster completion and lower cost than comparably effective research agents. Evidence analysis shows broader source inspection and greater coverage of supporting documents in the research loop, highlighting efficient evidence processing as a route to more capable autonomous research.
[IR-7] Web Scale on a Budget: Tail-Aware Retrieval Evaluation
链接: https://arxiv.org/abs/2610.17986
作者: Boaz Meivar,Shai Avidan
类目: Computer Vision and Pattern Recognition (cs.CV); Information Retrieval (cs.IR)
备注: 19 pages, 8 figures, 10 tables
Abstract:A common practice in instance retrieval is to grow the gallery size to improve a benchmark’s discrimination capability. One prominent example is ILIAS, which adds up to 100 million distractor images. The prevailing assumption is that faithful evaluation demands web scale, and that indexing all 100M images through the backbone is the unavoidable price of a comparable number. That price is steep: hundreds of GPU-hours and ~200 GB of descriptors per model, paid again for every method or ablation. We show the web-scale indexing price is largely avoidable: evaluation itself obeys a scaling law. The number of distractors that outrank a positive is Binomial in the gallery size. We define the per-query retrieval hardness, a gallery-scale agnostic quantity that yields a retrieval metric at any scale, and demonstrate it can be derived from a small random sample. The underlying sampled-rank machinery is long established in recommender-system and biometric evaluation, but to our knowledge has never been adapted to visual retrieval. We make the law per-query and estimable, and validate it at true 100M scale on ILIAS and INSTRE, and on the full 759k-distractor Google Landmarks v2 pool. Concretely, within a 5% sample (a flat 20x less indexing), every backbone’s tail-aware (EVT) estimate has converged onto the large-scale mAP, to within ~0.01 (mean absolute error 0.005), across five backbones, three benchmarks, and two distractor pools (mAP@1000 at 100M on ILIAS/INSTRE; the official mAP@100 at the full 759k pool on GLDv2); the recommender-systems count baseline over-predicts everywhere (+0.037 to +0.068). This is an order of magnitude better than naively reporting the metric on the same subset (+0.08 to +0.13). We release code and the per-query hardness to facilitate faster, cheaper, and greener retrieval evaluation.
[IR-8] FASHION-RQ: Mining Hierarchical Semantic Structure for Automated Evaluation of Fashion Retrieval
链接: https://arxiv.org/abs/2610.17841
作者: Tripti Shukla,Shwetha Somasundaram,Akshat Gupta,Rohit Garg,Paul Duff,Ian Anderson
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Evaluating fashion retrieval systems remains challenging due to the limited availability of benchmarks that capture the fine-grained, open-ended queries encountered in real-world search and recommendation. Existing benchmarks often rely on manual curation, predefined taxonomies, or synthetic prompts, limiting scalability and semantic coverage. We introduce FASHION-RQ: Mining Hierarchical Semantic Structure for Automated Evaluation of Fashion Retrieval, an automated framework that derives user-type queries and relevance sets directly from catalog structure. Our key insight is that hierarchical discrete representations learned by Residual Quantized Variational Autoencoders (RQ-VAE) organize products into semantically coherent clusters at progressively finer granularities. We leverage this structure to construct FASHION-RQBENCH, a scalable benchmark without manual query annotation or predefined taxonomies. We propose four complementary tasks for evaluating ranking, relevance classification, fine-grained sibling discrimination, and paraphrase robustness, alongside query- and attribute-level relevance metrics. Experiments across recent retrieval models demonstrate that FASHION-RQBENCH differentiates model capabilities across query types and attributes. A human study further shows that benchmark queries achieve comparable naturalness and higher intent clarity than real user queries, supporting their validity for real-world evaluation. The code and FASHION-RQBENCH dataset will be publicly released upon acceptance.
[IR-9] Mitigating Popularity Bias in Sequential Recommendations with Sparse Prior-Regularized Objective
链接: https://arxiv.org/abs/2610.17823
作者: Anton Pembek,Anton Klenitskiy,Alexander Savchenko,Alexey Vasilev
类目: Information Retrieval (cs.IR)
备注:
Abstract:Popularity bias is a well-known challenge in recommender systems, often leading to the over-exposure of popular items and reduced catalog coverage. In this work, we propose PR-entmax, a sparse prior-regularized objective that makes popularity correction context-dependent. We derive our method from a generalized variational principle in which an item popularity prior is incorporated directly into a sparse probability mapping, yielding a popularity-aware active support: the set of items receiving non-zero probabilities and hence non-zero gradients is determined jointly by contextual relevance and item popularity. As a result, the prior affects not only item probabilities, but also which items participate in gradient updates during training. The method requires no architectural changes and can be integrated into existing training pipelines by modifying only the loss function. At inference time, the prior term is removed, and items are ranked by the learned contextual relevance alone. Experiments across multiple benchmark datasets and sequential recommendation architectures show that PR-entmax consistently reduces popularity bias while preserving or improving recommendation relevance, advancing the relevance-debiasing trade-off.
[IR-10] he Scored Population the Operating Point and the Pipeline: Three Objects That Decide What an Open-Web Event System Looks Like
链接: https://arxiv.org/abs/2610.17814
作者: Shengqi Zhou,Zhigang Liu,Wenxiang Zhu,Boyong Du,Kairui Yang,Haicheng Huang,Weijian Bu
类目: Information Retrieval (cs.IR)
备注: 12 pages, 5 figures
Abstract:Open-Web streams can reference real-world events without making them suitable for organization: a correctly linked event may serve only as background, while an event absent from the knowledge base may be central to an item. This distinction reveals an evaluation blind spot in multistage Web systems: pipelines determine not only how inputs are scored, but also which inputs receive scores at all. Using open-Web event admission as a case study, we examine three connected objects (the scored population, the selected operating point, and behavior on the complete input stream) through shared-subset comparisons, admission-loss decomposition, and controlled extraction-recovery interventions. We find that reported ranking differences depend on the scored population, scoring gaps exhibit distinct backbone-dependent failure modes, and recovering missing scores can increase false acceptance under an unchanged threshold. On Original-302, Qwen candidate validity scores 255 of 302 items while direct admission scores all 302; evaluating both on the shared subset narrows their AUC gap. For Mistral, span-safe recovery raises false acceptance from 73.1% to 92.5% with the threshold fixed. Across 602 human-annotated evaluation items within a 3,290-record resource, these findings establish a reporting framework that considers score availability, threshold selection, and complete-stream decisions together.
[IR-11] SSEmb: Training a Generalizable Speech Retrieval Model Using Only Automatic Speech Recognition Data
链接: https://arxiv.org/abs/2610.17813
作者: Sangkwon Park,Dongwook Lee,Youngho Cho,Donghun Kang,Heeseung Kim,Sungroh Yoon
类目: ound (cs.SD); Information Retrieval (cs.IR)
备注:
Abstract:End-to-end speech-to-text retrieval is becoming an important capability for voice agents, allowing spoken queries to retrieve relevant documents without intermediate transcription. Training these retrievers typically depends on large collections of spoken query-document pairs that are costly to acquire and difficult to extend across domains and languages. We introduce \ours, which replaces such task-specific supervision with speech-transcript pairs from standard ASR corpora. A lightweight adapter projects the output of a frozen speech encoder into the retrieval space of a frozen pretrained text embedding model. \ours is competitive with cascaded ASR-then-retrieve systems while reducing inference latency, and outperforms existing publicly available speech embedding models. It also transfers to text-to-speech and speech-to-speech retrieval without additional training. Controlled analyses validate the speech-transcript alignment paradigm, demonstrating robustness to reduced training data and consistent gains from larger corpora. Further ablations of loss functions and trainable modules establish practical guidelines for the framework. Model weights and codes are to be released.
[IR-12] More Reader Data or a Better Teacher? Updating a Model of Crowd Highlights in Late 2026
链接: https://arxiv.org/abs/2610.17768
作者: Kazuki Nakayashiki,Keisuke Watanabe
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL)
备注: 10 pages, 4 tables, 1 figure. Ancillary files include the pre-registration with its amendments, the results write-up, the aggregate artifacts every reported number is generated from, and the prompt, scoring, target and contrast code
Abstract:A model trained on readers’ highlights can show a newcomer where earlier readers converged on a page, even on pages with few readers. A year of stronger language models suggests two upgrades: use them directly, or use them as teachers. We compare both with a third, more reader data, on one social highlighting platform, with the comparisons, the selection rules and a sealed set of fresh pages fixed in advance. On 215 development pages, seven language models released by October 2026 rank sentences zero-shot at an average precision of at most 0.3744 against the sentences half the readers marked most, and 0.3931 when fused with reading position, while a 3-billion-parameter model trained on reader highlights alone reaches 0.4346. Used as the teacher that breaks ties among readers’ counts, the stronger fusion changes a trained model by +0.0013 [-0.0012, +0.0040], whereas going from half to all of its training pages adds +0.0242 [+0.0179, +0.0301]. A 27-billion-parameter backbone on the reader-only recipe adds +0.0225 on the development pages; on 131 fresh pages, read once and too few for a confirmatory comparison, its difference from the 3-billion model is unresolved, both score above the fusion, and in an exploratory split their lead over the previous model sits on pages from domains that their fresh training pages also cover. In this recipe, more reader data improves a trained model where a stronger tie-breaking teacher yields no material gain, and frontier models used as rankers still trail a small model trained on readers.
[IR-13] RALI@iKAT 2026: Retrieval-aware User Profile Denoising via Personalized Conversational Dense Retrieval SIGIR2026
链接: https://arxiv.org/abs/2610.17673
作者: Yuchen Hui,Fengran Mo,Jia Ao Sun,Yuxing Tian,Wenhao Xu,Jian-Yun Nie
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL)
备注: 12 pages, 4 figures. Presented at the SCAI Workshop at SIGIR 2026. Ranked 1st in both retrieval performance and response quality at iKAT 2026
Abstract:Modern conversational information access systems continuously extract user information from conversations to build user profiles for personalized retrieval. As user profiles accumulate over time, they inevitably become noisy, making effective user profile denoising a fundamental challenge. Existing personalized conversational query reformulation methods rely on the black-box reasoning capability of LLMs to expand the original query with relevant user profile phrases, producing a verbose query for retrieval. Such rewrite-then-retrieve pipelines reduce user profile denoising to the explicit selection of profile phrases in the sparse token space, separating personalization from downstream relevance modeling. Consequently, they often fail to identify user information that is truly beneficial for retrieval. To enable retrieval-aware user profile denoising, we propose Personalized Conversational Dense Retrieval (PCDR), which learns to jointly project the conversation history and user profile into a latent embedding space dedicated to retrieval. By inheriting the context denoising capability from conversational dense retrievers, PCDR requires only a handful of personalized relevance annotations for training. We evaluate PCDR on the iKAT 2026 tasks, where it serves as the core retrieval component of a conversational AI agent assisting users with complex information needs. Comments: 12 pages, 4 figures. Presented at the SCAI Workshop at SIGIR 2026. Ranked 1st in both retrieval performance and response quality at iKAT 2026 Subjects: Information Retrieval (cs.IR); Computation and Language (cs.CL) ACMclasses: H.3.3 Cite as: arXiv:2610.17673 [cs.IR] (or arXiv:2610.17673v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2610.17673 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[IR-14] AiSAQ-in-Filtering for DRAM-free Filtered Information Retrieval
链接: https://arxiv.org/abs/2610.17627
作者: Kento Tatsuno,Daisuke Miyashita,Kazunari Sumiyoshi,Hiroaki Yamanaka,Kosuke Nakata,Jun Deguchi
类目: Information Retrieval (cs.IR)
备注: 5 pages, 7 figures
Abstract:Filtering search is one of the crucial applications of approximate nearest neighbor search (ANNS). In filtering search, label information is used in conjunction with unstructured vector data. ANNS provides efficient similarity search in large-scale datasets, however, the memory footprint of storing both vector data and their associated label becomes a significant cost factor as dataset sizes grow. To address this problem of increasing DRAM usage, we propose AiSAQ-in-Filtering, with a novel quantization method called LiPQ (Label-inclusive Product Quantization). LiPQ combines unstructured data and structured labels into a single representation, leading no growth of encoded data. Our proposal increased recall@1 of single-label searches by up to 55% and fewer disk I/Os compared with existing methods. This also retains the main advantages of AiSAQ, which can reduce DRAM footprint by up to 1/78 and index switching time comparing with existing methods. This proposal can also be applied to sparse-dense hybrid search beyond label filtering such as semantic and keyword search of documents.
[IR-15] Reference Wording Shapes Retrieval of Robot Task Instructions
链接: https://arxiv.org/abs/2610.17549
作者: Jahnavi Yelamanchi
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Information Retrieval (cs.IR); Robotics (cs.RO)
备注: 6 pages, 4 figures, 4 tables
Abstract:Identifying a robot task from a short instruction is sensitive to which descriptions are available for retrieval. We investigate this effect in the natural-language instructions included with the RLBench manipulation benchmark. We extract 296 fixed descriptions from 62 publicly available task files and compare five training-free lexical rankers over 80 random choices of one indexed description per task. Exact strings assigned to conflicting task labels are excluded from the primary score. Character n-gram TF-IDF achieves 61.3% mean top-1 accuracy on the resulting held-out instructions, compared with 55.6% for BM25 and 54.9% for word TF-IDF. Across gallery choices, character TF-IDF ranges from 55.2% to 66.8%; its 5th-95th percentile range is 56.9%-64.8%. Adding a second indexed description improves character TF-IDF by 7.0 percentage points on matched queries. These results quantify the influence of reference wording and representation choice on instruction-to-task retrieval under a consistent evaluation protocol.
[IR-16] Computed not Memorized: Bit-Factorized Identifiers for Efficient Generative Retrieval
链接: https://arxiv.org/abs/2610.17467
作者: I-Fan Lin,Zhen Zhang,Faegheh Hasibi,Suzan Verberne
类目: Information Retrieval (cs.IR)
备注:
Abstract:Generative retrieval is known for its efficient end-to-end indexing and retrieval. In particular, numeric Document Identifiers (DocIDs) are generally more efficient because they are shorter and generally fixed-length sequences over a small vocabulary, requiring fewer decoding steps during beam search. However, they tend to generalize poorly to unseen documents and must be learned during indexing-stage training. In this paper, we propose a numeric DocID design for generative retrieval that enables better adaptation to unseen documents without this http URL DocID design inherits the language model’s knowledge: each DocID is a sequence of tokens factorized into bits, and each token is scored as the sum of its bits. Unlike existing numeric DocIDs, whose tokens carry meaning only through embeddings memorized during indexing training, each bit here derives its meaning from the model’s representation. Therefore, the DocIDs of unseen documents remain scorable without this http URL on three benchmarks show that our DocID design substantially improves unseen-document retrieval while maintaining comparable performance on seen documents. It achieves over 20 percentage points higher Hit@10 on unseen documents and up to 7 times lower inference latency than existing numeric DocID designs on NQ320K and TriviaQA, while performing only 3 percentage points below weight-update methods on MS MARCO without additional training. These results show that an appropriate DocID design can heavily reduce retraining costs when adapting generative retrievers to new documents.
[IR-17] HASTE: Hierarchical and Attribute-aware Supervision for Generative Recommendation
链接: https://arxiv.org/abs/2610.17368
作者: Tripti Shukla,Shwetha Somasundaram,Akshat Gupta,Rohit Garg,Paul Duff,Ian Anderson
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Generative recommendation models have shown promising results by representing items as semantic IDs and generating these IDs autoregressively. We introduce an approach showing that explicitly incorporating hierarchical and attribute-level supervision into generative recommendation can simultaneously improve training efficiency and long-tail recommendation performance. Existing approaches typically optimize semantic ID prediction with a standard causal cross-entropy objective, which does not explicitly exploit the hierarchical structure of semantic IDs and can require substantial training to learn effective item representations. Moreover, data-sparse tail items often suffer from poor representations due to limited supervision, resulting in weaker long-tail recommendation performance. We address these challenges with HASTE (Hierarchical and Attribute-aware Supervision for generaTive rEcommendation), a unified training framework that introduces hierarchical and attribute-specific supervision. HASTE’s Hierarchical Loss explicitly exploits the multi-level structure of semantic IDs, providing supervision at each level of the item hierarchy and substantially accelerating training convergence. Its Attribute-Specific Losses encourage the model to preserve fine-grained item attributes, leveraging the language model’s inherent semantic knowledge to improve representations and recommendations for tail items. We extensively evaluate HASTE on a fashion recommendation benchmark and across multiple domains of the Amazon dataset. HASTE improves recommendation performance across datasets and item-frequency regimes. Notably, the Hierarchical Loss achieves approximately 2x faster convergence than the standard causal cross-entropy objective, substantially improving training efficiency while maintaining strong recommendation quality.
[IR-18] From Conceptual EneMonIE to an Executable Backend: A Modular Architecture for Energy Information Extractionand Grounded Question Answering
链接: https://arxiv.org/abs/2610.17243
作者: Janik Lesage,Khadim Lô,and Jean-Sébastien Dessureault
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注:
Abstract:The original EneMonIE proposal describes a Web-based semantic system for continuously monitoring energy trends through targeted information extraction from heterogeneous online media. This article presents EneMonIE-Backend, an implementation-derived redesign that converts that conceptual pipeline into a modular, executable, and application-independent backend. The architecture separates five concerns: data sources, acquisition and normalization, semantic processing, knowledge organization, and persistent storage. It operationalizes configurable connectors for RSS feeds, URL collections, arXiv, social-media JSON files, and open-data CSV files; it also introduces question-driven Web search and scraping. Relevant documents pass through energy categorization, named entity recognition, temporal and event extraction, source sensitive sentiment analysis, and extractive summarization. The resulting semantic representations support entity co-occurrence graphs, cross-document interlinking, cluster-level information fusion, and relational persistence in SQLite. Three grounded question-answering paths are exposed at the backend level: live Web evidence, the stored corpus, and selected configured resources. Answer generation uses a language model when configured and otherwise falls back to deterministic extractive output. Unlike the original proposal, the presented architecture deliberately excludes the user-interface and server-side application layers, allowing the semantic core to be embedded in different applications. A deterministic smoke validation confirms relevance filtering, persistence, clustering, evidence retrieval, and document deletion. The work provides a reproducible architectural basis for future evaluation on larger multilingual energy corpora.
[IR-19] racing Money Flows in the Multi-Token Cryptocurrency Network
链接: https://arxiv.org/abs/2610.16981
作者: Natkamon Tovanich,Célestin Coquidé
类目: ocial and Information Networks (cs.SI); Computational Engineering, Finance, and Science (cs.CE); Computers and Society (cs.CY); Emerging Technologies (cs.ET); Information Retrieval (cs.IR)
备注: 14 pages, 4 figures, 1 table, 19 references, 15th International Conference on Complex Networks and Their Applications
Abstract:Tracing money flows in cryptocurrency transaction networks is challenging because funds can move across accounts, transactions, and token layers. We propose a transaction-aware graph in which nodes represent account–asset states within transactions, while edges encode observed transfers, inferred asset transformations, and temporal inter-transaction sequences. Applied to Aave borrow–repay flows, we compare graph construction and edge-pruning strategies and extract shortest paths using a heuristic weighting. Our results reveal a trade-off between reachability and interpretability and uncover frequent token sequences underlying how borrowed assets are transformed before repayment. This work lays a foundation for general, interpretable multi-token flow tracing.
[IR-20] WeGR: Unified Generative Retrieval for Industrial Search Systems in WeChat
链接: https://arxiv.org/abs/2610.16979
作者: Jiajun Han,Guoxin Ma,Felix,Liangsheng Xu,Ruiqi Yu,Yingdi Guo,Zongbing Wang,Xinhua Ye,Chenyan Hua,Xin Xie,Pengcheng Yang,Haijun Wu
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
备注: 11 pages, 4 figures, conference
Abstract:Generative retrieval (GR) offers a promising paradigm for industrial search by unifying candidate generation within an autoregressive framework. However, its effectiveness is limited by two key challenges. First, existing Semantic ID (SID) tokenizers struggle to incorporate collaborative signals while maintaining effective codebook utilization, limiting their ability to distinguish items and achieve high-quality retrieval. Second, existing reinforcement learning (RL) methods suffer from severe reward sparsity under exact SID matching, leaving many sampled trajectories without effective learning signals and hindering preference optimization. To address these challenges, we propose WeGR, a unified generative retrieval framework consisting of two complementary components. Specifically, the Collaborative and Collapse-free Tokenizer (CoCo-VAE) incorporates query-to-item and item-to-item collaborative signals while mitigating codebook collapse through codebook refresh and complementary regularization, enabling more informative and balanced SID representations. Building upon these SIDs, Dynamic Horizon Policy Optimization (DHPO) progressively expands both reward evaluation and policy optimization horizons, reducing learning difficulty while leveraging behavioral reward signals at different levels of the SID hierarchy, without relying on auxiliary reward models. Extensive offline experiments demonstrate the effectiveness of WeGR in improving retrieval and ranking performance. Moreover, a two-month online A/B test in WeChat E-commerce Search demonstrates relative improvements of 2.539% in orders, 1.508% in click-through rate, and 1.381% in conversion rate, while reducing online CPU and GPU usage by 25.5% and 26.1%, respectively. The system has been successfully deployed across multiple search scenarios in WeChat, serving billions of users, generating tens of millions of PVs daily.
[IR-21] KNN Implementation Details Can Dramatically Change Performance: An Example from Cover Trees NEURIPS2026
链接: https://arxiv.org/abs/2610.16940
作者: Amol Khanna,Edward Raff
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: 40th Conference on Neural Information Processing Systems (NeurIPS 2026). Track on Evaluations and Datasets
Abstract:Nearest neighbor search algorithms are widely used and have a long history, but we find that key implementation details are seldom reviewed in articles about them. These details can materially impact performance, and thus change the conclusions one makes about a given algorithm’s superiority. In this article, we highlight these impacts by reviewing the Cover Tree as an example, producing RustKNN, a unified Rust implementation of three construction variants of the Cover Tree with three query modes. Our analysis reveals: (1) the optimal query strategy depends on k and on the dataset (single-tree search is fastest at k=1 on three of our four largest datasets, while dual-tree search is fastest at k\ge5 on the three largest), and choosing the wrong query mode can cost up to a 7.5\times factor in runtime; (2) the cover tree base parameter, set to 2 in theory and 1.3 in practice, is not a safe default for dual-tree search: 1.3 is optimal on only one of seven datasets, and on covtype it is 1.6\times slower than the optimum; (3) among five micro-optimizations, only early-exit distance computation has a consistent effect, and that effect grows with dimension (none at d=32 , 1.31\times at d=54 , and 1.49\times at d=784 ); and (4) all-nearest-neighbor and held-out benchmarks yield different implementation rankings. All main timing tables report medians over 20 independent allocations with bootstrap 95% confidence intervals. RustKNN is open source at this https URL.
[IR-22] Comparing GenAI to Machine Learning and Manual Review for eDiscovery
链接: https://arxiv.org/abs/2610.16925
作者: Gordon V. Cormack,Maura R. Grossman,Jeremy Pickens
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: 8 pages; 2 figures
Abstract:Keeling et al. [Keeling2026Generative] report on a case study recently conducted by Redgrave LLP in which they assert that Relativity’s Generative AI Review tool (aiR) achieved 88% recall while Relativity Assisted Learning (RAL), its machine-learning tool, achieved 64% recall. These results are not comparable because the methods employed have different objectives, because of inter-assessor disagreement, and because they represent different recall-precision and cost-benefit trade-offs. Our analysis of the Redgrave data indicates that, when these differences are taken into account, aiR and manual review are comparable. Furthermore, established machine-learning models employing Continuous Active Learning (CAL) show state-of-the-art effectiveness, and superior efficiency, when trained only on a sample of 2,000 first-pass-reviewer responsiveness determinations.
[IR-23] From Volpiano to Pixels: 30000 Lines of OMR Ground Truth for Medieval Chant
链接: https://arxiv.org/abs/2610.16862
作者: Robert Klugseder
类目: Computer Vision and Pattern Recognition (cs.CV); Information Retrieval (cs.IR)
备注:
Abstract:Optical music recognition (OMR) for medieval chant lacks training data linking manuscript images to musical symbols across notational families. Using existing Volpiano transcriptions from the Cantus Database, we created 30,567 image-target pairs from ten sources in eleven exports, including 339 verse fragments. Their target sequences contain 1,952,143 tokens. The pipeline locates staff systems, infers clefs and maps known pitches to staff positions. Six detector families handle the different layouts and notations. Where heuristics failed, two small U-Nets learned staff lines and clef letters from manual annotations. Reviews of sampled export lines found zero to four pitch or clef errors, but used different error definitions and do not provide a uniform audit of the corpus. A CRNN trained from scratch achieved a symbol error rate of 9.6 % on 3,086 test lines. The earlier model, trained mainly on Austrian sources of the 15th and 16th centuries, scored 45.3 %. In three separate training runs, each excluding one manuscript, error rates on the excluded manuscript were 21.5-31.0 %. Model-assisted alignment recovered a further 2,686 lines.
[IR-24] Separating Prompt Harmfulness from Interaction Safety via Partial Generation
链接: https://arxiv.org/abs/2610.16761
作者: Ram Vikas Mishra,Shashank Aggarwal,Jahnavi Kumar,Amit Awekar
类目: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注:
Abstract:LLM-based information systems need safety guardrails. Most guardrails classify the prompt alone. If the prompt looks harmful, the system rejects it before any answer is generated. This is costly. Many rejected prompts would have received a safe response. Users lose help they should have received from the information system. This cost grows as conversational and generative search becomes common. Prompt-level labels treat harm as a property of the query. But harm depends on what the model actually writes. Existing input-output guards do look at responses. But they judge a finished response, so the decision comes after full generation. However, we observe that safety of model interaction with the input prompt can be inferred early with only partial output. So the system can start generating, watch the partial response, and stop only if it turns harmful. We test this idea in three steps. First, we build a dataset of 41,944 prompt-response pairs with seven open weights LLMs from three families: Llama, Gemma, and Qwen. Second, we show that prompt harmfulness and interaction safety differ. Prompt labels and response labels agree on only 50.46% of items (Cohen’s \kappa = 0.06). Of the prompts that make any model respond harmfully, only 3.5% do so for all seven LLMs. Third, we show that partial responses carry most of the signal. Fine-tuned classifiers predict response-level harm well (best AUPRC 0.885, F1 0.807). About one third of the response tokens recover almost full-response AUPRC.
[IR-25] EHR-Bench: A Chinese Multi-Task Benchmark for Recruitment AI
链接: https://arxiv.org/abs/2610.16689
作者: Ruijia Yang,Bing Xu,Zulong Chen,Zhipeng Xu,Jia Xu,Yipeng Yu,Ying Zhou,Zeyi Wen
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: 36 pages, 9 figures, 41 tables
Abstract:Recruitment applications span diverse operations, from drafting job descriptions and interpreting queries to organizing profiles, assessing fit, and recommending candidates and jobs. Systematic LLM evaluation therefore requires task coverage that reflects this range of business needs. We introduce EHR-Bench, an open Chinese benchmark comprising twelve tasks across six capability families and five recruitment scenarios. The tasks cover generation, extraction, classification, retrieval, semantic similarity, and reciprocal ranking within one evaluation framework. An internal recruitment dataset guides task definitions, the occupation taxonomy and sampling targets, output requirements, and scoring protocols. Following these specifications, we construct 8,336 task instances from public materials and controlled augmentation, with source-aware expert review and documented provenance. We evaluate twelve LLMs through task-level, capability-level, and aggregate comparisons. Closely ranked models exhibit distinct task strengths, and performance varies across input sources, reference evidence, and candidate pools. Five shared models retain the same observed Overall ordering on a private operational reference, with variation in capability and task correspondence. EHR-Bench provides a reproducible foundation for systematic model assessment, task-specific model selection, and further research on LLMs for recruitment. The project repository is available at this https URL.
[IR-26] From a Large-Scale Teacher to Efficient Multimodal Students for Search Relevance Prediction
链接: https://arxiv.org/abs/2610.16569
作者: Tomasz Gniazdowski,Mohammadreza Momenifar,Nadia Vase,Shahram Khadivi,Wiktor Walentynowicz,Nicola Ueffing,Jan Pawlowski,Piotr Szymanski,Abraham Bagherjeiran
类目: Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注:
Abstract:Relevance prediction between user queries and items is a fundamental component of e-commerce search systems. Traditionally, high-quality relevance labels are obtained through human judgment, which is reliable but expensive, slow, and difficult to scale. Recent advances in large language models (LLMs) enable automated relevance judgment at scale, achieving performance comparable to human annotators. However, such systems often rely on large commercial LLMs with complex prompts, resulting in high inference costs and limited throughput. In this work, we propose a framework that bridges this gap by distilling a high-performing LLM-based relevance judgment system into efficient multimodal student models. We first curate a small, high-quality human-annotated dataset to serve as ground truth. Using this dataset, we develop a reliable LLM-as-a-judge pipeline through careful prompt engineering and iterative refinement. This pipeline is then used to generate large-scale relevance labels, which serve as training data for compact multimodal models. Our distilled models achieve performance close to the teacher system while significantly improving latency, cost, and scalability. This approach enables practical deployment of high-quality relevance prediction systems in production environments. Subjects: Information Retrieval (cs.IR); Machine Learning (cs.LG) Cite as: arXiv:2610.16569 [cs.IR] (or arXiv:2610.16569v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2610.16569 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[IR-27] ARMED: Adaptive Retrieval-Augmented Medication Recommendation ICTAI2026
链接: https://arxiv.org/abs/2610.16440
作者: Kyungkeon Chung,Se Un Park,Hyejin Lee,Junyoung Park
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: Accepted at ICTAI 2026
Abstract:Medication recommendation requires jointly considering diagnoses, comorbidities, longitudinal patient history, and treatment interactions. Although recent retrieval-augmented generation approaches using language models have shown promise for clinical recommendation, most existing systems rely on fixed retrieval pipelines that cannot adapt to diverse clinical retrieval scenarios and may introduce noisy or irrelevant evidence. We propose ARMED, an adaptive clinical RAG framework that dynamically expands retrieval according to retrieval sufficiency and current-admission relevance. ARMED first observes the initial retrieval outcome and selectively performs supplementary or fallback multi-query retrieval when additional evidence is needed. The framework also incorporates a relevance gate that activates only clinically relevant same-patient history and organizes retrieved evidence into structured query-level clinical evidence. Experiments on MIMIC-IV demonstrate that ARMED consistently outperforms previous static retrieval baselines and improves overall medication recommendation quality. By combining locally runnable small language models with adaptive retrieval, ARMED supports medication recommendation without relying on external language model APIs or task-specific fine-tuning.
[IR-28] ackling the Modality Inhibitance Effect with Wavelet-based Modality Integration in Multi-modal Recommendation
链接: https://arxiv.org/abs/2610.16229
作者: Rongqing Kenneth Ong,Andy W. H. Khong
类目: Information Retrieval (cs.IR); Multimedia (cs.MM)
备注: Published in IEEE Transactions on Knowledge and Data Engineering (TKDE) 2026
Abstract:Harnessing rich modality features has recently gained prominence in multi-modal recommendation systems (MRSs). To delineate user modality preferences, existing works focus on modality fusion, which can be achieved via attentive summation or concatenation. In this work, we introduce the modality inhibitance effect - a phenomenon which limits existing MRSs when these models transit from dual-modal to tri-modal configurations resulting in significant performance degradation. In particular, apart from the modal-specific contamination encapsulated in the newly introduced modality, the inclusion of new features results in excessive informational redundancy, thereby severely impeding effective integration within current approaches. To address the modality inhibitance effect, we propose a new multi-scale modality wavelet (MMWave) multi-modal recommender, which aims to fully capitalize on essential modality features and suppress modal-specific noise. This is acheived by employing wavelet decomposition from a multi-scale (time \ frequency) perspective, ensuring that only essential (non-redundant) features of each modality are extracted. We then account for the latent associations between item modality features by designing a new structure-aware multi-modal graph learning module that captures high-order latent structural information pertaining to items. To modulate the relative contributions of modality features and achieve effective integration, we then propose a new adaptive modality regulator component that employs a new cross-weight integration strategy. Experiments conducted on four real-world datasets validate the efficacy of our proposed model. The source code for this work has been made publicly available at this https URL.
[IR-29] Recall Begets Recall: Recurrent Memory Ranking for Personalized LLM s
链接: https://arxiv.org/abs/2610.16137
作者: Pengyue Jia,Shida Yan,Shunyu Liu,Yingyi Zhang,Leszek Rutkowski,Chenyang Lyu,Dacheng Tao
类目: Information Retrieval (cs.IR)
备注:
Abstract:Memory ranking is the stage in personalized Large Language Models (LLMs) that decides which of the candidate memories returned by a retriever enter the limited context, so that the answer is grounded in the right parts of the user’s history. Existing memory rankers operate in a single pass, scoring each candidate once against the raw query and cutting the ranked list at a fixed top-K. The single pass misses memories that the answer needs but that look unrelated to the query on their own, and the fixed cut ignores how many memories the query needs. To address both problems, we propose ReMind, a memory ranker that runs as a recurrent and self-terminating loop, following the cognitive theory of memory search: each recalled item joins the cue for the next recall (compound cueing), and recall stops once enough has been collected for the task (metamemory monitoring). For compound cueing, ReMind adds the top-ranked memory to the query context at each step and reranks the remaining candidates under the expanded cue, so that each recall makes the cue more specific and later steps look for what is still missing rather than for what merely resembles the query. For metamemory monitoring, a sufficiency predictor estimates at each step whether the collected memories are enough to answer the query and stops the loop once they are, so that the number of selected memories adapts to each query. Experiments on public benchmarks show that ReMind achieves the best results compared to state-of-the-art memory rankers.
[IR-30] UniSGR-V2: Stepwise Target-Side Feature Anticipation and Crossing for Generative Recommendation
链接: https://arxiv.org/abs/2610.16132
作者: Jun Yang,Jiasi Chen,Jiawei Sun,Ziyue Guo,Jianan Yan,Tao Luo,Xiaoyi Zeng
类目: Information Retrieval (cs.IR)
备注:
Abstract:Semantic-ID-based generative recommendation unifies retrieval and ranking through autoregressive decoding but differs structurally from discriminative models, which can directly access rich target-item side features. The attributes are compressed into a finite codebook during quantization, while the target item remains undetermined during decoding and its features cannot be looked up. This target-side feature vacuum weakens generation at each semantic-ID level through high conditional entropy and cascading errors, and final ranking through missing target-side feature crossing. Existing methods interact with candidate representations or generate discrete attribute tokens, but none reconstructs a continuous target-side representation inside decoding to support both stepwise generation and ranking. We propose TacGR (Stepwise Target-Side Feature Anticipation and Crossing for Generative Recommendation). Before decoding each codeword, TacGR anticipates a continuous target-side feature vector and crosses it with the semantic prefix and user behavior sequence inside the decoder. The crossed representation conditions current-level generation and is directly used for ranking. Continuous-vector regression risks a learnable target collapsing with its prediction and prediction noise propagating into generation through attention. TacGR addresses these challenges with a semantic-anchoring auxiliary tower and a unidirectional visibility mask, respectively. For extremely long industrial behavior histories, it introduces gated convolutional denoising to suppress noise and adapts the pyramid memory funnel to reduce inference cost. Offline experiments on a large-scale Lazada dataset show significant gains over OneRec, UniRec, and UniSGR. Online A/B testing improves IPV, Order, and GMV by 0.79%, 5.55%, and 3.09%, respectively. Subjects: Information Retrieval (cs.IR) Cite as: arXiv:2610.16132 [cs.IR] (or arXiv:2610.16132v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2610.16132 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[IR-31] OT-FlowRec: Optimal Transport-Guided Multi-Bridge Flow for Multimodal Sequential Recommendation
链接: https://arxiv.org/abs/2610.16038
作者: Miao Sun
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
备注:
Abstract:E-commerce user preferences evolve dynamically within sessions, exhibiting concurrent and multi-centric interest distributions. While existing multimodal sequential recommendation methods leverage global attention to compress historical interactions, they inadequately capture the concurrent nature of multi-interest behaviors and fail to model the underlying multi-centric distribution of user intent. To tackle this issue, we propose OT-FlowRec, an optimal transport-guided multi-bridge flow matching framework that explicitly models multi-interest distributions to facilitate multimodal sequential recommendation. Unlike single-path generative models, OT-FlowRec constructs a mixture of parallel, interest-specific transport flows to model the multi-centric distribution characteristics, where each flow preserves a distinct user interest cluster via an optimal transport plan. Then, a geometric transport mechanism is designed to establish minimum-cost paths from the latent multi-interest distribution to target preference prototypes to provide multi-interest awareness. Moreover, a progressive time-sampling warmup strategy dynamically biases optimization weights toward the final states of the generation process, bridging the gap between continuous global distribution fitting and practical top- k ranking metrics. This approach achieves explicit interest disentanglement, ensuring the preservation of a multi-peaked preference representation instead of collapsing it into a unimodal average. Extensive experiments on three public datasets demonstrate that OT-FlowRec outperforms existing state-of-the-art baselines.
[IR-32] Personalized LLM Recommendation with Semantic IDs: Autoregressive Beam Search vs. Multi-Token Prediction NEURIPS2026
链接: https://arxiv.org/abs/2610.15864
作者: Sejoon Oh,Fernando Amat Gil,Evan Smothers,Dan Iter,Mark Thornburg,Dawit Mureja Argaw,Sourabh Medapati,Thea Wang,Ria Verma,Qiuling Xu,Harshit Syal,Ehsan Gholami,Moumita Bhattacharya,Ashish Rastogi
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted to NeurIPS 2026 PALM Workshop
Abstract:Large-scale personalization is shifting from embedding-table retrieval to LLM-based recommenders that generate the Semantic ID (SID) of the item to recommend, where each SID is a short sequence of discrete hierarchical content codes. A single instruction-tuned model can then unify retrieval, content understanding, and language, and recommend brand-new titles from their content alone using SID; then, how should the model generate the SID? Traditionally, models rely on a single language modeling head that generates tokens autoregressively, typically using constrained beam search. In contrast, an emerging body of work explores multi-token prediction (MTP), where dedicated heads predict all tokens in parallel, substantially reducing the decoding latency associated with beam search. We report an industry-scale, apples-to-apples comparison of these two generation paradigms: single-head autoregressive decoding vs. MTP. We introduce a new auxiliary ranking loss, defined over the joint SID score, to train the MTP objective. We evaluate the SID training objectives on both quality and training/inference cost, inside a multi-task LLM that jointly does personalized retrieval, content question-answering (QA), and instruction following over a large streaming-media catalog. Our experiments show single-head beam search consistently beats MTP on most product-critical metrics: ranking and content QA, and trains at roughly 2.6x lower dollar cost on Qwen3-8B scale; MTP wins only on raw inference throughput. Around this finding, we detail an efficient training and inference stack and report end-to-end gains over the previous non-SID baseline equipped with an auxiliary ranking head: +77% item cold-start ranking, +22% content-QA accuracy, +2.8x faster training, with language metric regression traced to SID vocabulary expansion. We address the language regression by reinforcement learning post training.
[IR-33] KhyontekAI at CheckThat! 2026: Multilingual Dense Retrieval and Cross-Encoder Reranking for Scientific Web Claims
链接: https://arxiv.org/abs/2610.15837
作者: Bhargab Jyoti Bhuyan,Pritam Deka
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL)
备注: 14 pages, 1 figure, 9 tables. Author version of the paper published in CEUR Workshop Proceedings, Vol. 4283 (CLEF 2026 Working Notes, Jena, Germany)
Abstract:Scientific claims shared through social media platforms frequently reference research findings without explicitly linking to the corresponding scientific publications. Automatically retrieving the correct scientific source for such claims is a challenging multilingual retrieval problem involving semantic matching, paraphrasing, noisy social media text, and cross-domain language variation. In this work, we present a multilingual two-stage retrieval framework for CLEF 2026 CheckThat! Task 1 on Source Retrieval for Scientific Web Claims. Our system combines fine-tuned dense retrieval models with cross-encoder reranking for improving multilingual scientific document retrieval. We evaluate multiple transformer-based dense retrieval architectures including multilingual-E5, BGE, and Contriever under different metadata configurations. We further investigate multilingual joint training, retrieval depth analysis, metadata enrichment, and language balancing strategies for improving retrieval effectiveness. Experimental results demonstrate that metadata enrichment consistently improves retrieval effectiveness, while fine-tuned cross-encoder reranking provides additional gains in retrieval precision. Our final multilingual system achieves MRR@5 scores of 0.6222 for English, 0.5112 for German, and 0.6254 for French, with an overall average MRR@5 score of 0.5862.
[IR-34] Local Citation Recommendation Without Graph Signals AACL
链接: https://arxiv.org/abs/2610.15832
作者: Benjamin Basseri,Pavlos Protopapas,Kemeng Zhang,Alberto Accomazzi,Stephanie Jarmak,Anna Kelbert
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL); Digital Libraries (cs.DL)
备注: Accepted at WASP 2026, co-located with AACL-IJCNLP 2026. 16 pages, 3 figures, 2 tables. Dataset: this https URL Code: this https URL
Abstract:Resolving correct citations can be a time-consuming task for research writers. We address local citation recommendation (LCR), where the goal is to retrieve appropriate references for a given sentence. Web tools generally depend on the existing citation graph, which may bias results toward citation count or more recent publications. LCR research using full paper texts has been limited by the lack of publicly available data. We present a new dataset, the Local Citation Recommendation Corpus (LCRC), containing 14,735 scientific statements derived from astrophysics review papers, each with one or more citation targets whose full text is accessible via API. Using this dataset we systematically compare retrieval methods and develop a citation recommendation system that avoids citation graph bias while achieving 8.0% improvement in Recall@25 and 6.9% improvement in NDCG@25 over dense retrieval baselines without model fine-tuning. To support reproducibility, we make our code and dataset publicly available.
[IR-35] Rethinking Evidence Modeling: Context-Coupled Graph Learning for Multimodal Recommendation
链接: https://arxiv.org/abs/2610.15523
作者: Mingqiao Zhang,Qiyao Peng,Yumeng Wang,Wenzhe Dong,Hongtao Liu,Jian Yang
类目: Information Retrieval (cs.IR)
备注:
Abstract:Multimodal graph recommendation integrates observed interactions with item images and text to learn user preferences. These inputs constitute Semantic-Behavioral Evidence (SBE). Many existing graph methods incorporate collaborative and multimodal semantic relations, yet their aggregation and state updates remain context-separated, limiting the continuous composition of preference evidence across layers and the joint use of multiple sources. To address this limitation, we propose CC-MGRec, a context-coupled graph learning method that organizes SBE within a common graph learning context for cross-relation modeling and compositional propagation. Its context-coupled multimodal graph integrates user-item collaborative relations and item-item semantic relations, coordinating their message scales through joint normalization. Recommendation-grounded semantic calibration uses training behavior to adjust semantic neighbor weights while preserving normalized collaborative propagation coefficients. Collaborative-semantic compositional propagation updates shared node states layer by layer to continuously compose collaborative and multimodal semantic evidence, implicitly forming Preference Evidence Chains (PECs), including collaborative-semantic and visual-textual PECs. Joint execution of relation messages on the same shared node states within each layer preserves propagation outputs while reducing intermediate output processing and fusion overhead. We provide theoretical analysis of the proposed mechanisms. Experiments on three Amazon datasets demonstrate improved recommendation quality and faster inference during graph propagation.
[IR-36] Pathwise Alignment for Collaborative Filtering: Bilateral Conditional Flow Regularization
链接: https://arxiv.org/abs/2610.15506
作者: Zhiwei Li,Guodong Long,Chengqi Zhang,Jing Jiang
类目: Information Retrieval (cs.IR)
备注: 25 pages, 13 figures. Includes appendices
Abstract:Collaborative filtering predicts user-item interactions from implicit feedback. We focus on matrix factorization (MF) trained with the Bayesian Personalized Ranking loss. This loss updates each user embedding only through that user’s observed items and sampled negatives, so sparse feedback gives MF few constraints. To constrain each user through other users, we propose Pathwise Alignment for Collaborative Filtering (PACF), which trains MF with one network shared by all training pairs. At points between user and item embeddings, a flow matching term trains the network to predict the user-to-item step, reversed and scaled for sampled negatives. A trajectory consistency term pulls each user’s predicted path to end at the observed item. Both terms update the embeddings, so the network is discarded after training, thereby maintaining the efficiency of MF at inference. We prove that a network with bounded sensitivity to its input point accurately fits users who share an item only if their embeddings are close, so these users constrain one another. Against eleven discriminative and generative baselines, PACF achieves the highest performance on all seven datasets. Its relative gains over MF are largest where feedback is sparsest. The shared network thus supplies the missing constraints through other users.
[IR-37] Federated Semantic Residuals for Multimodal Recommendation
链接: https://arxiv.org/abs/2610.15505
作者: Zhiwei Li,Guodong Long,Chengqi Zhang,Jing Jiang
类目: Information Retrieval (cs.IR)
备注: 23 pages, 14 figures. Includes appendices
Abstract:Federated Recommendation Systems (FedRSs) learn collaborative scores from interaction histories on device and rank items by them, so a higher score means a higher rank. Item text and images add semantic information that can reorder the list. Existing Federated Multimodal Recommenders (FedMRs) fuse this content into their models, which can obscure how content reorders the list. We therefore propose the SEmantic Residual Adapter (SERA), which keeps content out of an existing FedRS and adds semantic residual scores to its collaborative scores. Content can then lift an item above another only when its semantic residual score exceeds the other’s by more than their collaborative score gap. This score grows with the item’s content similarity to the user’s interaction history and is capped at a residual bound. We prove that for any content, an item can enter the list only if its collaborative score is within twice this bound of the lowest one in the list. A larger bound lets content lift more items, including irrelevant ones, so the bound trades off how far content reaches against accuracy. With a validated bound, SERA improves all three FedRSs on all six datasets without adding personalized parameters. It outperforms FedMRs in most comparisons and lifts some FedRSs above centralized baselines.
[IR-38] oward Explainability in Multimodal Recommendation
链接: https://arxiv.org/abs/2610.15484
作者: Chen Liang,Jinghui Li,Xiuze Zhou,Zongyue Wang,Yuanguo Lin,Han Yu
类目: Information Retrieval (cs.IR)
备注: 19 pages, 2 figures, 4 tables
Abstract:Multimodal recommender systems effectively alleviate data sparsity and cold-start problems by integrating heterogeneous information, yet complex cross-modal representations and model decision-making processes also introduce new interpretability challenges. Despite the rapid growth of research on the interpretability of multimodal recommendation in recent years, relevant methods and evaluation criteria remain relatively fragmented, lacking a unified conceptual framework. To address this gap, this paper categorizes existing methods into modality explanation and post-hoc explanation based on the timing and manner of explanation generation, and further constructs a progressive system for each category. Building on this foundation, this paper synthesizes four categories of evaluation metrics from both objective quantitative and subjective perceptual perspectives, and reviews relevant datasets along the dimensions of application domain and data source. Through a comprehensive analysis of existing research, we further propose future research directions. This paper aims to establish a unified analytical perspective across multimodal representation, recommendation decision-making, and user preferences, providing a research foundation for building transparent and trustworthy multimodal recommender systems.
[IR-39] Understanding ChatGPT Citations: Evidence from Production Web Search Across Model Generations
链接: https://arxiv.org/abs/2610.15382
作者: Yunxuan Jiang,Shiqi Zhang,Anqi Xing,Xiangyu Chang,Xiaokui Xiao
类目: Information Retrieval (cs.IR)
备注: 10 pages, 5 figures
Abstract:Generative engines such as ChatGPT Web retrieve webpages to answer user queries, but only a subset are shown as citations. Understanding this retrieval-to-citation process is important for generative engine optimization (GEO), which aims to improve content visibility in generative-engine responses. However, prior studies mainly analyze final responses, leaving the internal execution process and its variation across model settings less understood. To this end, we conduct an observational study of four ChatGPT Web settings from ChatGPT 5.2 to 5.5, covering Instant and Thinking modes at the URL, hostname, and entity levels. We examine how ChatGPT rewrites queries, retrieves candidate webpages, and selects references. The Instant settings remain mostly single-round, while GPT-5.4 Thinking frequently performs multiple retrieval rounds. Across repeated runs, candidates retrieved more often are also more likely to be cited. We then evaluate Google Search and the OpenAI Responses API as external proxies. Responses API candidates provide broader coverage, while its citations more closely match sources cited by ChatGPT Web. Both proxies match ChatGPT Web better at the hostname and entity levels than at the exact-URL level. Even when combined, they recover only about one quarter of cited URLs. We further test whether these external signals remain predictive when combined. Within the external candidate set, a supervised model combining them reaches AUCs above 0.74 for GPT-5.3 and GPT-5.5 at all three levels. These findings suggest that GEO should distinguish retrieval visibility from citation visibility, account for rewritten queries and run-to-run variation, and consider hostname- and entity-level presence in addition to exact URLs. Differences across settings also suggest that GEO strategies should be re-evaluated when the deployed setting or serving environment changes.
[IR-40] Retrieving Complementary Evidence for Multi-Hop RAG with Semantic Gain Graphs
链接: https://arxiv.org/abs/2610.15285
作者: Zeliang Li,Xiaofen Xing,Kailing Guo,Xiangmin Xu
类目: Information Retrieval (cs.IR)
备注:
Abstract:Multi-hop retrieval-augmented generation (RAG) requires finding documents whose evidence is useful in combination. Query relevance and graph connectivity help locate related content, but neither directly measures the additional information one document contributes given another. We introduce SG ^2 RAG, a framework that represents this complementarity through semantic gain. Offline, a large language model assesses the relevant and nonredundant information contributed by one chunk given another within neighborhoods proposed by semantic similarity. These ordered assessments are aggregated into symmetric edge weights to construct a reusable Semantic Gain Graph. Online, controlled Personalized PageRank starts from query-relevant seeds and ranks complementary evidence beyond the initial retrieval set. Experiments on MuSiQue, 2Wiki, and HotpotQA with two language model backbones show that SG ^2 RAG achieves the best average EM, F1, and Recall@10 among the evaluated single-retrieval methods. Under identical candidate support and propagation, semantic gain improves average F1 and Recall@10 over similarity weighting by 4.08 and 4.31 points. Analyses of supporting documents and improvements in three iterative retrieval frameworks further support its effectiveness as a reusable retrieval component.
[IR-41] PANINI-DR: Controlled Deep Research through Generative Semantic Workspaces ICLR2027
链接: https://arxiv.org/abs/2610.15208
作者: M. Yigit Turali,Shreyas Rajesh,M. Efe Lorasdagi,Chenda Duan,Kartik Sharma,Vwani Roychowdhury
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: Under review at ICLR 2027. 37 pages, 1 figure, 22 tables. M. Yigit Turali and Shreyas Rajesh contributed equally
Abstract:Deep research requires connecting multiple steps of search, reasoning, and computation, with intermediate findings guiding what to investigate next. As investigations grow, agents must maintain research state across model calls and check findings before errors propagate into subsequent work. We introduce PANINI-DR, a multi-agent harness that separates coordination of the overall investigation from the research and verification of individual subquestions. A shared workspace, built on the Generative Semantic Workspace representation, records subquestions, their dependencies, findings, and supporting evidence. Researchers investigate individual subquestions, while a separate verification agent checks proposed answers against those subquestions and their evidence. Verification feedback guides further investigation, and recorded findings inform subsequent assignments. We evaluate PANINI-DR on BrowseComp-Plus, FRAMES, and MoNaCo against seven state-of-the-art deep-research harnesses, all using a shared language-model backbone and frozen corpora. PANINI-DR achieves the strongest average performance across the three benchmarks. On BrowseComp-Plus, it achieves 73.3% accuracy, outperforming the next-best harness by 16.0 points. Its advantage is strongest on hard questions, where it answers nearly twice as many as the next-best harness. Ablations further show that structured planning and persistent state improve performance, as does checking each finding against its cited evidence before it is reused. Because a model can answer correctly from memory alone, we also analyze the evidence retained by each harness. PANINI-DR achieves the highest answer recoverability from saved evidence on all three benchmarks, while complementary claim-support audits show stronger evidence-backed answers. Accounting for this evidence changes how harnesses compare and further strengthens PANINI-DR’s advantage.
[IR-42] MoSAIC: Learning DAG Scheduling Heuristics from Optimal Schedules through Motif-Aware Recovery and LLM -Guided Reasoning
链接: https://arxiv.org/abs/2610.15205
作者: Shiva Ahir,Hashmath Shaik,Alex Doboli,Nadeem Yaseen,Simona Doboli
类目: Hardware Architecture (cs.AR); Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注:
Abstract:Efficient task scheduling in heterogeneous computing systems remains a fundamental challenge in embedded and high-level synthesis flows. This paper presents MoSAIC (Motif-aware Scheduling with AI-driven Cognition), a learning-based framework that bridges the gap between exact and heuristic scheduling approaches for DAG-based applications. The system uses the reasoning capabilities of large language models (LLMs) to learn the relationships between DAG structures and scheduling priority functions. The framework first leverages the OR-Tools CP-SAT solver [4] to generate optimal schedules, which are then used to supervise the learning of compact priority-based scheduling functions through Bayesian Optimization. These learned heuristics capture relationships between scheduling decisions and structural graph features such as rank, depth, and fanout, enabling the construction of a reusable knowledge base across diverse DAG instances. Experimental results across synthetic graph families and real-world benchmarks, including PolyBench and MachSuite, show that the learned heuristics consistently improve over classical approaches such as HEFT [2], while maintaining low runtime overhead. The results indicate that combining optimal supervision with learning-based generalization provides a practical approach for scalable and adaptive task scheduling.
[IR-43] Faithful Query Resolution for Conversational Search
链接: https://arxiv.org/abs/2610.15034
作者: Saber Zerhoudi,Ivica Kostric,Jelena Mitrovic
类目: Information Retrieval (cs.IR)
备注:
Abstract:Conversational query rewriting makes context-dependent requests explicit for retrieval, but additional context can introduce an information need the user did not express. This raises a question about the interaction between query fidelity and retrieval breadth. This paper examines that interaction in behavior-adaptive conversational search, using iKAT SCAI 2026 as the evaluation setting. We compare five query-generation configurations within a behavior-adaptive assistant using BM25 and SPLADE retrieval. A minimal-faithful resolver preserves the user’s wording and resolves only context-dependent references; weighted reciprocal rank fusion combines its rankings with those from a learned rewriter and the raw utterance. Offline evaluation shows that adding these formulations recovers recall over the faithful query alone, with a slight reduction in nDCG@10. Across our five iKAT SCAI 2026 submissions, the weighted configuration obtains the highest mean retrieval scores, reaching nDCG@5 of 0.534, a 5.3% relative improvement over the strongest alternative. However, this retrieval advantage does not coincide with the highest conversational satisfaction scores. These findings motivate separating faithful query interpretation from retrieval expansion, while assessing their effects on conversational utility independently.
[IR-44] What Drives AI Citations? Content Retrieval and Model Knowledge in Generative Engine Optimization (GEO)
链接: https://arxiv.org/abs/2610.15033
作者: Mahe Chen,Nick Koudas
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL); Machine Learning (cs.LG); Social and Information Networks (cs.SI)
备注: 11 pages
Abstract:Generative search engines such as ChatGPT, Gemini and Perplexity now decide which Web pages users see as sources, and a growing industry of generative engine optimization advises publishers on how to earn these citations. We study what shapes citation in two 90-day collections of answers to commercial questions in contrasting markets: 158,099 answers with 468,037 citations. Comparing pages of the same website, content matters in one way: pages whose title and paragraphs match the questions people ask are cited more, whereas FAQ blocks, structured data, summary boxes and page speed appear effective only across websites. Page properties nevertheless explain little of the variation between pages of a website. Replaying the questions through the engines’ APIs shows why. Every page ChatGPT cites is one its own search retrieved, and which pages that search retrieves, for sub-queries the model rewrites on every answer and that rarely coincide with Google’s or Bing’s results, explains far more of the variation than page properties (32% against 2.8%). Content acts mainly through this step, by matching the engine’s queries in meaning. The model’s own knowledge shapes it too: ChatGPT writes brands it knows into its searches, prefers websites it knows, and names brands without any supporting page, and controlled experiments show that brand names reorder its rankings of products with identical facts. Pages are thus cited when searched; brands are also named when remembered. The findings replicate in a market of small B2B vendors. We release data and code.
[IR-45] Data Curation and Model Souping: Unlocking Qwen 3-0.6B for MS MARCO Passage Ranking
链接: https://arxiv.org/abs/2610.14816
作者: Shuguang Han
类目: Information Retrieval (cs.IR)
备注: 12 pages, 9 tables
Abstract:This report studies passage retrieval and re-ranking on the MS MARCO benchmark using lightweight Qwen3-0.6B models. Across three experimental settings — BM25-then-reranking task, corpus-wide dense retrieval task, and dense retrieval then re-ranking task — the results demonstrate that competitive ranking performance can be achieved without complex architectural modifications, custom loss functions, or cross-encoder distillation. Instead, substantial gains stem from data-centric strategies: corpus-targeted continued pre-training, rigorous data curation, hard-negative mining, training–inference distribution alignment, and weight merging via Model Soup. In particular, Model Soup delivers marked performance gains while preserving the inference efficiency of a single model. These findings highlight how data curation and weight merging can push sub-billion-parameter models to state-of-the-art performance.
[IR-46] Self-improving data synthesis can scale embedding models across domains and tasks
链接: https://arxiv.org/abs/2610.14730
作者: Zucan Lyu,Rui Tang,Jianlyu Chen,Peisheng Jin,Zheng Liu,Defu Lian
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 30 pages, 6 figures; supplementary information included (27 pages). Code: this https URL Data: this https URL
Abstract:Adapting embedding models to new tasks requires training data that reflect task-specific information needs and relevance criteria. However, manual data design is costly and difficult to scale. Here we introduce SYNTRA, a novel retrieval data synthesis method whose self-improvement loop iteratively refines reusable synthesis instructions using feedback on trial outputs. This process requires only an unlabelled corpus and a few relevant query-document pairs. Experiments across 24 retrieval datasets spanning 15 domains demonstrate the effectiveness of the synthesized data. These data also improve widely used, strong embedding models on tasks with and without existing labelled training sets. Self-improvement enhances retrieval on all six evaluated tasks, outperforming human calibration on four. Moreover, retrieval performance continues to improve on MS MARCO as we scale the training dataset to 32 times its initial size. Overall, with minimal human intervention, SYNTRA enables scalable adaptation of embedding models across domains and tasks.
[IR-47] he Company a Question Keeps: Auditing Open Jev-Style Models for Retrieval
链接: https://arxiv.org/abs/2610.14631
作者: Jianbo Chen
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL)
备注: 15 pages, 2 figures, 14 tables
Abstract:Jev’s documented interface evaluates typed questions independently against shared evidence. We audit request independence in seven open Jev-style checkpoints on 276 constructed three-passage states from 92 TREC retrieval queries. Paired interventions separately change requested questions, displayed evidence, and candidate eligibility. Request expansion induces strict binary flips in 9.1%/8.0% of Nimble’s judgments and changes 21.4%/18.9% of selection-or-abstention actions in TREC DL2019/2020 under a threshold of 0.5 fixed before inference. Correct-action rates change by 0.0/+8.2 percentage points, respectively, with 95% query-bootstrap intervals of [-6.0,6.0] and [3.8,12.6]. Field masking traces all 70 request-set flips to the schema fields serialized into each scoring prompt. In Von, fixing a shared separator’s position eliminates two eligibility-induced preference flips in selected diagnostic states. Laya-typed’s default CUDA path produces five passage swaps without strict binary flips, and both yearly action-change intervals include zero; in an FP32 rerun, its probability movement vanishes at the model’s four-decimal reporting resolution.
[IR-48] Decoupling Knowledge Forgetting and Refusal Learning in MLLM Unlearning
链接: https://arxiv.org/abs/2610.14605
作者: Haodong Chen,Shengtao Wen,Dong Liang,Lingbing Guo,Xiang Chen
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注:
Abstract:Current evaluations of multimodal unlearning typically focus on whether a model stops outputting target information. This surface-level criterion, however, masks two critical failure modes. First, knowledge suppression techniques often induce confabulation, leading the model to hallucinate facts about the unlearned subject. Second, refusal training merely alters observable behavior without erasing the underlying knowledge; such models might decline open queries but still identify correct answers under multiple-choice probing. To overcome these limitations, we propose Forget-Aware Self-Distillation (FASD). This method is designed to achieve honest ignorance, a state where the model genuinely loses the target knowledge and learns to acknowledge its absence. FASD separates these goals into two complementary channels. The behavior channel distills appropriate refusals from an ignorance-prompted copy of the model itself. Crucially, target tokens are masked from the teacher distribution to avoid accidentally reteaching the forgotten concepts. Simultaneously, the knowledge channel directly suppresses ground-truth answer spans independently of the model’s generated responses. Experiments demonstrate that FASD successfully unlearns target knowledge, prevents information leakage in open-ended tasks, and preserves performance on retained data. The resulting model not only forgets the specified information but accurately recognizes the boundaries of its own knowledge. Code and data will be released upon publication.
[IR-49] Fine-Tuning Strategy Not Just Data: A Statistically Rigorous Audit of Gender and Stereotypical Bias in Embedding Models for HR Recommendation Systems RECSYS
链接: https://arxiv.org/abs/2610.14579
作者: Ekta Yadav
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL); Computers and Society (cs.CY); Machine Learning (cs.LG)
备注: 8 pages, 4 figures, 5 tables. Accepted at RecSys in HR '26 (The 6th Workshop on Recommender Systems for Human Resources)
Abstract:Embedding models underpin modern job-candidate matching: resumes, job descriptions, and queries are encoded as dense vectors, and recommendations come from nearest-neighbor retrieval over that space. Any social bias in the embeddings therefore feeds directly into HR recommendations. We audit six multilingual sentence-embedding models (LaBSE- and E5-family, both public and custom fine-tuned) using Bias-in-Bios (occupational gender bias) and StereoSet (stereotypes across profession, race, gender, and religion). We treat statistical validity as a first-class concern: we identify a non-independence error in a naive significance test, where pooling non-independent pairwise similarities inflated a trivial 0.0046 mean difference to p \approx 0 , and replace it with a profession-level paired Wilcoxon signed-rank test, individual-resampling bootstrap confidence intervals, and permutation tests for StereoSet. Under the corrected methodology, knowledge-distillation (KD) fine-tuning of E5-Base significantly increases measured bias on Bias-in-Bios ( \Delta = -0.177 , p 10^-8 ; more negative means more bias), although this measure is sensitive to global embedding geometry. The KD result is corroborated by pro-stereotype patterns in the StereoSet profession category (uncorrected p 0.05 ). Binary-classification fine-tuning of the same base model shows no statistically distinguishable change ( p = 0.218 ). Fine-tuning strategy is thus an actionable design lever for practitioners building HR recommenders. We also report which benchmark categories are underpowered, and note that two domain-adapted models could not be included in significance testing due to a software-access constraint, which limits our conclusions.
[IR-50] LLM -based User-Product Relevance Evaluation for Dynamic Product Advertising CIKM2026
链接: https://arxiv.org/abs/2610.14550
作者: Wen Zhang,Xiao Bai,Hsiang-wei Chao,Mengfan Zhu,Tong Zhao,Mingxuan Ju,Xiaodong Liu,Congfei Zhang,Yunzhi Zhou,Yajun Wang,Jinchao Li,Yu Zhang
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 8 pages, 2 figures, CIKM 2026
Abstract:Dynamic Product Advertising (DPA) retrieval systems are a core component of modern e-commerce advertising; however, their evaluation remains challenging. Conventional offline engagement metrics are limited by sparse, biased feedback, while online evaluation is costly and time-consuming. In this work, we propose a user–product relevance metric for offline evaluation of DPA retrieval systems. The metric first compresses the engagement history into a compact set of intent representatives via Semantic-ID-grouped, event-stratified segmentation, then assesses the relevance of a recommended product to each individual intent and aggregates them as a user-level relevance score, yielding a stable and interpretable measure of retrieval quality. To compute this metric at scale, we leverage large language models (LLMs) as semantic relevance judges, capturing nuanced functional and semantic relationships that heuristic methods often miss. We propose an agentic, human-aligned prompting framework that generates multi-category relevance guidelines and distills teacher-model judgments into an efficient student model for scalable evaluation. The resulting metric strongly correlates with offline and online business metrics, enabling faster and more reliable DPA retrieval iteration before A/B testing.
[IR-51] A Stateful Research Information Infrastructure for Scientific Discovery
链接: https://arxiv.org/abs/2610.14538
作者: Lingyuan Kong,Zhengran Wu
类目: Digital Libraries (cs.DL); Information Retrieval (cs.IR)
备注: Project Website: this https URL
Abstract:Artificial intelligence (AI) is being applied to an expanding range of scientific activities, including literature analysis, hypothesis generation, software development and automated experimentation. Yet these capabilities rely on a more fundamental condition: scientific information must remain identifiable and traceable across systems and over time. In practice, different versions of a work and their associated records and files are often scattered across services and local directories. This fragmentation forces researchers to identify and contextualize the same work repeatedly, disrupting continuity from discovery to interpretation and reuse. Here we present Gewu Insight, a research information infrastructure that addresses this discontinuity by organizing scientific works as persistent, stateful research objects. It unifies venue-, topic- and author-based discovery records through a shared paper identity. Within a local library, metadata, versioned resources and research context can then accumulate around the same work. These capabilities are integrated into an iterative workflow for scientist-led, AI-assisted research, allowing earlier interpretations to be revisited and refined. A bounded evaluation showed high candidate precision across the three discovery modes, with paper identity, resource distinctions and research state preserved in controlled workflows. By making continuity an explicit property of research infrastructure, Gewu Insight can serve as an inspectable scientific memory layer shared by researchers and AI assistants, supporting human judgment and providing a foundation for future, more epistemically grounded scientific AI.
[IR-52] EVISTA: Unified Multimodal Video Search with Evidence-Aware Temporal Path Optimization
链接: https://arxiv.org/abs/2610.14510
作者: Minh-Huy Tran,Van-Thinh Vo,Minh-Khoi Nguyen,Duy-Tan Nguyen,Anh-Quan Nguyen-Tran,Loi Nguyen Khanh
类目: Information Retrieval (cs.IR)
备注: 15 pages, 6 figures
Abstract:In the era of big data, precise event retrieval from large video collections has become increasingly important and challenging. In particular, locating multiple events from the same video in their temporal order is difficult when different events are described by different types of evidence, such as visual content, on-screen text, and spoken words. We develop EVISTA, a unified multimodal video retrieval system for the Ho Chi Minh City AI Challenge 2026, which integrates visual-semantic, OCR, ASR, object, and metadata search over a shared keyframe index, and propose TEMPO-P, an evidence-aware temporal path optimization method for ordered multi-event retrieval. Building EVISTA for this task reveals two limitations of independent event retrieval: a relevant video may be missed when one event is weakly retrieved, and independently retrieved event candidates may be temporally inconsistent or unable to represent events defined primarily by text or speech. TEMPO-P addresses these limitations by aggregating candidate videos across events and modalities, jointly scoring keyframes using visual, OCR, and ASR evidence, and selecting a compact temporally ordered path within each video through dynamic programming. On five organizer queries and 36 additional multi-event queries, TEMPO-P matches visual-only temporal alignment on visually described events (R@1 0.42 for both) and retrieves queries containing text-defined events, which visual-only alignment cannot represent (R@1 0.38). TEMPO-P is integrated into EVISTA as its ordered-event retrieval mode.
[IR-53] RES-MR: Risk-Aware Reasoning for Explainable and Safe Medication Recommendation
链接: https://arxiv.org/abs/2610.14474
作者: Cong Wang,Jin Li,Shoujin Wang,Yishuo Li,Huilin Gu,Wenpeng Lu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: 12 pages
Abstract:Medication recommendations seek to deliver personalized drug combinations based on patients’ clinical information, where ensuring transparent decision-making and personalized safety are of paramount importance. However, existing approaches predominantly operate as ``black boxes’', offering limited interpretability and hindering clinician trust. Furthermore, they typically rely on fixed, patient-agnostic safety constraints, neglecting the heterogeneity of risk tolerance across individuals, which poses significant risks to vulnerable populations. To address these gaps, we propose \textbfRES-MR, a novel Risk-aware Reasoning framework for Explainable and Safe Medication Recommendation using LLMs. RES-MR follows a two-stage training paradigm: (1) an explainable clinical reasoning distillation stage to elicit diagnostic capabilities by fine-tuning on synthesized reasoning trajectories, grounded in patient-specific knowledge from graph retrieval; and (2) a risk-aware policy optimization stage to dynamically balance therapeutic efficacy with personalized safety. This stage employs disentangled drug factors to construct patient-specific reference prototypes that capture desired therapeutic drug features and risk features to avoid. These prototypes enable risk boundary calibration and guide the model toward safe recommendations via reward shaping and clipping. Extensive experiments on benchmark datasets demonstrate that RES-MR significantly outperforms state-of-the-art baselines in terms of accuracy, safety, and explainability. Our code and data are available at: this https URL.
[IR-54] Co-adaptive Rule Routing and Constraining for Structured Knowledge Retrieval ICLR
链接: https://arxiv.org/abs/2610.14465
作者: Zhisheng Qi,Haoyu Han,Jiaxi Li,Bing He,Huiyuan Chen,Franck Dernoncourt,Ryan A. Rossi,Mahantesh M. Halappanavar,Jin Lu,Yu Zhang,Tianming Liu,Yu Wang
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
备注: Under review at ICLR
Abstract:Structured knowledge retrieval acquires task-relevant evidence from structured knowledge sources for downstream tasks. Existing methods rely on unconstrained retrieval or predefined query-rule supervision, such as author-paper rules for scientific discovery and anatomy-drug rules for precision medicine. Such supervision is rarely available in practice. We propose Co-Adaptive Rule Routing and Constraining Retrieval (CoRE), which jointly learns which reasoning rule to assign to each query and how to retrieve evidence under that rule, without predefined query-rule pairs. A routing agent selects a rule, and its corresponding retriever traverses the knowledge base and ranks evidence that satisfies the rule’s constraints. Because routing and retrieval are interdependent, CoRE alternates between reinforcement learning for the routing policy and contrastive learning for the rule-constrained retriever. The selected rules provide supervision for retriever training, while retrieval performance rewards the routing policy. This closed-loop process enables both components to co-adapt without explicit query-rule supervision. Experiments on STaRK-QA datasets across scientific discovery, biomedical reasoning, and e-commerce knowledge management demonstrate CoRE’s effectiveness. Ablations show the benefit of co-adapting the router and retriever. Generation-based routing transfers across datasets and benefits from multi-domain pretraining followed by target-specific fine-tuning. Analysis of rule usage across queries also informs automatic rule refinement and discovery.
[IR-55] Understanding and Mitigating Training Instability in Large-Scale Recommender Systems: A Cascade Perspective
链接: https://arxiv.org/abs/2610.14394
作者: Yuening Li,Zeyu Wang,Zhen Zhang,Jiayi Liu,Zhimeng Jiang,Yuji Roh,Weikai Xu,Mingyan Gao,Onkar Dalal,Chuan He,Liang Liu
类目: Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注:
Abstract:Training large-scale recommender systems on dynamic industrial platforms, such as Short-Form Video (SFV) services with billions of daily interactions, faces critical stability challenges. Rapid distribution shifts in user data and increasing model complexity frequently trigger a divergence cascade, typically initiating with gradient explosion, propagating to weight explosion, and culminating in loss explosion. This G-W-L cascade severely impedes model convergence and performance. This paper provides a systematic analysis of this G-W-L cascade within an industrial SFV recommender context. We then propose and evaluate a structured mitigation framework where interventions are explicitly mapped to counteract specific stages of the cascade: gradient (G) control, weight/activation (W) stabilization, and loss (L) shaping. Our contribution provides a structured framework for understanding and tackling training divergence, offering practical insights for deploying robust, continuously updated recommenders in large-scale, non-stationary settings. We empirically validate the effectiveness of these strategies on a major industrial SFV recommendation system.
[IR-56] Simple Linear Corrections for Compression in Late Interaction
链接: https://arxiv.org/abs/2610.14196
作者: Mihir Agarwal,Sabyasachi Basu,Viraj Agrawal
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Late-interaction (multi-vector) models represent documents using a set of token-level vectors, providing a middle ground between cross-encoders and dense encoders in terms of quality and cost of embedding. However, compared to the latter, the indices are costly to store and score. Pooling and pruning based compression techniques are deployed widely but deteriorate in quality noticeably with aggressive compression. We investigate whether a lightweight change to the scoring function can recover retrieval quality after document-vector compression. For each combination of encoder, compressor, and vector budget, we learn a global linear correction by distilling the uncompressed model’s score distributions over candidate documents. The encoder and compressor remain frozen, and the correction can be applied to each query at search time or once, offline, to the stored vectors. We evaluate the method with ColBERTv2 and Jina v4 on MS MARCO and BEIR, and with ColPali2 on the ViDoRe v3 visual document retrieval benchmark. Experiments cover four training-free compression methods, multiple vector budgets, in-domain adaptation, and transfer from MS MARCO to BEIR. In the in-domain setting, the correction improves over baseline compression in 154 of 160 BEIR settings, while transfer from MS MARCO is more variable across models and datasets. We also find that matrices trained with synthetic queries can improve over baseline compression on ViDoRe v3. These results show that a small learned correction can recover some of the quality lost to compression without retraining the encoder or recompressing the documents. Subjects: Information Retrieval (cs.IR); Computation and Language (cs.CL); Machine Learning (cs.LG) Cite as: arXiv:2610.14196 [cs.IR] (or arXiv:2610.14196v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2610.14196 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[IR-57] Do you “see” what I “see”? A Multi-panel Visual Question Answer Dataset for Large Language Model Chart Analysis
链接: https://arxiv.org/abs/2610.14173
作者: Daniel J. Evans,J. P. Naiman,JooYoung Seo,J. Stephen Downie
类目: Digital Libraries (cs.DL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: 16 pages, 4 figures, accepted to The 30th International Conference on Theory and Practice of Digital Libraries, 2026
Abstract:Digital libraries increasingly host scientific publications whose figures encode primary research findings, yet making this visual content searchable, accessible, and reusable remains an open challenge. Large language and multimodal models (LLMs/LMMs) provide the possibility to automate many figure parsing tasks such as determining data trends or extracting data from plots. In this work, we consider if these models can reliably support task-oriented visual data analysis when presented with synthetic, but realistic, scientific figures that contain multiple charts. Our Visual Question Answering (VQA) process scaffolds our tasks along a four-level description model from basic conceptual answers to interpretive and analytical understanding. Additionally, we outline our process for synthetically generating our single and multi-panel (charts in which there are multiple panels present) plots and testing the data with various LMMs using quantitative analysis. In general, we find LMMs become less accurate with increasing figure complexity as the number of panels, number of plot elements, and difficulty of questions increase. Finally, because LMM accuracy is expected to change rapidly, our analysis also considers cost relative to performance as a metric to inform future user decisions. We provide the detailed question-answer pairs used in this analysis, along with the code used to generate random plots, question-answer pairs and detailed bounding-box metadata for chart elements. The code and data are available at this https URL and this https URL_qa_multipanel
[IR-58] An Edge-First Multimodal Asset Management Studio and Retrieval Engine for Heterogeneous AI-Generated Media
链接: https://arxiv.org/abs/2610.14046
作者: Chaoyu Lin
类目: Multimedia (cs.MM); Human-Computer Interaction (cs.HC); Information Retrieval (cs.IR)
备注: 8 pages, 1 figure, 4 tables. Source code and reproducibility artifacts are available at this https URL
Abstract:AI-generated image collections combine visual content with heterogeneous provenance such as prompts, model parameters, and workflow graphs. Managing these collections locally requires coordinated metadata extraction, responsive browsing, and retrieval without mandatory remote inference. We present Berry AI Studio, an edge-first asset management system built with Vue, Tauri, Rust, SQLite, and ONNX Runtime. Berry integrates format-specific metadata parsing with raw-payload retention, generation-aware cancellation of queued thumbnail work, a persistent multi-tier thumbnail manifest, cursor pagination, and local image-text retrieval. We evaluate these mechanisms in five independent repetitions on a pinned source revision. With 50,000 catalog records, the median warm first-page database call was 285.3 ms; at offset 40,000, keyset pagination reduced median raw-query latency from 57.9 to 0.46 ms. A controlled browser harness recorded a 0.3 ms median-of-run P95 update time during virtualized scrolling, while cancellation prevented all 1,200 stale speculative decodes in the tested trace. A versioned 10-case corpus produced 53/53 exact metadata-field matches, and controlled retrieval and grouping tests exercised the authentic storage and clustering paths. The implementation, benchmark scripts, raw logs, and checksummed result packages are publicly available.
[IR-59] PsyBer-Agent : A Psychology-Driven User Agent for Recommender Systems
链接: https://arxiv.org/abs/2610.13989
作者: Chong Zhang,Qijie Shen,Haichao Zhang,Can Wang,Xiaobo Jin,Jia Wang
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注: 49 pages, 15 figures
Abstract:Evaluating recommender systems before deployment requires simulators that remain realistic over long, closed-loop interaction cycles. Current LLM-based agents generally function as surface imitators that generate realistic short-term responses but fail to capture the underlying user dynamics such as fatigue, affect, and shifting preferences. We propose PsyBer-Agent, a psychology-inspired simulator that models these dynamics through its Psy-Engine. This engine represents each user with three evolving, model-defined latent states. To calibrate simulated behavior against real interaction logs, we align log-conditioned reference trajectories with simulator-generated latent-state trajectories via Unbalanced Gromov–Wasserstein optimal transport. This approach provides structured supervision for a lightweight policy without the computational burden of fine-tuning the backbone LLM. We also introduce WebSim and the PsyBer Benchmark, which encompass multimodal scenarios and seven recommender backbones. Our framework evaluates behavioral fidelity, latent-state consistency, robustness, and responses to social-cue perturbations. On the tested horizons, PsyBer-Agent shows lower behavioral discrepancy than the prompt-only simulators while maintaining stable performance under large-scale concurrent simulation for up to 5M queued simulation tasks under bounded concurrency. Code is available in the supplementary material and at: PsyBer-Agent: this https URL, WebSim Platform: this https URL.
人机交互
[HC-0] What Human-AI Interaction Leaves Behind: Residual Human States – A Trajectory-Based Framework and Multi-Study Analysis
链接: https://arxiv.org/abs/2610.18765
作者: David Herrera-Poyatos,Carlos Peláez-González,Cristina Zuheros,Francisco Herrera
类目: Human-Computer Interaction (cs.HC)
备注: 29 pages, 13 tables. Code: this https URL
Abstract:Human-AI Interaction is commonly evaluated by what an artificial agent enables during use, yet interaction can also alter the human who acts next. We introduce Residual Human States, a trajectory-based framework for representing the potentially multidimensional human state carried forward after interaction with AI, relative to a relevant alternative trajectory. The framework distinguishes assisted output from human-state transition and derives four empirical propositions: multidimensionality, interaction-policy dependence, horizon dissociation, and temporal persistence. Its central claim is that assisted performance does not by itself determine the human state carried forward. We examine these propositions through four empirical cases with different inferential roles. Published medical-training evidence illustrates how one AI interaction history can be associated with both slightly higher observed accuracy on some later unaided cases and transmission of a systematic decision bias on others. A second diagnostic-learning study shows that later unaided behaviour differs across AI-support policies, while also revealing that aggregate accuracy cannot by itself distinguish improved discrimination from a shifted response criterion. In a classroom experiment, generative-AI tutoring substantially improved assisted practice performance without a corresponding detectable gain on a subsequent unaided exam. A longitudinal reanalysis of personalised AI dialogue further shows that residual contrasts can evolve across post-interaction horizons and that such change may arise from the comparison trajectory as well as from the focal intervention. Evaluating interactive AI systems therefore requires attention not only to contemporaneous performance, but also to which human states alternative interaction trajectories produce, relative to which counterfactuals, and how durably. Comments: 29 pages, 13 tables. Code: this https URL Subjects: Human-Computer Interaction (cs.HC) Cite as: arXiv:2610.18765 [cs.HC] (or arXiv:2610.18765v1 [cs.HC] for this version) https://doi.org/10.48550/arXiv.2610.18765 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[HC-1] Bio-Inspired Electrotactile Feedback: A Framework and Impacts of Multimodal Cues
链接: https://arxiv.org/abs/2610.18746
作者: Melika Emami,Amirhossein Bayat,Rahim Tafazolli,Atta Quddus
类目: Human-Computer Interaction (cs.HC); Emerging Technologies (cs.ET)
备注:
Abstract:Electrotactile stimulation, as a haptic feedback method, often struggles to evoke high-quality natural sensations. Consequently, despite its practical advantages, it is generally less preferred in human-machine systems than mechanical approaches. In this paper, we propose a bio-inspired stimulation framework that generates signals modelled on afferent neural activity derived from a touch model. By producing more biologically plausible spiking patterns, the proposed approach aims to enhance the realism of perceived tactile sensations. To evaluate these stimulation patterns in a realistic task context and to quantify the influence of visual and movement cues, we conducted a series of subjective experiments under three conditions. The results indicate that bio-inspired modulation was effective: in particular, bio-inspired co-modulation of amplitude and frequency achieved a favourable pressure-irritation trade-off, increasing the probability of perceiving pressure while maintaining low irritation (pressure: 0.78; irritation: 0.24) relative to the other patterns. Visual cues and hand movement further improved feedback quality by significantly increasing the probability of reporting pressure (p 0.001) without a corresponding increase in irritation, and by increasing mean naturalness ratings. Overall, these findings demonstrate the potential of bio-inspired electrotactile patterns to improve haptic feedback quality, and highlight the importance of evaluating such methods under realistic, multimodal interaction scenarios.
[HC-2] AIRRI: Towards a Human-Readable Machine-Resistant Reading Interface
链接: https://arxiv.org/abs/2610.18724
作者: Emory Salaun,Syed Ayon,Kaleel Mahmood,Shaun Wallace
类目: Computer Vision and Pattern Recognition (cs.CV); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
备注: 12 pages, 2 figures, 6 tables
Abstract:The rapid advancement of image-to-text extraction systems and large language models (LLMs) poses new challenges for educational assessment by making assignment materials easy to transcribe and solve with minimal human effort. To address this problem, we present AIRRI (Artificial Intelligence Resilient Reading Interface), a framework for generating machine-resistant educational content, a dataset of higher-education assignments for evaluation, and an initial evaluation of the human readability of AIRRI machine-resistant text designs. AIRRI employs two categories of machine-resistance techniques. First, AIRRI applies image transformations including character spacing, word spacing, line spacing, character opacity variation, stripe overlays, and image noise. Second, AIRRI applies decision-based adversarial attacks: RayS, SurFree, and ADBA. We evaluate AIRRI against three image-to-text systems: Tesseract, EasyOCR, and the transformer-based GOT-OCR2 model. Our results show that AIRRI substantially degrades automated text extraction performance. AIRRI increased character error rates for EasyOCR (6.9% to 55.9%), Tesseract (2.3% to 100.0%), and GOT-OCR2 (30.7% to 48.1%). Word error rates increased for EasyOCR (24.9% to 96.7%) and Tesseract (10.7% to 100.0%). Among the evaluated decision-based attacks, ADBA consistently produced the largest degradation in image-to-text performance, while RayS and SurFree produced more moderate reductions. EasyOCR and Tesseract were generally more susceptible to both image-based and adversarial perturbations than the transformer-based GOT-OCR2 model. Our findings suggest that AIRRI can significantly reduce the reliability of image-to-text-to-LLM workflows, increasing the effort required for AI-assisted assignment completion.
[HC-3] Knowing the User Is Not Knowing What We Established: Grounding Awareness in Personal Agents
链接: https://arxiv.org/abs/2610.18684
作者: Jiajun Chen,Xinyue Chen,Yicheng Jiang,Hua Shen
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
备注: Under Review
Abstract:Personal-agent memory has largely focused on learning what users know, prefer, and want. However, knowing the user does not necessarily mean knowing what the user and the agent have established together. An assistant’s proposal should not become an authorized instruction merely because it is remembered. We introduce Grounding Awareness, the ability to distinguish accepted, unacknowledged, revised, and scoped arrangements when selecting actions. We develop a controlled benchmark to test whether agents can tell the difference between proposals users have accepted and those they have not, and whether they apply accepted arrangements only where appropriate. Across multiple agents and memory systems, we find that strong memory performance does not ensure Grounding Awareness. More surprisingly, providing more complete user information can increase execution of unacknowledged proposals, while persistent memory can cause agents to treat these proposals as if they were user instructions. We trace these failures through memory writing, retrieval, and action selection. Finally, we introduce Grounding Layer, a lightweight, training-free extension that explicitly tracks grounding status and checks consequential actions, substantially improving agent behavior without modifying models or memory systems.
[HC-4] Communicating Uncertainty in AI-Supported Nurse Scheduling: Expert Co-Design and Evaluation of an Explainable AI Prototype
链接: https://arxiv.org/abs/2610.18679
作者: Alice Rollwagen,Emily Schiller,Anton Hummel,Kathrin Ebertsch,Andreas Riener,Jan-Philipp Steghöfer
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Nurse scheduling is a complex optimization problem in which nurses are assigned to shifts to meet staffing demands while complying with working-time limits, qualification requirements, and other operational and regulatory constraints. Artificial intelligence is increasingly being used to support this process by forecasting future staffing needs and providing advice on how to improve shift scheduling. Effective human-AI collaboration in nurse scheduling requires users to understand, interpret, and appropriately act on AI forecasts and advice. However, it remains insufficiently understood how healthcare professionals use transparency mechanisms like uncertainty communication and AI advice explanations when making staffing decisions. Applying a human-centered design approach, we iteratively developed a prototype decision support system for nurse scheduling and evaluated it through a mixed-methods user study with expert users. Using thematic analysis, questionnaire data, and eye tracking, we examine how healthcare professionals interact with the system and how uncertainty communication and explanations of AI advice influence understanding, decision confidence, trust, and usability. We found that the uncertainty communicated in AI forecasts was frequently misinterpreted and reduced decision confidence. Explanations were generally perceived positively, but did not significantly affect decision confidence or advice acceptance. We conclude with eight design implications derived from our findings, providing insights for the design of more transparent and effective AI-supported decision support tools in healthcare.
[HC-5] Known-Voice Cloning and the Limits of Personality Transfer for Character Design
链接: https://arxiv.org/abs/2610.18669
作者: Johanna Magdalena Kuch,Maud van Lier,Daksitha Senel Withanage Don,Patrick Gebhard,Tanja Schneeberger,Elisabeth André
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Robotics (cs.RO)
备注: 17 pages, 3 figures, 1 table. This preprint has been accepted for publication in Connected Futures - Nature, Robots, and Society (Robophilosophy 2026), in the Lecture Notes in Computer Science (LNCS) series, volume 17299 (expected publication: December 2026)
Abstract:Recent advances in zero-shot voice cloning systems make it possible to reproduce a recognizable voice from a short recording and attach it to arbitrary virtual characters. Prior work suggests that familiar vocal identity can carry over associated personality impressions to new targets such as robots or virtual characters. However, these works have mainly examined familiarity and personalization effects rather than directly testing whether a cloned familiar voice systematically transfers personality impressions to an embodied virtual character. Such transference raises the concern that cloned voices could be used to manipulate people into engaging with virtual characters as capable social actors. Across two studies, we investigated whether personality impressions transfer when a virtual character speaks with the cloned voice of a familiar person. Study 1 examined this question in a real-world university setting using the cloned voice of a lecturer familiar to the participating students. Study 2 examined the same question online using voices of social media content creators and members of their communities. Drawing on the concepts of identity activation, trait transfer, and person reconstruction, we show that evidence of personality transfer was limited in both studies. Observed effects were selective and did not generalize across traits or study contexts. In the online study in particular, recognizable voice identity cues activated expectations about the speaker but did not reconstruct the broader person impression. Qualitative responses further indicated that judgments were frequently shaped by perceived artificiality and by whether voice and appearance fit together. Our findings moderate common concerns that cloned voices straightforwardly manipulate people into perceiving artificial characters as capable social actors.
[HC-6] From Mind Mapping to Dream Interpretation: Understanding Muslim Users Engagement with GenAI Tools in Religious Context
链接: https://arxiv.org/abs/2610.18549
作者: Ishrat Jahan,Soomanib Kamruzzaman,K M Nazib Hossen,Novia Nurain,A. B. M. Alim Al Islam
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Generative Artificial Intelligence (GenAI) tools are increasingly integrated into everyday life, including sensitive domains such as religion. Prior works examined alignment with religious values and discussed the potential risks. However, to gain empirical insights into the adoption of these tools in users’ religious lives, in this study, we conducted an online survey with 178 Muslim participants in Bangladesh and collected 726 prompts on religious use cases, along with open-ended responses about using these tools in religious contexts. Using reflexive thematic analysis, we identify three key roles these tools play in users’ religious lives: (1) religious assistants, (2) perceived religious scholars, and (3) faith counselors. However, participants expressed concerns about misguidance, overreliance, and violations of Islamic tradition, and reported adopting protective strategies to avoid these pitfalls. Based on these, we propose a set of design guidelines for faith-sensitive GenAI systems that would preserve religious boundaries while supporting religious practices.
[HC-7] How Safe Is ChatGPT Health Advice? Evidence from Real User Conversations in Four Global South Countries
链接: https://arxiv.org/abs/2610.18465
作者: Shreyasi Roy Chowdhury,Kiran Garimella
类目: Computers and Society (cs.CY); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC)
备注: 25 pages, 17 figures, 8 tables
Abstract:People increasingly bring their health questions to ChatGPT, yet safety evaluations of LLM health advice rest almost entirely on clinician-authored vignettes, synthetic prompts, or simulated user profiles. The gap is widest for the Global South, where concerns about demographic bias and unsafe recommendations are well documented but have never been tested on real usage. We present, to our knowledge, the first demographically grounded safety evaluation of real-world health conversations with ChatGPT. From a consent-based data-donation corpus of 202,590 conversations from 1,252 users in India, Nigeria, Brazil, and Pakistan (December 2022 to February 2026), each linked to self-reported age and gender, we identify 15,555 health conversations, classify their health intent using an LLM classifier adapted from an established taxonomy, and score response safety under both a context-blind and a user-context-aware rubric. Roughly one in five health conversations receives a high-risk rating. Risk concentrates in symptom and condition queries and recurs through identifiable failure modes: unsafe medication or dosage guidance, harmful self-diagnosis, missing safety warnings, encouraging over-reliance on the model, and failure to refer users to professionals. Supplying the user’s own accumulated context raises the high-risk share by 13-17 percentage points, and about one in four conversations rated safe in isolation turns out to be unsafe for the specific person asking. Safety evaluations built on synthetic cases are likely to miss these harms.
[HC-8] Understanding the “Why”: Investigating the Design of Privacy Support Tools (PSTs) through the Lens of Psychological Needs
链接: https://arxiv.org/abs/2610.18366
作者: Sara Hahn(1,2,3),Verena Zimmermann(4),Simon Althaus(2,3),Andreas Heinemann(1,3),Ephraim Zimmer(2,3),Alina Stöver(2,3),Nina Gerber(2,3) ((1) Darmstadt University of Applied Sciences, (2) Technical University of Darmstadt, (3) ATHENE - National Research Center of Applied Cybersecurity, (4) ETH Zurich)
类目: Human-Computer Interaction (cs.HC)
备注: 4 pages, 3 figures
Abstract:Privacy Support Tools (PSTs) help users control their digital data, but face low adoption due to resignation and overwhelm. While current regulations aim to enhance privacy control, users still struggle to manage their privacy. To address this, we explore how PSTs can be designed for long-term adoption by applying the theoretical lens of psychological needs, as their fulfillment motivates users to use a digital product. We conducted in-depth interviews with 15 users and identified Security, Autonomy, and Competence as key psychological needs. From our results, we derive implications and recommendations for the need-sensitive design of PSTs.
[HC-9] Exploring Object Recall in an Object-Rich AR Office Scene
链接: https://arxiv.org/abs/2610.18183
作者: Sangita Kunapuli,Shane Dirksen,Radha Kumaran,Yanxiu Jin,You-Jin Kim,Misha Sra,Tobias Höllerer
类目: Human-Computer Interaction (cs.HC)
备注: 11 pages, 9 figures. Accepted to IEEE ISMAR 2026 and published in IEEE Transactions on Visualization and Computer Graphics (TVCG), Vol. 32, 2026
Abstract:Augmented reality (AR) changes how users perceive and attend to their surroundings. Prior work suggests AR can reduce attention to the physical world, but it remains unclear whether this holds in environments dense with both virtual and physical objects, and which factors most affect object recall. This study examines recall in a mixed-reality office scene after an object-selection task, testing object-level factors (virtuality, size), context-level factors (semantic congruency, saliency), and user-level factors (cueing, dual tasking). Cueing informed participants of a later recall test, while dual tasking added cognitive load through a secondary audio task. Virtual objects were recalled more accurately than physical ones. Cueing produced a modest improvement in recall of present objects. Using playback software to reconstruct each participant’s view throughout the trials, we computed object-level saliency and found a weak positive correlation with recall, slightly stronger than the effects of virtuality, but not reaching statistical significance. These findings have implications for AR interface design and for building mixed-reality environments that better support cognition during real-world tasks.
[HC-10] CompanionSafe: From Multi-turn Safety Evaluation to Guardrails for AI Companions
链接: https://arxiv.org/abs/2610.18095
作者: Xinyang Liao,Lingyu Li,Kaicheng Shen,Yang Yao,Na Lu,Yan Teng,Xia Hu
类目: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Human-Computer Interaction (cs.HC)
备注: 40 pages, 33 figures, Code and dataset are available at this https URL
Abstract:AI companions provide users with emotional support and ongoing companionship, but their sustained relational interactions also introduce distinct safety risks. Prior work largely targets general harms or specific companion platforms and risk categories, while relying on manually collected conversations, limiting scalable and user-grounded evaluation and mitigation. To address this, we introduce CompanionSafe, a unified evaluation-to-mitigation framework for AI companion safety. Its evaluation component, CompanionEval, provides scalable multi-turn evaluation by constructing 1,300 relationally grounded settings from diverse personas, matched companion characters, and dialogue scenarios across four turn tiers. Its mitigation component, CompanionGuard, converts benchmark-exposed failures into turn-localized supervision for companion-specific risk detection. Our results show that even strong LLMs and existing guard models remain vulnerable to companion-specific relational harms, while both closed- and open-source models exhibit increasing safety risks as interactions become longer. Together, CompanionEval and CompanionGuard form a practical evaluation-to-mitigation loop for AI companion safety, turning relational safety failures exposed during multi-turn interaction into supervision for companion-specific guardrails.
[HC-11] Safety for Whom? Benchmarking Understanding Beyond Refusal in Egocentric Multimodal Assistants
链接: https://arxiv.org/abs/2610.17728
作者: Yewon Han,Jisu Jeon,Suhyun Choi,Juan Yeo,Minsoo Jo,Gihoon Kim,Hyunggyu Lim,Taesup Kim
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:
Abstract:Egocentric multimodal assistants see through the user’s eyes and guide what the user does next, so their outputs directly shape physical actions, including those that affect bystanders. Yet existing safety evaluations do not distinguish whom a response would harm, and score only whether a model refuses. Pooling all harms into one refusal rate hides two failures: a model may protect one target but not another, and it may refuse without understanding the risk, or understand it and comply anyway. We present STOU, a benchmark of 2,902 egocentric video-audio and video-text queries whose potential harm targets the user (Self), co-present individuals (The Others), or society at large (Others). Our evaluation decouples safety understanding from refusal behavior: beyond refusal, we score whether the model recognizes the user’s intent, the presence and target of harm, and its downstream consequences. Across 10 open and closed source omnimodal and multimodal LLMs, we find that refusal varies systematically with the harm target, and that models often recognize a harm yet still assist with it. These results show that refusal rates alone cannot certify an assistant as safe, and that safety alignment must account for who would be harmed. Project page: this https URL
[HC-12] InSight: A Mobile Assistant Helping Visually Impaired People Navigate Unfamiliar Environment
链接: https://arxiv.org/abs/2610.17660
作者: Junwei Zheng,Jiale Wei,Di Wen,Yufan Chen,Ruiping Liu,Chengzhi Wu,Kunyu Peng,Kailun Yang,Jiaming Zhang
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Navigating unfamiliar outdoor environments is one of the most demanding everyday tasks for people with visual impairments (PVI). However, existing navigation systems cannot simultaneously satisfy the needs of informative perception and precise localization during navigation. This paper explores the core navigation demands of PVI through a preliminary study. Based on the results of the pre-study, we present the design of InSight, a mobile assistant that combines scene descriptions with precise localization to support PVI navigating unfamiliar outdoor environments. The offline mode of InSight runs entirely on device for real-time navigation, while the online mode streams camera frames to a self-hosted server for arbitrary object detection and accurate localization. Finally, a user study with 21 visually impaired participants validates that InSight could assist users in navigating unfamiliar outdoor environments with increased success rate and confidence compared to mainstream navigation applications like Apple Maps.
[HC-13] Get In Decide Prove: A Merchant-Side Readiness Benchmark for Agent ic Commerce
链接: https://arxiv.org/abs/2610.17378
作者: Adrian de Valois-Franklin,Alex Bogdan,Anas Ibrahim,Ibrahim Shaer,Tamaz Andguladze
类目: Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
备注: 23 pages, 7 figures, 10 tables. Companion dataset: this https URL
Abstract:AI agents now browse and buy on people’s behalf, and benchmarks measure how well the agents perform. Far less is known about the other side of the transaction: whether a merchant’s website lets an agent in, publishes terms precise enough to decide a purchase, and leaves evidence of what was agreed. This paper presents a merchant-side benchmark built on those three questions and reports a pilot of it. The central measurement, decision sufficiency, rates each criterion of a concrete buyer mandate (total cost, cancellation or return terms, service date, variant identity) as determinate, indeterminate, or unknown from the merchant’s own published evidence. Readiness scores carry their evidence coverage and missingness bounds; access is reported separately for a declared bot, a declared headless browser, and a user-present browser; and each observed term is mapped to the RAILS admissibility class of the evidence that could support it in a dispute. Four large merchants received the full assessment: three refused the declared profiles at their offer pages, and no mandate could be confirmed as met, because each merchant left a total or a refund term undetermined. An access census of 555 sites of merchants that two payment processors publicly name as clients found the home page unreadable to both declared profiles at 105 sites (18.9%). A declared assessment of 330 merchants, scored by deterministic rules without a rater, found that evidence coverage rather than the score decides who can be ranked: 115 merchants (35%) reached the ranking threshold, rankable profiles scored a median of 89 to 90, and material terms was the one pillar well below the others, because the all-in total was decidable at 18 merchants. The pilot shows that the measurements can be taken; it does not validate them or rank merchants, and the weights and the projected clearing band await validation against outcomes.
[HC-14] Brain2Face: A Study on the Relation of EEG Signals and Spatio-temporal Head Avatars
链接: https://arxiv.org/abs/2610.17356
作者: Camilo Andrés Martínez Martínez,Olaf Dünkel,Kartik Teotia,Maurice Rekrut,Christian Theobalt,Marc Habermann
类目: Computer Vision and Pattern Recognition (cs.CV); Human-Computer Interaction (cs.HC)
备注: Project page: this https URL
Abstract:Driving digital head avatars is classically done by hand-crafted artist design or by image-based facial performance capture. In this work, we ask whether spatio-temporal tracking of the humans’s head geometry and appearance can be performed solely from brain signals, i.e., Electroencephalography (EEG). Since prior work mostly focused on activity and emotion recognition, this task is relatively unexplored without even having publicly available high-quality datasets and benchmarks. Thus, we contribute a dataset featuring paired EEG and human head geometry as well as appearance annotations. We synchronize 64-channel EEG with per-frame expression latent codes of a personalized Gaussian head avatar. Second, we introduce a baseline to the task of expression recovery of a head avatar from EEG signals. To this end, we propose a dedicated EEG-to-expression model that takes a temporal sequence of EEG signals as input and regresses the latent expression parameters of the 3D Gaussian head avatar. Lastly, we validate our capture and algorithmic design choices through a benchmark protocol that we introduce to this task. We believe this work paves the way towards better understanding of the relation between EEG signals and facial surface as well as appearance changes.
[HC-15] Human Situational Awareness and Workload during Heterogeneous Autonomous Vehicle Hazard Monitoring: A Cross-Platform Study
链接: https://arxiv.org/abs/2610.17339
作者: Ziyue Wang,Yang Xing,Argyrios Zolotas,Adolfo Perrusquia,Weisi Guo,Antonios Tsourdos
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:As autonomous ground and aerial vehicles become increasingly capable, human operators are transitioning from direct control to supervisory monitoring roles. Maintaining Situational Awareness (SA) during rare but safety critical hazard events remains a fundamental challenge, particularly when operators must monitor heterogeneous autonomous platforms with different spatial and cognitive demands. This study develops a unified experimental framework for investigating human SA and workload during supervisory monitoring of autonomous ground and aerial vehicles under hazard conditions. Eighteen participants completed six within-subject conditions combining three interface modalities, namely, visual-only, haptic-only, and visual-haptic, with two platform types, ground vehicle and aerial drone. SA was measured using Situation Awareness Rating Technique (SART), while workload was assessed using the NASA Task Load Index (NASA-TLX). The key findings are: 1) Aerial monitoring produced significantly lower overall SA, higher situational demand, and greater workload than ground monitoring, driven by elevated Demand and reduced Understanding rather than diminished Supply. 2) Workload was consistently coupled with situational demand across all conditions, suggesting a stable SA-workload mechanism. 3) Both haptic-only and visual-haptic significantly improved overall SA relative to visual-only without increasing workload, demonstrating that interface design can enhance awareness within the existing workload envelope. These findings establish a cross-platform behavioural baseline for designing multimodal interfaces that enhance operator SA without elevating workload in heterogeneous autonomous vehicle systems.
[HC-16] Calibrating LLM Explanations to Human Decisions with Reinforcement Learning
链接: https://arxiv.org/abs/2610.17306
作者: Dong Shu,Ziyang Guo,Negar Kamali,Matthew Groh,Jessica Hullman
类目: Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
备注:
Abstract:Explanations of LLM outputs are considered an important way to make model behavior legible, helping human users make good decisions on downstream tasks. However, default training approaches for improving explanations target model task performance or human preference rather than human decision quality. To bridge this gap, we introduce \textbfHD-RL, a reinforcement learning framework for optimizing LLM explanations for human decision making. Our approach calibrates a human decision simulator on observed human decisions, applying a proxy reward model to optimize the LLM to generate explanations that help people to make good decisions. We apply this framework to AI-generated text detection. In our initial data collection phase, we observed that while LLMs possess strong autonomous detection capabilities, the explanations they generate for users yield near-random human accuracy (51.3%). By applying the HD-RL framework to the open-source Qwen 3.6 model, we generate optimized explanations that improve human performance, achieving a GLMM-estimated human accuracy of 63.7% in a randomized between-subjects experiment with repeated measures and outperforming RLVR, RLAIF, and other baselines by over 8%.
[HC-17] ouchTwin: Human-AI Haptic Authoring for 3D-Scanned Tabletop Scenes through Language and Touch
链接: https://arxiv.org/abs/2610.17302
作者: Wanli Qian,Xinlei Yu,Heather Culbertson
类目: Human-Computer Interaction (cs.HC)
备注: submission CHI2027
Abstract:Creating haptic experiences from real-world environments remains difficult: visual scans capture appearance but not material feel, while manual haptic authoring requires specialized expertise. We present TouchTwin, a human-AI approach that turns phone scans of tabletop scenes into editable haptic scenes. TouchTwin uses AI to propose material regions that users can experience and revise through language and spatial edits. A model trained on measured tool-surface interactions generates the haptic responses, allowing users to work with material descriptions rather than low-level signal parameters. In a study with N=12 participants and physical reference objects present, automatic drafts scored 4.0/7 for realism and 5.4/7 as useful starting points. After a median of 6.6 minutes of editing, realism increased to 5.6/7, while scoring the saved material layers against ground truth showed material-family accuracy improving from 52% to 92%. Participants also used the same workflow to author material feels without a physical reference.
[HC-18] Particle Selection in Virtual Reality
链接: https://arxiv.org/abs/2610.17094
作者: Boaz Keren-Gil,James Gain,Lucia Marchetti,Patrick Marais,Alexander Sivitilli
类目: Human-Computer Interaction (cs.HC); Instrumentation and Methods for Astrophysics (astro-ph.IM)
备注:
Abstract:Point-cloud representations of 3D data are used in many fields, from astronomical surveys of galaxies to medical imaging and laser scans of buildings. A task common to all of these is interactive selection: picking out a meaningful subset of points for closer analysis. The data is inherently 3D, yet the traditional workflow uses a 2D mouse on a 2D projection and inherits the limitations of a flat screen. This paper compares Virtual Reality (VR) and desktop environments for 3D particle selection, asking whether VR’s six-degree-of-freedom hand interaction yields measurable gains in accuracy, speed, usability, workload and flow. In a counterbalanced within-subjects experiment (N = 34), participants selected 90 target galaxies from a catalogue of 16,816 galaxies, using mouse-driven selection with keyboard camera control on the desktop and a hand-held spherical brush that adds or removes particles in VR. We measured task accuracy, completion time, and three standard questionnaires: the System Usability Scale (SUS), the NASA Task Load Index (NASA-TLX) and the Flow State Scale (FSS). Accuracy was comparable (91.6% VR vs. 93.1% desktop; p = 0.858), but every other measure favoured VR. Tasks were 23% faster (p = 0.001, d = 0.93), usability was higher (SUS 84.6 vs. 54.0; d = -1.78), workload was lower (NASA-TLX 20.4 vs. 34.7; d = 0.90), and flow was higher (FSS 83.2 vs. 70.2; d = -1.23). Because the underlying problem, making a 3D selection through a 2D interface, is not specific to astronomy, we expect the pattern to carry over to point-cloud selection in other fields.
[HC-19] Better together? A performance theory of human-machine skill policies under generalization difficulty
链接: https://arxiv.org/abs/2610.16863
作者: Riccardo Zanardelli
类目: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY); Machine Learning (cs.LG); Theoretical Economics (econ.TH)
备注: 104 pages (65 main text + appendices), 21 figures, 27 tables. Extends arXiv:2509.14057 in performance terms; the utility analysis is left to future work. Reanalyzes the public Collab-CXR dataset (doi: https://doi.org/10.17605/OSF.IO/Z7APQ )
Abstract:Organizations pairing a human and a machine on a task decide, often by default, which executes each instance and what happens when both do. We account for what this attains under the squared rule, with performance distributed over instances to capture generalization difficulty. A gross stake lies between the skills’ mean and best per instance. It splits at the best single skill into a systematic part, known from separate assessments and secured by commitment, and a residual, the expected value of perfect information on which skill each instance favors, known only from paired observation and split further by classes. Skills set the stake; workflows decide the take, via two channels. Routing directs work, weighting or selecting responses; with binary truth it is capped by the best skill per instance. Reading responses as joint evidence is not routing; what they jointly reveal caps it. Synergy lets skills interact before responding; bounded by no skill either way, its lift over independent responses must clear a break-even and can fall below one. As algorithms, arrangements form a ladder stating what each reads, holds and records, and a procedure whose checks keep earlier gains when class means are known and lose little on finite records given a margin. The theory extends to many skills, each worth what it adds where it beats all others held, and to competing skills, where at parity a single failure’s allowed cost decides. In a public chest-radiograph study with 227 radiologists, they are near parity with the model and the stake is mostly residual; assisted reading, no better on average originally, returned less than averaging both responses; recalibrating the model alone beats any deployable arrangement of raw responses. Organizations can learn from their records what is at stake, which arrangement collects it and what evidence each step needs; pricing costs and errors is future work.
[HC-20] Integrating Feature Understandability into Feature Selection for Human-Centred Tabular Explanations
链接: https://arxiv.org/abs/2610.16627
作者: Nicola Rossberg,Bennett Kleinberg,Barry O’Sullivan,Luca Longo,Andrea Visentin
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Machine Learning (cs.LG)
备注: 24 pages, 5 figures, preprint
Abstract:Existing human-centred XAI research predominantly evaluates explanations after they have been generated. However, this fails to provide actionable insights into improving accessibility. This study incorporates human-based feature understandability into feature selection to improve the generation of human-centred explanations. Natural language explanations for machine learning models trained on tabular data often cite influential features in the input data to explain a given decision. As a consequence, users’ ability to understand the produced explanation hinges on their understanding of the cited features. To quantify this natural-language feature understandability, Rossberg et al. proposed the Feature Understandability Scale, which assigns numerical `understandability scores’ to features based on user feedback. To evaluate the degree to which these scores can improve machine learning explanation quality, this study collects understandability scores across two datasets, proposes an understandability-aware feature selection algorithm to select features that are both understandable and informative, and trains a random forest based on these feature subsets. The SHAP-based explanations generated by this model are evaluated using user feedback, and changes in explanation quality are assessed. The individual evaluation of the explanations showed that while the understandability-based explanations scored higher on average, the difference was not statistically significant. In contrast, the comparative evaluation found that participants preferred these explanations significantly more frequently than expected at the chance level. It is concluded that the integration of understandability can yield more understandable explanations at minimal cost to classification performance, although changes to quality require elucidation in larger cohorts.
[HC-21] Foresee: Making Procedural Futures Tangible through Generative Previews
链接: https://arxiv.org/abs/2610.16469
作者: Yusi Sun,Ying Jiang,Jiayin Lu,Jing Qian,Yin Yang,Yong-Hong Kuo,Chenfanfu Jiang
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Procedural guidance can show users how to act without explaining what an action enables later. We present Foresee, an interface combining next-step guidance with consequence previews, method comparisons, and cross-step access. A hybrid pipeline combines structured rendering, image editing, and video generation. A literature-informed framework and six-participant formative study shaped its design. A 24-participant study compared AR guidance, AR with generated video, and Foresee across digital painting, gameplay, and cocktail making. Exploratory painting comparisons showed higher correct completion without a recorded error than AR and higher post-task quiz accuracy than generated-video guidance under task-wise correction. Interviews revealed that users could overlook valued explanations, match visible outcomes without understanding later possibilities, or wait for concrete demonstrations when symbolic cues were unclear. These findings inform generative previews that explain what actions enable or constrain while supporting ongoing execution.
[HC-22] What UI Design Can Do For “Good” Oversight Work: A Three-Month Action Research Study
链接: https://arxiv.org/abs/2610.16410
作者: Cedric Faas,Lisa Dargasz,Julius Gabelmann,Felix Jahn,Markus Langer,Kevin Baum,Anna Maria Feit
类目: Human-Computer Interaction (cs.HC)
备注: 29 pages, 10 figures, Pre-Print
Abstract:With AI automating consequential tasks, human oversight is required for ethical, legal, and practical reasons. Yet these oversight tasks can be demanding and often make for poor work. Empirical evidence on overseers’ lived work experiences remains underexplored, yet critical for designing ``good’’ oversight. We provide longitudinal, in-the-wild insights through action research accompanying (N=8) tutors who oversaw an AI-supported learning assistant over three months. Our thematic analysis combines insights from twelve weekly meetings and two workshop iterations and was enriched by four repeated questionnaires. We identified stimulation, mastery, and relatedness as key challenges, stemming from the task’s repetitiveness, unclear impact of interventions, and limited connections with students and peers. Informed by our initial findings, we deployed targeted UI modifications with mixed success. This paper documents the long-term evolution of the work characteristics of a real-world oversight task and discusses which oversight work characteristics can and cannot be supported by interface design.
[HC-23] STEPwise: Supporting Teachers in Diagnosing Students Mathematical Problem-Solving Skills through LLM -Generated Questions and Rubric
链接: https://arxiv.org/abs/2610.16344
作者: Yoonsu Kim,Hayeon Doh,Eunhye Kim,Jinho Son,Juho Kim
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Problem-solving is central to mathematics education. However, diagnosing where students’ reasoning breaks down requires individualized probing and consistent criteria that teachers cannot sustain at classroom scale. We introduce STEP (STage-based Evaluation of Problem-solving skills), co-designed with six mathematics education experts, which operationalizes problem-solving as four stages and eight diagnostic sub-skills. By specifying what evidence to elicit and assess at each stage, STEP enables an LLM pipeline to generate stage-specific questions and rubrics for any math word problem, reducing the manual design of diagnostic materials. An expert review (N=8) assessed the pedagogical validity of the generated materials and informed STEPwise, a teacher-in-the-loop system for diagnostic material generation and process-level assessment. In a classroom study with 5 teachers and 64 students, STEPwise helped teachers and students locate reasoning breakdowns and supported students in reflecting on their reasoning. Together, STEP and STEPwise offer a structured approach to fine-grained diagnosis at classroom scale.
[HC-24] BehaviorFormer: Discovering Behavioral Patterns via Clusterable Next-Token Prediction
链接: https://arxiv.org/abs/2610.16322
作者: Tobias Schneider,Danial Fahrner,Soren Antebi,Daniel Uedelhoven,Lennard Bodden,Sandra Halscheidt,Rafet Sifa
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注: Accepted at IEEE Conference on Games 2026
Abstract:Discovering behavioral patterns in open-ended games is challenging due to the lack of clear objectives or ground-truth annotations. In this work, we extend a next-token prediction Transformer with the deep clustering algorithm DCSS to learn meaningful representations with well-separated clusters. In a controlled Minecraft session with 26 players, the method discovers 23 distinct behavioral patterns that generalize across the studied players, further refining structures present in self-supervised representations into clearly separable clusters. We group the clusters into seven interpretable categories covering stationary, roaming, travel, mining, base work, logging, and farming behavior.
[HC-25] FlowAvatar: Real-Time Full-Body Avatars from Sparse Egocentric Inputs on Consumer XR Devices
链接: https://arxiv.org/abs/2610.16277
作者: Chenfeng Gao,Taeyoung Yeon,Sungheon Park,Vasco Xu,Anish Prabhu,Mar Gonzalez-Franco,Karan Ahuja
类目: Computer Vision and Pattern Recognition (cs.CV); Human-Computer Interaction (cs.HC)
备注: Accepted to IEEE ISMAR 2026 (Conference Track). 10 pages, 9 figures. Code: this https URL
Abstract:Prior avatar embodiment research has studied tracking completeness (how faithfully the body, hands, and face are sensed and reproduced) and personalized appearance in isolation, each requiring specialized lab equipment. How these factors interact on consumer hardware remains an open question, because no existing system unifies body pose, face, hands, and personalized appearance on a standalone headset. We present FlowAvatar, a system enabling real-time personalized look-alike avatars with full-body pose, hand articulation, facial expression, and eye gaze tracking from sparse egocentric inputs on consumer XR devices (Meta Quest Pro). Our pose model achieves competitive spatial accuracy and state-of-the-art temporal consistency on the AMASS benchmark against prior works with comprehensive spatial and temporal losses validated through architecture and loss ablations, while running in real-time. Two user studies (N=12, N=10) provide converging evidence that tracking completeness and personalized appearance serve complementary roles: tracking completeness broadly improves perceived motion quality, while only their combination achieves positive total embodiment.
[HC-26] Access|Use|Sustain|_-ability: Exploration and Quantification of a Linguistic Structure and Its Use in HCI
链接: https://arxiv.org/abs/2610.16206
作者: Alarith Uhde,Paul Haimes
类目: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY)
备注: 8 pages, 4 figures, accepted at the International Journal of Asia Digital Art and Design
Abstract:Language plays a crucial role in research and technology design, throughout problem formulation, method descriptions, analysis of findings, and interpretation. In this paper, we focus on a particular, trending linguistic pattern in the Human-Computer Interaction literature, which we call -abilities. Examples include usability, accessibility, and explainability. These -abilities have become increasingly popular, with potential impacts on the discipline. They take a verb, turn it into an adjective, and finally nominalize it. We approach the -abilities phenomenon with a survey of the HCI literature between 1988 and 2025. All in all, we identified 349 different -abilities. Moreover, the use of -abilities notably increased since the late 2000s, both in the absolute number of different -abilities and the frequency of use of the most popular ones. The majority of -abilities we found had a positive notion, and most -abilities with a negative notion are prefixed counterparts of their positive opposite (e.g., un+sustainability). We critically discuss implications of their increased use. First, -abilities change the literal meaning of design intentions. For example, acceptability reduces the concrete verb accepting into only its potential (acceptable), and finally nominalizes it, claiming the existence of acceptability as a somewhat stable and real, abstract characteristic of a technology. Second, the prefix-based opposite construction can weaken the design goal. Un+sustainability avoids clarification of the -ability’s meaning (here e.g., ecological or financial), and does not spell out the expected impact of designs that are not sustainable (e.g., the destruction of nature). Future work should compare the use of -abilities in English with other languages to better understand the ways in which they transform meaning.
[HC-27] Exploring Mixed Reality for Engagement in Robotics: A Comparison of Robot Control Methods
链接: https://arxiv.org/abs/2610.16186
作者: Samantha Edvell,Dinh Tung Le,Gavin Paul
类目: Human-Computer Interaction (cs.HC)
备注: 10 pages. Submitted to ACRA 2026
Abstract:This paper presents a Mixed Reality (MR) robotics experience that provides an accessible introduction to interacting with a physical robot. The application combines a TurtleBot with a virtual animal-tending activity and three control methods: Joystick, Waypoint, and Path-Drawing. A mixed-methods user study with 10 participants investigated changes in robotics interest and confidence following the experience and how the control methods influenced engagement, workload, and preference. Confidence increased following the experience, with the exploratory analysis indicating a significant pre–post difference, while the increase in interest was not statistically significant. The control methods produced clear differences in perceived workload and ease of use, with Path-Drawing generally more demanding and difficult to use. Joystick was preferred by eight of ten participants and was associated with better perceived performance and lower frustration. However, no statistically significant differences were detected in the other engagement dimensions, suggesting that the most preferred method was not necessarily more engaging across all aspects of the experience. These findings highlight the importance of considering both usability and engagement when designing how users interact with robots in introductory MR experiences.
[HC-28] A Mechanomyography-Based Integrated Sensing-Stimulation System for Fatigue-Aware Neuromuscular Therapy
链接: https://arxiv.org/abs/2610.15987
作者: Xiao Jin,Cong Liu,Zhenhua Yu
类目: Human-Computer Interaction (cs.HC)
备注: 12 pages, 14 figures
Abstract:Muscle fatigue limits the effectiveness of functional electrical stimulation (FES), yet fatigue monitoring during active stimulation remains challenging because electromyography (EMG) is severely contaminated by stimulation artifacts. To address this issue, this study proposes an integrated mechanomyography-based sensing framework (MMG-FES) for artifact-resistant fatigue monitoring during stimulation. Fatigue states were first characterized during voluntary movement using established EMG spectral indicators, including median frequency (MF) and mean power frequency (MPF), and then linked to MMG-derived features, including high-frequency energy ratio (HFER) and spectral bandwidth (BW), for fatigue recognition under stimulation conditions. Experimental results showed consistent fatigue-related trends, with EMG spectral features decreasing and MMG features exhibiting coordinated spectral redistribution across repeated contractions. Using these features, the proposed framework achieved fatigue-state recognition accuracies of 0.93 and 0.96 with logistic regression and random forest, respectively. These findings support MMG as a physiologically meaningful and stimulation-compatible modality for fatigue-aware closed-loop neuromuscular stimulation.
[HC-29] Vision: “What Dont I Know About Users Preferences?” How Uncertainty Estimation Improves Proactive Assistant Alignment
链接: https://arxiv.org/abs/2610.15948
作者: Zhiyu An,Ziyi Xuan,Yu Yang,Wan Du
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注: Published in Companion of the 2026 ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp Companion '26) PEACE Workshop
Abstract:Proactive Smart Assistants (PSAs) increasingly use general knowledge, learned patterns from group-level data, patterns from individual user’s data, and recent sensor observations to anticipate which action a user may prefer. Existing systems typically select actions from predicted user preference likelihoods, but these likelihoods alone do not reveal how reliable the prediction is, which evidence supports it, or why ambiguity remains. In this vision paper, we argue that uncertainty estimation can provide this missing information and substantially change appropriate downstream behavior. Using the example of a home assistant, we introduce the questions that uncertainty estimation aims to answer, i.e. how sure is the assistant about the predicted preference likelihood, whether the confidence come from group-level assumptions or individual’s data, and whether likelihood ambiguity reflect the user’s intrinsic preference variability or insufficient data for the particular situation. We further introduce uncertainty estimation applications in neighboring domains, such as human trajectory prediction, personalized activity recognition, information-collection, and model-based control. Finally, we draw on insights from neighboring domains to envision how future uncertainty-aware proactive assistants may address key challenges in proactive interaction.
[HC-30] FreeBrush: Motor Signals-Based Toothbrushing Monitoring System for Electric Toothbrushes
链接: https://arxiv.org/abs/2610.15800
作者: Pinhao Lin,Xuanyu Liu,Guoheng Ma,Haoxian Liu,Zongqi Yang,Jin Zhang
类目: Human-Computer Interaction (cs.HC)
备注: 15 pages. Includes a 2-page supplementary document as an ancillary file. Extended version of our workshop paper “BrushSense: Toothbrushing Region Recognition Using Motor Signals” (UbiSense 2026)
Abstract:Oral health is vital for overall well-being, yet many users brush different tooth surfaces for insufficient or excessive durations. Toothbrushing monitoring systems can help improve brushing habits by reminding users of under-brushed and over-brushed surfaces, but existing approaches often rely on external sensors, increasing cost and reducing practicality. We observe that the intrinsic motor signals of electric toothbrushes can reflect the brushing surfaces and durations. When brushing different tooth surfaces, the brush head experiences different constraints from the surrounding oral cavity, changing the motor’s driving currents and position feedback. Based on this observation, we propose FreeBrush, a toothbrushing monitoring system that uses the intrinsic motor signals of electric toothbrushes to recognize brushed tooth surfaces and estimate brushing durations without any external sensors. We prototype the system on a commercial off-the-shelf electric toothbrush and conduct a comprehensive evaluation with 18 participants. FreeBrush achieves 92.8% accuracy for 16-surface recognition and 92.0% accuracy for brushing duration estimation. The system further shows robustness across different toothpastes, brush heads, vibration modes, usage scenarios, and long-term use, and is capable of running on edge devices. By recognizing brushed tooth surfaces and estimating the brushing duration, FreeBrush can help users identify under-brushed and over-brushed surfaces and build balanced brushing habits.
[HC-31] A Co-Located MMG-FES Wearable Interface for Action-Updated State-Gated Stimulation
链接: https://arxiv.org/abs/2610.15605
作者: Cong Liu,Xiao Jin,Zhenhua Yu
类目: Human-Computer Interaction (cs.HC)
备注: 9 pages, 8 figures. Submitted to IEEE Robotics and Automation Letters
Abstract:Functional electrical stimulation (FES) assists movement, but fixed schedules cannot track how a repeatedly activated muscle changes, and surface electromyography (EMG) near the stimulation electrodes is often obscured by artifacts. Prior work recorded mechanomyography (MMG) passively during closed-loop FES cycling and estimated FES-induced fatigue or predicted fatigue for FES control, without gating stimulation with stimulated-site MMG. In this study, we present VG-FES, a wearable module co-locating microphone-based MMG, programmable charge-balanced biphasic stimulation, and embedded state-gated control at one on-skin thigh site. After each action, we compare rectus femoris (RF) mean power frequency (MPF) with an early-session baseline of the same direction, giving a relative state. A threshold on its window means enables fixed-duration bursts separated by rest periods. During onboard stimulation, MMG remained recordable, whereas near-site EMG was artifact-dominated. Under prolonged external stimulation, cycle-wise MMG tracked an independent local-pressure reference (Spearman rho = 0.968). In leave-one-session-out evaluation on 2284 squat actions (11 sessions, eight participants), MMG-based action segmentation reached an event-detection F1 of 0.923 and direction classification a balanced accuracy of 0.925. Online, over 15.69 min of self-paced squats, one participant received four bursts while MMG acquisition and action monitoring continued. Mechanical feedback from the stimulated site offers wearable FES and assistive robotics a per-action basis for state-gated stimulation.
[HC-32] Side-Vent Design for Microphone-Based Mechanomyography Sensors: Modeling and Experimental Evaluation of Contact Dependence
链接: https://arxiv.org/abs/2610.15600
作者: Cong Liu,Xiao Jin,Zhenhua Yu
类目: Human-Computer Interaction (cs.HC)
备注: 10 pages, 9 figures, 1 table. Submitted to IEEE Sensors Journal
Abstract:Wearable muscle sensing for rehabilitation, assistive robotics, and electrical-stimulation feedback needs mechanomyography sensors less sensitive to fit changes during wear. In microphone-based mechanomyography, cavity pressure depends on rim contact, a dependence housing design has rarely targeted. This study evaluates a side-vented housing, whose vent opens a pressure-relief path in parallel with the rim to lower contact dependence, and quantifies the acoustic-gain cost. Thermoviscous finite-element simulations compared twenty cavity configurations under prescribed source motion. Paired bench sweeps over 10-200 Hz on both housings compared direct contact with a Gap-ring spacer, a radially grooved spacer ring. At 100 Hz, side venting reduced the simulated acoustic-gain variation across 0.10-0.50 mm rim clearances from 25.8 to 10.0 dB and cost 5.5 dB of gain at 0.20 mm clearance. Velocity fields and branch-flow calculations identified this path. At the higher speaker drive level, the Gap-ring spacer caused median attenuations of 13.6 dB (vented) and 30.1 dB (unvented) relative to direct contact. In this front end, a 35.9-dB signal-gain increase at 100 Hz raised the calculated operational-amplifier voltage-noise gain by 1.6 dB. A single-participant demonstration of paced elbow flexion and extension compared envelope timing and spectra between housings, finding similar envelope timing (correlation 0.935) and broadly similar spectra (correlation 0.956). Vent area trades acoustic gain against contact dependence and can therefore serve as a housing design parameter.
[HC-33] AuthScribe: Supporting Agency and Authorship in LLM -Assisted Reflective Journaling
链接: https://arxiv.org/abs/2610.15439
作者: Lanping Hu,Yinsi Zhou,Zhenchang Xing,Xiaoyu Sun,Shidong Pan
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Large language models are increasingly used in reflective journaling, where articulating an experience is part of interpreting it. This raises questions about how writers retain authorship, exercise agency, and assess personalized assistance. We present AuthScribe, an LLM-assisted reflective journaling system that combines contextual anchors, writer-steered AI support, Autobiographical Citations, and read-only Provisional Drafts that require re-expression before AI-generated suggestions enter the journal. Retrieval and LLM inference run on-device. In a 14-day field deployment with 14 participants, grounding shifted from situating current experiences to reconnecting them with prior self-authored interpretations. Perceived agency remained stable through Week 1 and shifted during Week 2. Participants described authorship through transforming AI-generated suggestions into personally meaningful expression, and trusted personalized prompts when they could trace them to prior writing. We discuss design implications for adoption friction, longitudinal agency, and traceable personalization in AI systems for personal reflection.
[HC-34] ArguScale: Evidence-Grounded Human-AI Sensemaking Across Concurrent Live Conversations
链接: https://arxiv.org/abs/2610.15319
作者: Panayu Keelawat,Caleb Wohn,Shiva Pochampally,Rezky Dwisantika,Eugenia H. Rho,Yan Chen
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Monitoring concurrent classroom collaborative reasoning discussions is challenging because instructors have to track collaboration-relevant behaviors across multiple ongoing conversations at once. Existing Human-AI conversational sensemaking systems can scale monitoring upward through summaries and visual abstractions, but these representations may distance users from underlying conversational evidence, e.g., relevant audio and transcript segments, needed to forage downward. Through our formative study, we developed ArguScale, an evidence-grounded Human-AI sensemaking system that organizes interactional signals into temporally aligned representations linked to transcript and audio evidence. A temporal signal-density view provides temporal context and supports rapid cross-group comparison, while selective evidence access allows instructors to inspect audio and transcript. We evaluated ArguScale in a mixed-design study with 20 participants. Compared to other conditions, ArguScale improved participants’ ability to efficiently and accurately identify collaboration-relevant cues. We further identify strategies participants used to forage for and make sense of conversational evidence under live time constraints.
[HC-35] Using Evidence-Centered Design to Guide Prompt Engineering in Measurement: An Illustration with Automated Essay Scoring
链接: https://arxiv.org/abs/2610.15231
作者: Jiawei Xiong,Cheng Tang,Hong Jiao,Hanna Choi
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:
Abstract:Generative artificial intelligence (GenAI) models, particularly large language models (LLMs), are increasingly used in educational settings for tasks such as item generation, automated scoring, and feedback delivery. The quality of these outputs depends not only on the AI model but also on how users construct the prompts that guide model behaviors. Current prompt engineering practices are largely ad hoc, relying on trial and error rather than principled design. This lack of structure introduces validity threats including construct misrepresentation, construct-irrelevant variance, and limited reproducibility. We propose that Evidence-Centered Design (ECD), a framework in educational measurement for structuring valid inferences from observed performance, can serve as a principled foundation for prompt engineering in educational applications. We map the three core ECD models, the student model, the evidence model, and the task model, onto the components of educational prompt design, and describe a workflow for ECD-based prompt engineering. We derive five design principles from this mapping and present an empirical study contrasting ECD-designed prompts with state-of-the-art (SOTA) prompting baselines on an automated scoring task. Our empirical study demonstrates the usefulness of this framework on automated scoring. Results suggest that treating prompting as assessment design rather than informal instructions strengthens the validity, reliability, and fairness of AI-generated educational outputs.
[HC-36] rust Hinders Bidirectional Human-AI Alignment NEURIPS2026
链接: https://arxiv.org/abs/2610.15136
作者: Michaela Benk,Kacper Sokol
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注: 40th Conference on Neural Information Processing Systems (NeurIPS 2026). Workshop: Dynamic Alignment in Human-AI Coupled System (BiAlign Workshop)
Abstract:Trust calibration remains the dominant framing of AI-assisted decision making: the goal is to close the gap between perceived and actual trustworthiness of AI systems so that users exhibit appropriate reliance on them. We argue that this approach, while useful for decision problems that are well-defined and clearly scoped, overly restricts what it means to make a decision with AI support. To overcome this limitation, we develop three dimensions of AI-assisted decision making. First, trusting AI is only one of several possible strategies for managing uncertainty. Second, trust calibration is often misapplied when AI models “large-world” decision problems, i.e., ill-defined situations where goals and states are only partially known or unknowable. Third, in such settings human-AI alignment requires addressing higher-order decision processes. We conclude by reflecting on the implications of our observations on the design of process-oriented human-AI coupled systems that jointly identify and negotiate uncertainty throughout the decision-making task.
[HC-37] MotionAnimator: Semantic-to-Motion Reasoning for Animating Part-Structured 3D Assets in XR
链接: https://arxiv.org/abs/2610.15074
作者: Jianan Jiang,Bin Li
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:3D digital assets are increasingly used in education, simulation, design, and immersive content creation, yet most remain behaviorally inert: they provide geometry and component structure but lack the semantic annotations, motion constraints, and executable behaviors needed to function as dynamic objects in Extended Reality (XR). Animating such assets therefore still requires manual rigging, object-specific templates, or substantial programming expertise. We introduce MotionAnimator, an intent-driven, part-aware XR animation authoring system that transforms semantically unlabeled, part-structured 3D assets into executable animations from natural-language instructions. Rather than generating code directly from text, MotionAnimator externalizes object understanding and motion reasoning into explicit semantic and motion representations. It grounds object identity, semantic parts, and functional directions, constructs a structured plan that coordinates object- and part-level motion, and translates the plan into executable Unity behavior. On a benchmark of 90 prompts across 30 objects, MotionAnimator demonstrated reliable semantic grounding and coordinated animation generation across diverse objects and animation intents. An exploratory user study further provided insights into users’ experiences with MotionAnimator and highlighted its potential for part-aware XR animation authoring. These findings demonstrate how explicit semantic and motion representations can bridge high-level user intent and executable behavior, enabling existing part-structured assets to support coordinated XR animations beyond object-level transformations and predefined behaviors.
[HC-38] EnRoute: Iterative Design and Evaluation of a Mobile App for Lower-Carbon Fashion and Transportation Choices
链接: https://arxiv.org/abs/2610.15026
作者: Angela Busheska
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:EnRoute is an iOS application designed to support more sustainable everyday choices in transportation and fast fashion consumption. We describe three design iterations and the user research and field evaluations that shaped each one. Interviews with 15 people from 12 countries informed Version I, which combined errand route planning with basic fashion guidance. A survey of 1,200 people in five countries found much stronger interest in sustainable shopping (78%) than in switching to lower-carbon transport (46%); respondents most often named limited public transport as the main barrier to lower-carbon travel. We therefore narrowed the focus to fashion. Version II added item-level garment scanning with a manual-entry fallback, a Sustainability Coefficient that rates items, and recommendations of more sustainable alternatives; in a two-week evaluation with 3,021 participants in eight countries, participants randomly assigned to receive a short tutorial and educational prompts viewed more recommended alternatives and reported larger gains in purchase intent than those without it. Version III shifted from “buy better” to “buy less,” adding a wardrobe inventory, outfit matching, and a “Real Price” that combines cost per wear with estimated environmental and social costs. In a four-week pre/post pilot with 500 participants, self-reported monthly garment purchases fell from 2.9 to 1.9. We report descriptive results, discuss the substantial limitations of these uncontrolled and self-reported measures, and outline what the iterations suggest about designing for sustainable consumption.
[HC-39] Design and Evaluation of a Collaborative Teleoperation Interface for Multi-Operator Multi-Robot Systems in Challenging Environments
链接: https://arxiv.org/abs/2610.15001
作者: Alperen Kenan,Juan José García Cárdenas,Adriana Tapus,Paul Bremner,Manuel Giuliani
类目: Robotics (cs.RO); Human-Computer Interaction (cs.HC)
备注: 9 pages, 8 figures, accepted for presentation at the 2026 IEEE Conference on Telepresence in Bristol, United Kingdom, November 9-13, 2026
Abstract:Multi-robot systems that combine the complementary capabilities of specialised platforms are critical for complex missions in challenging, inaccessible environments. Current deployments assign an operator or operator team to each robot, with collaboration relying on operator-to-operator communication to manually coordinate heterogeneous platforms with incompatible interfaces. This operator-dependent approach increases cognitive load and limits scalability across robots from different manufacturers. This study designs and evaluates a shared-visualisation collaborative teleoperation interface to improve operator coordination, situational awareness, and the task performance in multi-operator, multi-robot systems. A comparative user study with 16 participants compared three teleoperation conditions: co-located operators with a shared interface, remote operators with a shared interface, and co-located operators with individual interfaces. Participants controlled two simulated robots in a nuclear power plant digital twin to complete disaster response tasks. Performance metrics, physiological data, and behavioural measures were recorded to assess task performance, cognitive load, and operator awareness across conditions. The co-located shared-interface condition achieved the fastest mission completion (292.8 s), outperforming remote operation (351.6 s) and individual interfaces (385.1 s), with the fewest collisions (0.9 per trial) and highest situational awareness accuracy (91.1 %). Physical co-presence was associated with reduced cognitive load, with NASA-TLX scores increasing from 38.2 in co-located to 57.7 in remote operation, supported by consistent physiological indicators. These results demonstrate that co-presence and shared interfaces provide complementary benefits, supporting workload regulation and mutual awareness respectively in multi-operator, multi-robot teleoperation. Comments: 9 pages, 8 figures, accepted for presentation at the 2026 IEEE Conference on Telepresence in Bristol, United Kingdom, November 9-13, 2026 Subjects: Robotics (cs.RO); Human-Computer Interaction (cs.HC) Cite as: arXiv:2610.15001 [cs.RO] (or arXiv:2610.15001v1 [cs.RO] for this version) https://doi.org/10.48550/arXiv.2610.15001 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[HC-40] How Should a Camera-First AI Agent Work?
链接: https://arxiv.org/abs/2610.14772
作者: Xiaofan Wu,Xi Zeng,Miaoxia Chen,Peishan Chen,Shuyan Li,Chen Wang,Jiyun Yao,Hanyong Zhong,Jiahao Zhu
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Logic in Computer Science (cs.LO)
备注: 8 pages, 4 figures, 3 tables. Position/contract paper on time-critical camera-first agents (systems/timing).Primary cs.AI;
Abstract:Camera-first agents must act on the visual world as it unfolds. We argue that the central obstacle for future camera-first agents is not recognition ability - multimodal models (vision-language and live/streaming alike) generalize well through language and remain the right tool for semantic reasoning - but time. Many visual targets are ephemeral: they appear, are query-relevant, and leave within a short window. Binding memory to the same end-to-end path as reasoning means that by the time a monolithic pipeline completes, the evidence is gone. We formalize this as Ephemeral-Target Interaction (ETI) and a Perceptual Deadline Contract separating a hard memory deadline from a soft reasoning deadline, grounded in the bounded, passive nature of sensory and working memory. We propose a two-deadline architecture: an always-on, bounded, non-semantic fast path; an active remembering mechanism that commits a moment to durable memory before it is overwritten (event-driven in v1); and a deferred deliberation layer (multimodal model + tools). We instantiate the contract on real MOT17 track windows, a consistency check, and calibrate the single-path baseline from measured frontier-model latency: even a short answer takes 1.2-9.3 s, and an agentic identify-and-search longer. Answers can be late; evidence cannot be missed.
[HC-41] Rethinking Opinion Measurement Interfaces for Human-in-the-Loop Optimization
链接: https://arxiv.org/abs/2610.14757
作者: Changkun Ou,Sven Mayer,Daniel Buschek,Andreas Butz
类目: Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
备注: 29 pages. Written in 2022, published as last revised. Dataset, analysis scripts, and system source code: this https URL
Abstract:Human-in-the-loop optimization refers to systems that present choices to a human decision-maker and query their judgment to optimize system parameters iteratively. The user interface typically provides either an absolute rating scale or asks for a preferred option. However, these approaches can potentially suffer from calibration problems, inconsistencies, and completeness violations of user preferences in such a sequential evaluation. This paper revisits the design space of opinion measurement interfaces for human-in-the-loop optimization. We conducted a study (N=360) using text summarization and image color enhancement tasks to evaluate human sequential feedback in six representative opinion measurement interfaces. Based on our analysis, we recommend using a listwise approach to support sequential opinion measurement, which counterweighs individual rating and preferential choice limitations and conveys more information for subsequent use. We further discuss the trade-offs in different user interface designs and provide guidelines to inform future research.
[HC-42] Impact of Expertise and Trust on AI-assisted Work
链接: https://arxiv.org/abs/2610.14660
作者: Qianou Ma,Pranav Khape,Rock Pang,Ken Koedinger,Tongshuang Wu,Siddarth Suri,Shamsi Iqbal
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Artificial intelligence (AI) agents promise to extend what users can accomplish beyond their existing expertise, but they also raise the risk of producing plausible-looking work that users may struggle to evaluate. We ask how task-relevant expertise and trust in AI shape agent-assisted work. To isolate these factors, we conduct a think-aloud study in which 20 professionals with varying data-science expertise use the same agentic tool to complete the same data-science hiring assessment, used at a major U.S. company. We find that surface productivity can come at the expense of understanding: participants who delegated more completed more of the task, but were also more likely to produce invalid solutions. Experts more often resisted this tradeoff, sacrificing completion to decompose the task and scrutinize agent outputs. Our findings call for evaluating agentic assistance not only by how much it accomplishes, but by whether users remain able to evaluate the work they produce.
[HC-43] Plottery: A Code-Integrated GUI for Matplotlib
链接: https://arxiv.org/abs/2610.14306
作者: Brian Hempel,Ruanqianqian Huang,Devamardeep Hayatpur,Sorin Lerner,Haijun Xia
类目: Human-Computer Interaction (cs.HC)
备注: Preprint. Final version is published in the proceedings of the 2026 IEEE Symposium on Visual Languages and Human-Centric Computing (VL/HCC 2026)
Abstract:Scientists have many tools to author plots: graphical direct manipulation apps like a spreadsheet’s chart builder, text-based programming tools like Matplotlib and ggplot2, or AI chat interfaces. Textual code is the most capable of these, but writing code, even with AI assistance, is a potentially poor fit for graphical chart design. We present Plottery, a Jupyter notebook extension that leverages API type annotations to add graphical direct manipulation to the popular Matplotlib plotting library. User interactions in Plottery’s in-situ graphical interface are immediately reified in the user’s code, so they may seamlessly mix-and-match between GUI, code, and AI interactions. In a user study, participants seamlessly alternated modalities, with the GUI used for straightforward visual changes, code for quick edits or duplicating code, and the AI for broad support. Plottery showcases graphical interaction that retains code’s benefits.
[HC-44] Defective Amplified Contested: Folk Conceptualizations of AI Slop
链接: https://arxiv.org/abs/2610.14231
作者: Zhixin Li,Jaime Banks
类目: Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
备注:
Abstract:AI slop has flooded social media feeds with overprocessed and derivative content–Jesus fusing with a shrimp, a humanoid orange chubby cat shoplifting, and a “hero” dog rescuing a human child after a plane crash. This might be no laughing matter, as waves of media coverage have taken a distinctly negative bent, emphasizing its low quality, ethical concerns, and threats to human creativity. Despite extensive attention from politicians and technologists, the perspectives of everyday users–the very audience most exposed to this phenomenon–remain largely unexplored. This descriptive work explored how Reddit users experience, conceptualize and contest AI slop. Inductive thematic analysis of posts and comments across six slop-related subreddits identified four themes covering constitutional defects, algorithmic overreach, user resistance, and entertainment outcomes–ranging from anxieties about eroding agency, moral norms, creativity, and existential questions to defensive distinctions between quality AI content and slop. We derive a user-driven definition: AI slop is audience-contested AI-generated content characterized by constitutional defects and enabled by platform logics. We operationalize slop through a facet-theoretical framework to inform future research into this fuzzy category.
[HC-45] Who Guides Whom? Negotiating Care Religious Authority and Technology in Bangladeshi Muslim Families
链接: https://arxiv.org/abs/2610.14229
作者: Susmita Biswas,Saiful Islam Tuhin,Faria Soroni,Sharifa Sultana,Jannatun Noor
类目: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY)
备注:
Abstract:Digital technologies are reshaping how families negotiate care, authority, and responsibility, yet these dynamics remain understudied in religiously situated family life. Through interviews with 38 adult members of Bangladeshi Muslim families, we examine how family members guide one another across religious, digital, and AI-mediated contexts. We find that authority did not simply follow age or family hierarchy: older members often remained sources of religious and familial guidance, while younger members increasingly provided expertise around digital risks, information credibility, and AI. Families also actively shaped digital environments and negotiated the role AI could occupy in everyday relationships. We develop the concepts of household epistemic governance, counter-curation, and domain-dependent authority to explain these shifting relations of knowledge and responsibility. Our findings suggest opportunities for family technologies that support negotiated boundaries, credible information practices, and culturally situated forms of care.
[HC-46] Latent World Action Model for GUI Agents
链接: https://arxiv.org/abs/2610.14228
作者: Seungjun Yu,Hojun Choi,Seojeong Park,Jaeyo Shin,Hyunjung Shim
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:
Abstract:Pixel-space world models have recently emerged as a promising approach for GUI agents, predicting future screens prior to execution and subsequently re-perceiving them for action selection. However, this two-stage paradigm faces two key challenges in GUI environments: (L1) the high computational cost of pixel reconstruction and (L2) the need to reconstruct large and diverse appearance-level details beyond what is required for action selection. In practice, this screen-generation objective can induce an identity shortcut problem, where future-state prediction largely preserves the current state rather than capturing action-induced changes. In this work, we reformulate GUI world modeling by shifting the prediction target from full-screen reconstruction to action-induced future information in latent space. Concretely, we leverage a large multimodal model as the latent world model, whose shared representations support two dedicated heads for both action selection and predictive transition modeling. The action head directly translates these latent representations into action selection, while the transition head captures action-conditioned future information. To mitigate the identity shortcut, we further introduce an action-aware contrastive objective that encourages the transition representations to capture discriminative action-induced changes. Extensive experiments across offline and online GUI settings show that our method achieves up to 31.9% and 40.6% relative gains in success rate over pixel-space world models, respectively, while substantially reducing inference latency by up to 98.9%. Our code is available at this https URL.
[HC-47] he AIs Poker Face: Contrasting Persuasion and Sycophancy in Voice-Based Multi-Turn Conversations
链接: https://arxiv.org/abs/2610.14124
作者: Ilia Frantzi,Bernie Hogan,Max M. Lang,Paul C. Bauer
类目: Human-Computer Interaction (cs.HC); Computation and Language (cs.CL)
备注:
Abstract:As interactions with large language models (LLMs) become increasingly routine, concerns about their capacity to influence users’ beliefs grow. Yet how such influence unfolds in sustained dialogue remains insufficiently understood. We conduct a randomized controlled experiment (N=616) where participants engage in multi-turn, spoken dialogue with an LLM under three conditions: non-directive, and two directive conditions maintaining opposing stances on a polarizing topic (the merits of feminism). We examine opinion change, rhetorical strategies, and sycophancy, distinguishing the semantic content of LLM argumentation from the pragmatic strategies used to manage the interaction. Both directive conditions shifted opinions toward their assigned stances, but asymmetrically: shifts were larger for pro-feminist positions, which also elicited distinct rhetorical strategies. The non-directive condition, likewise shifted opinions pro-feminism. Social sycophancy, though near-universal, was not associated with opinion change; instead it shaped users’ subjective experience. Our findings highlight distinct roles for argumentation and sycophantic accommodation in AI influence.
[HC-48] Refuses in Chat Complies with Arms: LLM Refusal Does Not Survive Robot Control
链接: https://arxiv.org/abs/2610.14043
作者: Edward Sun,Christopher Leet,Sravanthi Machcha,Achyutha Menon,Tzu Kit Chan,Yuchen Cui,Jay Chooi
类目: Human-Computer Interaction (cs.HC); Robotics (cs.RO)
备注: 7 pages
Abstract:Frontier language models reliably refuse to explain how to cause physical harm. We ask whether this refusal survives when the same models control a robot. We introduce RoboHarm, a physical AI safety benchmark of harmful robot instructions, and test seven frontier models on it. We hold each model fixed and vary only the interface, from text to Code as Policies to vision agents that command end-effector poses or joint angles. In a MuJoCo simulation twin of a bimanual YAM arm robot, refusal falls from 73% of text requests to 18-25% once the model controls the robot, even when it is told what the objects are. On the physical robot, six models with end-effector control refused 18% of 612 trials against 19% in sim, and three with joint control refused 41% of 252 against 31%. GPT-5.6 Sol refused every text request and none of its trials on the real robot. Case studies of the robot transcripts show that models miss the hazard, name it without inferring its consequences, or act despite recognizing it, and their refusals often hinge on uncertainty about what an object is rather than on the harm of the action. Safety alignment learned in text thus fails to generalize to action, leaving current models with a poorly understood safety gap in the physical world.
[HC-49] Showing versus Doing: Teaching by Demonstration NEURIPS2016
链接: https://arxiv.org/abs/2610.14004
作者: Mark K. Ho,Michael L. Littman,James MacGlashan,Fiery Cushman,Joseph L. Austerweil
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注: Published at NeurIPS 2016 (oral presentation). 12 pages, including supplementary materials
Abstract:People often learn from others’ demonstrations, and inverse reinforcement learning (IRL) techniques have realized this capacity in machines. In contrast, teaching by demonstration has been less well studied computationally. Here, we develop a Bayesian model for teaching by demonstration. Stark differences arise when demonstrators are intentionally teaching (i.e. showing) a task versus simply performing (i.e. doing) a task. In two experiments, we show that human participants modify their teaching behavior consistent with the predictions of our model. Further, we show that even standard IRL algorithms benefit when learning from showing versus doing.
[HC-50] EmoXpt: Analyzing Emotional Variances in Human Comments and LLM -Generated Responses
链接: https://arxiv.org/abs/2501.06597
作者: Shireesh Reddy Pyreddy,Tarannum Shaila Zaman
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC)
备注: 7 pages, 10 figures, 5 tables. This paper has been accepted and presented at the 2025 IEEE 15th Annual Computing and Communication Workshop and Conference (CCWC)
Abstract:The widespread adoption of generative AI has generated diverse opinions, with individuals expressing both support and criticism of its applications. This study investigates the emotional dynamics surrounding generative AI by analyzing human tweets referencing terms such as ChatGPT, OpenAI, Copilot, and LLMs. To further understand the emotional intelligence of ChatGPT, we examine its responses to selected tweets, highlighting differences in sentiment between human comments and LLM-generated responses. We introduce EmoXpt, a sentiment analysis framework designed to assess both human perspectives on generative AI and the sentiment embedded in ChatGPT’s responses. Unlike prior studies that focus exclusively on human sentiment, EmoXpt uniquely evaluates the emotional expression of ChatGPT. Experimental results demonstrate that LLM-generated responses are notably more efficient, cohesive, and consistently positive than human responses.
计算机视觉
[CV-0] LoopWAM: Closed-Loop Generalist Policy Improvement from Deployment Rollouts
链接: https://arxiv.org/abs/2610.18873
作者: Wenxuan Guo,Yingping Liang,Zhaoguo Wang,Ziyuan Li,Angyuan Ma,Zhenyu Wu,Chengkun Wang,Bingyao Yu,Ze Chen,Borui Zhang,Xiuwei Xu,Meng Zhang,Erjin Zhou,Jiwen Lu
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Scaling expert demonstrations alone does not enable generalist robot policies to improve from deployment. Real-world rollouts contain successes, suboptimal behavior, and failures: imitating every recorded action can reinforce poor behavior, while filtering out failures discards evidence of physical interactions and their consequences. We introduce \textbfLoopWAM, a closed-loop world-action model that learns directly from this full quality spectrum, using rollouts from both itself and other deployed policies. Source-dependent noise sampling unifies behavior learning and action-conditioned world modeling, while quality steering and quality-weighted supervision retain full video supervision across qualities and regulate action imitation. Video and action tokens interact throughout parallel denoising, allowing learned consequences to inform action generation directly. On RoboChallenge Table30-V2, three deployment-and-training rounds raise success from 23.67% to 43.33%, a 19.7-percentage-point gain, with a final score of 56.80. LoopWAM also achieves 94.5% average success on RoboTwin~2.0 and 49.5% overall success on RoboCasa365 without additional embodied pretraining, obtaining the best overall results among the compared methods across all three benchmarks. On challenging shirt folding task, two rounds of deployment-driven improvement after single-task post-training raise success from 13.20% to 55.56%. These results demonstrate closed-loop improvement from mixed-quality experience under both multi-task learning and single-task post-training. \hrefthis https URL\textcolororangeProject Page
[CV-1] EgoExplorer: A Vision-Motion-Touch Observatory for Human Loco-Manipulation In Situ
链接: https://arxiv.org/abs/2610.18872
作者: Yunsong Zhou,Xuekun Jiang,Haiqi Shen,Hangxu Liu,Chaojian Shi,Yujie Liu,Jingyuan Li,Haoran Yang,Jiafei Cao,Dongzhan Zhou,Dong Wang,Yao Mu,Chunhua Shen,Weinan Zhang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this http URL
Abstract:Learning humanoid loco-manipulation from human demonstrations rests on observing how locomotion, posture, and object interaction evolve together as tasks unfold. Scaling collection across diverse environments, in turn, favors lightweight, portable capture systems. Yet, existing datasets provide only partial views of this coordination. Thus, we introduce EgoExplorer, a 100-hour vision-motion-touch observatory of human loco-manipulation built with a wearable system. With an emphasis on professional activities, its continuous recordings align egocentric video with reconstructed whole-body motion, fine-grained finger articulation, and tactile observations, accompanied by step-level descriptions. To achieve this coverage with sparse body sensing, we develop the Calibrate-Complete-Constrain pipeline. Concretely, reusable calibration fixes participant geometry and sensor attachments before motion inference to disentangle mounting offsets from articulation. A temporal prior then completes unobserved joint motion, while layered constraints promote measurement agreement, temporal continuity, and geometric consistency, properties that motion plausibility alone does not guarantee. Against tracker-fitted references, the method reduces body-point error by 71.9% relative to the strongest baseline; matched evaluations also show gains in hand stability and contact estimation. Preliminary studies explore multimodal generation and conditioning alongside humanoid retargeting with whole-body control in simulation.
[CV-2] Cepheid: Decoupling Scale from Structure for Cross-Family Motion Completion
链接: https://arxiv.org/abs/2610.18871
作者: Chih-Hai Su,Tsung-Wei Ke,Yu-Lun Liu
类目: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR); Machine Learning (cs.LG)
备注: 31 pages. Project page: this https URL
Abstract:Sparse-to-dense motion completion recovers dense mesh motion from sparse vertex displacements. Residual priors trained on one geometry family frequently fail to beat parameter-free harmonic extension off-family. Moreover, cross-family training meets a scaling barrier: residual amplitude varies by more than an order of magnitude across geometry families and collapses as observations become denser. Since this scale depends on unknown target motion, it cannot be measured directly at test time. We decouple residual structure from physical scale by learning a dimensionless prior, estimating amplitude from observables and a lightweight correction fitted only on training sources. Our unified model surpasses harmonic extension across all evaluated families, including unseen cloth and legged bodies, without skeletons, registration, or cross-mesh correspondence. The same prior applies without retraining to point-tracked animal video and a robot grasping a soft body. Project page: this https URL
[CV-3] Dog-Breed Text-to-image Flow Matching from Pretraining to Fine-Tuning on 4 Consumer GPUs
链接: https://arxiv.org/abs/2610.18870
作者: Reyhaneh Esmailizadeh,Xingjian Leng,Zhanhao Liang,Liang Zheng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We generally perceive that training text-to-image diffusion models requires heavy compute: it is true if we aim at a comprehensive frontier model. This understanding prevents people from training their own models when getting started in the field. We develop MiniDog, a framework that enables pretraining diffusion transformers in 55 minutes and fine-tuning in another 15 minutes on four 3090 GPUs. For data, we use 26k dog images from ImageNet and caption them for pretraining and use 2k images generated by FLUX.1-schnell for fine-tuning. We precompute the image and text features and cache them for efficiency. For architecture, we use E2E-INVAE for its semantic latent space, a small transformer with 22M parameters, a small Qwen model for text encoding, and in-context conditioning. For training, we use iREPA and an EUPE perception encoder for acceleration. For evaluation, our repository supports FID, HPSv2 score, PickScore, and various VAE reconstruction metrics.
[CV-4] PLAYGROUND: Progressive Layout Generation with Render-Grounded Planning
链接: https://arxiv.org/abs/2610.18865
作者: Jaeho Lee,Junhwan Heo,Junghyun Park,WonJun Moon,Eunju Yang,Seungho Jang,Seongchan Kim,Seungryong Kim
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Generating editable graphic designs requires arranging visual and textual elements into a coherent composition while preserving their content and editability. Although recent MLLM-based approaches employ iterative generation and visual refinement to reflect the real-world design workflows, their high-level plans are predetermined prior to execution, limiting their adaptability as the design evolves. To address this limitation, we propose PlayGround, a progressive layout generation framework that dynamically adapts to the current rendered canvas. PlayGround establishes a progressive design loop comprising a Progressive Planner, a Layout Generator, and a Visual Verifier, guided by a Structure-Aware Retriever. Given an initial set of unordered elements, the Structure-Aware Retriever initially infers a coarse layout structure to retrieve structurally similar reference designs. Subsequently, at each turn, the Progressive Planner examines the current rendered canvas alongside the remaining elements, forms the next functional group, and plans its composition. The Layout Generator then predicts editable layout attributes, while the Visual Verifier evaluates rendered candidates and provides feedback when refinement is necessary. To train the Progressive Planner with step-wise supervision, we construct a two-stage data generation pipeline that extracts functional groups and turn-level planning trajectories from completed designs. Experiments on Crello and LICA demonstrate the effectiveness of PlayGround over existing methods. Our code and weights will be publicly released.
[CV-5] Visual Tokens on Demand: Visual Budgeting for Scalable and Efficient Multi-Image Reasoning in Multimodal LLM s
链接: https://arxiv.org/abs/2610.18858
作者: Aimon Rahman,Joymallya Chakraborty,Vidit Bansal,Abhishek Dan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal large language models have advanced rapidly, but most remain optimized for single-image inputs. Multi-image reasoning is costly because each image adds many visual tokens, increasing encoding and attention demands. Existing techniques that address the above issue, including visual token pruning, token merging, visual token compression, learned resampling, hierarchical compression, retrieval-augmented visual reasoning, and memory-based architectures, have largely been designed for single-image scenarios, where redundancy is primarily spatial and local. Multi-image reasoning introduces additional challenges: redundancy may occur across images, task-relevant evidence may be sparse or distributed, and aggressive per-image compression can remove details needed for comparison, counting, temporal reasoning, retrieval, or evidence grounding. Moreover, existing token-reduction methods are often either (i) training-free but inference-inefficient, typically relying on accessing attention matrices to get token-importance that may be impractical to access with memory-efficient attention implementations such as FlashAttention, or (ii) training-dependent, requiring the base MLLM to be retrained or adapted to a new compressed-token paradigm. In this work, we study scalable multi-image reasoning and understanding without retraining the base model and without accessing attention matrices at inference time. To this end, we propose mmVB, a lightweight plug-and-play module that assigns query-conditioned per-image token budgets via cross-image context, adding only 1.6M trainable parameters while reducing visual input tokens by 9x. This reduction yields 5x faster time-to-first-token and 7x fewer prefill FLOPs, while maintaining 94.4% of full-resolution accuracy across eight multi-image reasoning benchmarks.
[CV-6] PixelMoE: Scaling Pixel-Space Diffusion with LLM -Style Mixture-of-Experts
链接: https://arxiv.org/abs/2610.18857
作者: Ruixiao Dong,Jianmin Bao,Bozhi Luan,Zijie Meng,Xin Tao,Pengfei Wan,Kun Gai,Li Li,Houqiang Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Pixel-space diffusion models provide an appealing alternative to latent-space diffusion models by directly modeling raw images without relying on a learned tokenizer. However, scaling pixel-space models has predominantly followed dense strategies, tightly coupling model capacity with active computation. In this work, we show that standard LLM-style Mixture-of-Experts (MoE) architectures can be effectively transferred to pixel-space diffusion. Building on this finding, we introduce PixelMoE, a pixel-space generative framework that sparsely scales model capacity with standard LLM-style MoE, without diffusion-specific routing mechanisms or auxiliary routing objectives. Specifically, PixelMoE combines standard Token-Choice (TC) routing with fine-grained routed experts for input-adaptive sparse computation, a shared expert for capturing common representations, and a bias-based load-balancing strategy. To complement sparse capacity scaling, PixelMoE further introduces Pixel-Space Hierarchical Flow Matching, which varies spatial resolution along the denoising trajectory from global structure formation to fine-grained detail refinement. At comparable active parameter counts, PixelMoE consistently outperforms its dense counterparts across model scales, demonstrating effective sparse capacity scaling in pixel space. On class-conditional ImageNet-1K, PixelMoE achieves FID scores of 1.47 at 256x256 and 1.53 at 512x512 without representation alignment or perceptual losses, comparing favorably with leading pixel-space and latent-space generative models. Code and models will be released soon.
[CV-7] WorldRecon: Generative Multi-View Mesh Reconstruction for Large-Scale 3D World
链接: https://arxiv.org/abs/2610.18846
作者: Mengchen Zhang,Yang Li,Chunshi Wang,Tong Wu,Chunchao Guo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Reconstructing complete, high-quality meshes of large-scale scenes from unposed multi-view images requires faithful geometric grounding while plausibly completing surfaces that are unobserved or only sparsely captured. We present WorldRecon, a generative multi-view reconstruction framework for recovering complete, large-scale 3D scene meshes from unposed RGB images. WorldRecon uses a coarse, noisy global point cloud and camera poses estimated from the input images by a reconstruction model as a global scaffold for whole-scene generation and partitions it into overlapping spatial blocks. To robustly leverage these imperfect geometric priors, we equip the native 3D generative model TRELLIS.2 with a dedicated reconstruction-based conditioning architecture that generates each block from its local point-cloud crop and block-relevant image observations, producing sparse structure, detailed geometry, and PBR materials. To maintain scene-level coherence, we introduce neighbour-aware latent inpainting and a causal world-generation procedure that propagates structural, geometric, and material context across overlapping blocks. Together, these designs scale multi-view generative reconstruction to large-scale scenes, yielding coherent, unified meshes. Extensive experiments demonstrate that WorldRecon achieves state-of-the-art reconstruction performance, reducing Chamfer distance by 14.9% and 43.1% while improving F-score by 17.3% and 12.0% over the strongest baselines on synthetic and real-world indoor scenes, respectively, thereby demonstrating faithful and complete generative reconstruction as well as strong sim-to-real generalization. Project page: this https URL.
[CV-8] RepProbe: Can Video Language Models Count Repeated Actions? A Diagnostic Benchmark
链接: https://arxiv.org/abs/2610.18844
作者: Mahsa Khoshnoodi,Sarah Adel Bargal
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Repetition counting requires identifying and enumerating action cycles over time. While state-of-the-art video language models describe actions fluently, they fail to reliably count how many times an action repeats. To investigate this limitation, we introduce RepProbe, a benchmark of 42K real and synthetic video clips containing exactly k in [1, 20] complete repetitions, with controls for clip position, temporal coverage, and duration. These controls allow us to study the counting failure directly and locate where in the model it arises. Counting accuracy collapses at high counts across the video language models we evaluate, and controlled experiments show this collapse is not explained by frame sampling alone. We compare model outputs with linear probes trained on the final-layer representations used to generate those outputs. We find a consistent representation-to-output dissociation: substantial count information remains linearly decodable even when models fail to report the correct count. Further analysis shows this failure arises primarily at the output stage, where answers concentrate on a small set of preferred numbers rather than the available count information. The dissociation holds on both real and synthetic videos and is not explained by clip duration, position, or scene content. These results indicate that models encode repetition structure but do not reliably report it in their outputs. Project page: this https URL
[CV-9] Scaling Online Optimal Transport Minibatches for Flow Matching
链接: https://arxiv.org/abs/2610.18838
作者: Peter Chen,Sergey Levine
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: 37 pages
Abstract:A central challenge in scaling flow models is maintaining stability in training and integration. In flow matching, this stability depends critically on how source samples are paired with data. Independent random pairing introduces substantial stochasticity, motivating minibatch optimal transport (OT) to construct online source-target couplings that reduce this randomness (Tong et al., 2024). However, as applications extend to larger and higher-dimensional datasets, small minibatches can limit these benefits, while scaling the coupling to substantially larger minibatches at every iteration is computationally prohibitive. Moreover, how OT coupling affects the full stochastic gradient and integration remains theoretically unresolved. In this work, we develop a scalable online entropic OT approximation that enables substantially larger source-target couplings while preserving the benefits of OT. We close the theoretical gap for both unregularized and entropic OT by proving strict reductions in full stochastic-gradient variance and integration error relative to independent pairing, and further characterize how these improvements strengthen as the online coupling minibatch grows. Experiments on single-cell gene interpolation-flow and image-flow training validate these predictions, showing that larger online couplings consistently improve training stability and final generation performance.
[CV-10] Plan-Bench: Can Existing VLMs Go Well Beyond Perception to Reason and Plan in Embodied Worlds?
链接: https://arxiv.org/abs/2610.18836
作者: Juntong Feng,Qingan Liu,Jingru Dong,Yue Zhu,Weiming Zheng,Yujie Cheng,Xinzhe Han,Shuai Hao,Yunzhi Zhuge,Haiwen Diao,Huchuan Lu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 24 pages, 11 figures, 10 tables
Abstract:Vision-Language Models (VLMs) have achieved strong performance in visual perception and spatial reasoning, leading to growing interest in using them as high-level planners for embodied AI. However, physical tasks require models to integrate spatial information, reason over temporal states, and generate multi-step plans under various constraints, raising the question of whether current VLMs can reliably support such reasoning and planning independently of execution. We introduce Plan-Bench, a benchmark for evaluating the high-level reasoning and multi-step planning capabilities of VLMs relevant to embodied worlds, while isolating the influence of action execution. It covers three complementary dimensions, spanning 16 subclasses and 699 curated samples. Extensive evaluation of 28 VLMs reveals room for improvement, with even the strongest model achieving only 75.0% accuracy. We further find that model scaling generally improves these capabilities. Ablation studies also indicate larger models show increasing potential as high-level planners. Overall, Plan-Bench provides a lightweight probe for evaluating VLMs’ reasoning and planning capabilities relevant to embodied environments and assessing their potential as decision makers for physical tasks. Our benchmark and code are available at this https URL.
[CV-11] Game2Real-Bench: Do Toy Games Reveal and Transfer to Real-World Visual Reasoning ?
链接: https://arxiv.org/abs/2610.18834
作者: Qingan Liu,Juntong Feng,Yue Zhu,Yujie Cheng,Xinzhe Han,Shuai Hao,Yunzhi Zhuge,Haiwen Diao,Huchuan Lu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 32 pages, 8 figures, 11 tables
Abstract:Visual reasoning is an important capability for vision-language models (VLMs) to understand visual evidence, spatial structure, constraints, and state changes. Existing evaluations mainly rely on real-world data or procedurally generated tasks. Real-world data preserve natural complexity, while procedural games provide controllable generation and verifiable solutions. However, it remains unclear whether abilities measured in toy games correspond to real-world visual reasoning and whether game examples can support transfer. We introduce Game2Real-Bench, a benchmark that studies these connections across four capability groups, including fine-grained perception, spatial reasoning, constraint reasoning, and sequential planning. We construct 1,000 procedurally generated game samples with executable verification and curate 400 real-world samples aligned by shared reasoning requirements rather than visual similarity. We evaluate 25 representative VLMs and observe strong game-to-real consistency, with Pearson and Spearman correlations of 0.935. We further study one-shot game-to-real transfer on four models. Constraint reasoning improves for several models, while fine-grained perception declines with game demonstrations. We also construct Game2Real-Dataset with nearly 24K verified game samples across 18 tasks and 65 QA families. Our benchmark, dataset, and code are available at this https URL.
[CV-12] WorldOracle: Benchmarking Action-Conditioned World Models against Real-World Executions
链接: https://arxiv.org/abs/2610.18826
作者: Haoyi Duan,Liliang Chen,Nan Wang,Yang Zhang,Yuhang Lu,Xiaotong Chen,Wei-Cheng Tseng,Shenyuan Gao,Zhaoshuo Li,Jingyi Jin,Pooya Jannaty,Dhruv Shah,Yuke Zhu,Ming-Yu Liu,Maoqing Yao,Guanghui Ren
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We introduce WorldOracle, a benchmark specifically designed to evaluate action-conditioned world models (ACWMs) against paired real-world executions. Given the same initial observation and low-level action sequence, WorldOracle asks whether a predicted rollout reproduces what happens in the real world, including physical consequences and task-state transitions. We evaluate models along four complementary dimensions: physical state correctness, interaction spatial alignment, key interaction area fidelity, and multi-view action coherence. To support this evaluation, we curate real-world training and evaluation suites that cover different interaction dynamics and embodiments. The evaluation suite contains long, multi-segment episodes with both successful and failed executions, enabling evaluation across varied action outcomes and behavior distributions. Experiments on seven representative ACWMs reveal distinct model strengths and failure modes. WorldOracle provides a systematic foundation for measuring and advancing action-conditioned world modeling for manipulation. We will release the datasets, evaluation code, and leaderboard.
[CV-13] DuplexBridge: Coordinating Understanding and Generation for Unified Image Editing
链接: https://arxiv.org/abs/2610.18819
作者: Yidi Liu,Jie Xiao,Chaojie Mao,Jingfeng Zhang,Xueyang Fu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Complex image editing requires faithful task interpretation and accurate execution, but a failed edit does not reveal which responsibility failed. We present DuplexBridge, which uses a unified model’s own understanding capability to enhance prompts for its own generation capability, without an external understanding module. A source-grounded bridge specifies targets, final-state changes, and preservation constraints. Understanding is supervised on fidelity to the original request without observing the generated image; generation is supervised on execution of the condition it actually receives. Reciprocal trust checks selectively admit global task feedback to each role while retaining independently valid local supervision. We introduce CE-Bench, a 1,200-task benchmark measuring joint satisfaction of task-derived change and preservation requirements. Across three model configurations, the full system improves complete compliance over direct editors by 21.75-25.75 percentage points, with gains also observed on KRIS-Bench, RISEBench, ImgEdit, and GEdit-Bench. From a common supervised Wan-Image initialization, gated joint training reaches 77.08% complete compliance, compared with 70.92% without gates and 75.00% when generation-local supervision instead checks the original task with both gates retained. These results support distinct local responsibilities and selective shared feedback for joint understanding-generation adaptation.
[CV-14] Registration is a Representation Problem: Solving Whole-Body PET/CT Registration MICCAI WWW
链接: https://arxiv.org/abs/2610.18804
作者: Neel Dey,Adrian V. Dalca
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Technical report. First place winning solution for the MICCAI Learn2Reg 2026 PSMAReg challenge. Code and volumetric visualizations are available at this http URL . Keywords: representation learning, medical image registration
Abstract:Raw image intensities are a poor surrogate for similarity across images in correspondence problems. Further, while auxiliary segmentation labels can provide semantic identification and improve matching, they can assign only coarse identities to fine-grained anatomy and often cover only a subset of structures, particularly in whole-body settings. We present an approach that instead establishes correspondence using complementary representations of anatomy and demonstrate its utility for longitudinal whole-body PET/CT registration in the Learn2Reg 2026 PSMAReg Challenge. Our 3D foundation model, anatomix, is used to derive fine-grained, per-voxel semantic representations that are consistent across volumes, and we further employ off-the-shelf segmentation models to obtain coarse anatomical identities. Together, these representations drive a GPU-accelerated iterative registration solver (FireANTs) in a hybrid multistage framework: the first stage estimates alignment from dense multi-channel features, after which a second refines the deformation using both anatomix- and segmentation-based losses. Lastly, because registration should not introduce non-biological tumor volume changes across time, we develop lesion-specific component-wise volume preservation schemes. Without any task-specific training, our system won first place in the PSMAReg Challenge leaderboard. Code and interactive visualizations are available at this http URL.
[CV-15] rust Before Autonomy: Rethinking Agent ic Remote Sensing for Crisis Response
链接: https://arxiv.org/abs/2610.18789
作者: Georgios Angelidis,Aggelos Psiris,Spyridon Evangelatos,Thomas Lagkas,Panagiotis Sarigiannidis,Iraklis Varlamis,Georgios Th. Papadopoulos
类目: Computer Vision and Pattern Recognition (cs.CV); Computers and Society (cs.CY)
备注:
Abstract:Agentic remote sensing is evaluated on how many questions it can answer, whereas crisis response depends on whether its answers can be relied on and how much of the decision can safely be handed to the agent. This paper argues that \emphautonomy should be earned, not assumed: An Earth-Observation (EO) agent should be granted autonomy in proportion to demonstrated reliability, epistemic honesty and verifiability. The position is supported by an evidential meta-analysis of twelve agentic RS/EO/GIS benchmark studies from peer-reviewed venues. Only one repeats runs, three contain unanswerable tasks and none measures calibration. Their own tables show that completed workflows are often not correct (task success exceeds tool-call correctness by up to 54pp on identical queries), that the gap to an expert trajectory widens overall with trajectory length (from 7 to 56pp), and that capability does not predict abstention (Spearman r from -0.25 to 0.50, all n.s.). The paper then contributes three EO-specific trust properties, an autonomy ladder for crisis EO tasks, and a readiness protocol that replays archived emergency activations before any live use. In this context, current evidence justifies assistive roles only.
[CV-16] GAD: Geometry Attention Distillation for Consistent Video Generation
链接: https://arxiv.org/abs/2610.18779
作者: Junfeng Ni,Ziyu Zhu,Fan Jiang,Yu Liu,Chiyu Wang,Jiawei Zhang,Zhaoxu Sun,Mu Xu,Siyuan Huang,Yixin Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Video diffusion models generate visually compelling short-term sequences but often struggle to preserve scene geometry and object structure over time. In this paper, we first demonstrate that current video attention cannot identify reliable correspondences across frames, with responses dispersed across multiple regions even after recognizable scene structure has emerged. We argue that reliable cross-frame associations are essential for information exchange as scene structure forms in video generation. Thus, we introduce Geometry Attention Distillation (GAD), which transfers correspondences from a frozen geometry model’s attention to video attention. Extracted from clean videos, these correspondences link the same static surface or moving object across frames. Since structure formation begins in deeper layers at higher noise and shallower layers at lower noise, we apply attention distillation at a student layer selected according to the noise timestep. After training, video attention focuses more consistently on corresponding locations. Experiments across all three video backbones demonstrate improved long-range scene consistency and dynamic temporal consistency, with geometric gains across generation lengths. On Wan2.2 5B, GAD improves cross-frame matching accuracy by at least 30% over the Wan baseline on both long-range benchmarks. Notably, GAD achieves these gains without additional inference modules, offering a promising approach to efficient and consistent video-based world simulation. Project page: this https URL.
[CV-17] he Federated Tumor Segmentation and Classification (FeTS) Challenge
链接: https://arxiv.org/abs/2610.18771
作者: Wenyi Tang,Carla Pitarch-Abaigar,Siddhesh Thakur,Verena Chung,Konstantinos Koukoutegos,Alexandros Karargyris,Holger R. Roth,Spyridon Bakas
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:International challenges have become the standard for developing and benchmarking biomedical image analysis methods using centralized multi-institutional data. Federated learning (FL) describes a paradigm-shift on multi-institutional collaborations, where models are trained across institutions while their data are always retained locally. The Federated Tumor Segmentation (FeTS) challenge, first held in 2021, was the first challenge dedicated to FL in medical imaging and editions in 2022 and 2024 pushed the frontiers of real-world method validation. FeTS++ 2027 returns after a two-year gap with a substantially larger and more diverse cohort, and with the two “pluses” that name the challenge: an official classification task alongside segmentation, and a move of the infrastructure to NVIDIA FLARE (NVFlare). Built on the clinically acquired, multi-institutional multi-parametric MRI data of the BraTS 2025 Lighthouse challenge, FeTS++ spans gliomas and meningiomas across patient cohorts from institutions in North America, Europe, Asia, and Africa, and focuses on ranking submitted methodological approaches according to their model utility and an efficiency-aware score. Task 1 asks participants to find the best strategy for aggregating knowledge from per-institution segmentation models, while Task 2 asks them to distinguish glioma from meningioma under the same federated protocol, in which the entity label is heavily confounded with the site of origin.
[CV-18] MIST: Multi-Scale Spatial Context for Gene Expression Prediction from Histology
链接: https://arxiv.org/abs/2610.18769
作者: Koushik Howlader,Tirtho Roy,Md Tauhidul Islam,Wei Le
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Predicting spatial gene expression from standard HE images could extend molecular analysis to tissues without spatial transcriptomic measurements. A standard Transformer applies self-attention over spots but ignores where each spot sits in the tissue, whereas a spot’s molecular state depends on tissue context at multiple spatial scales. Building on this insight, we introduce MIST, a Multi-context Spatial Transformer that augments the self-attention backbone with two context pathways applied in every layer: a coordinate-invariant local pathway that attends over spatial k-nearest-neighbor spots, and a slide pathway that broadcasts a whole-slide summary to every spot; the model predicts all spots in a single forward pass. Because coordinates enter only through pairwise distances, the local pathway is invariant to translations, rotations, and reflections when image features are fixed. In intra-cohort evaluation across six human organ cohorts from HEST-1k, MIST improves mean Pearson correlation over the strongest evaluated baseline by 9.7%, 11.6%, and 11.4% for 10-, 50-, and 100-gene panels, respectively, with the lowest average MSE and MAE. MIST further improves leave-one-organ-out transfer by 6.0% over the next strongest baseline, attains the best expression-derived domain recovery (highest mean ARI and NMI at every clustering resolution, and higher NMI than the closest baseline on 84% of held-out slides), and reproduces spatial expression patterns of selected cancer-marker genes. Together, these results show that adding local and slide context to a vanilla Transformer backbone is an effective approach to spatial gene expression prediction from histology.
[CV-19] LayerCanvas: Infographic Generation in AI Learning Assistants via Hybrid Layer-Set Diffusion
链接: https://arxiv.org/abs/2610.18763
作者: Keshav Gupta,Jeremy Owen,Christopher Lloyd,Shaunak Mishra,Manav Sharma
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Online learning platforms are increasingly deploying AI assistants to provide personalized, on-demand tutoring. These can leverage infographics to communicate complex information and improve the learning experience. However, automating effective infographic generation remains a challenging task, requiring a combination of visual richness with precise typography, structured layouts, and continued editability. Existing representation methods satisfy these goals only in part: Scalable Vector Graphics (SVG) representations preserve geometry, hierarchy, and text but provide limited aesthetic flexibility, while raster diffusion models produce richer appearances by flattening content and layout into a single image. Conditioning a diffusion model on a rasterized SVG does not resolve this conflict because the raster preserves approximate placement but discards element ownership, occlusion relationships, and the structure needed to modify and recompose the result. We introduce LayerCanvas, a hybrid approach that retains SVG as the semantic, structural, and typographic program while using diffusion to aestheticize non-text content. Rather than generating one flattened raster, LayerCanvas partitions SVG elements into z-ordered sets of mutually non-overlapping elements and processes them through communicating parallel RGB generation streams. The resulting layers are independently editable and deterministically recomposed, while text is rendered from preserved vector metadata. Our approach combines rich visual synthesis with accurate typography and localized editing. LayerCanvas substantially improves text fidelity over state-of-the-art raster image generators while achieving the highest overall aesthetic quality. We further demonstrate non-destructive content editing, localization, asset reuse, and animation.
[CV-20] Context-aware Bayesian Fusion for Multimodal Object Detection
链接: https://arxiv.org/abs/2610.18741
作者: Arunachalam Venkatachalam,Mhafuzul Islam,Marwan Yusuf,David Van Hamme,Ben Stoffelen,James Foresi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Reliable detection of pedestrians and road users is essential for the safe operation of autonomous vehicles. However, this task becomes increasingly challenging under adverse environmental conditions, such as poor weather and low lighting, where the performance of state-of-the-art pre-trained object detectors and on-board sensors tends to degrade, which limits their effectiveness and reliability. Moreover, retraining models for each operating context imposes significant constraints due to high data requirements, computational resources, and time. Considering this, we propose a novel context-aware multimodal late fusion approach that integrates a lightweight gating network within our probabilistic calibration framework. The gating network extracts environmental context and dynamically modulates the detection outputs of pre-trained object detectors based on the inferred context, without retraining of the detectors. Our methodology is evaluated on a modified KITTI dataset, in which the camera images and lidar pointclouds are augmented for varying fog levels. Experimental results show that our approach yields an approximately 14% relative improvement over non-adaptive late fusion baselines, achieving a consistent average precision of 56% across different fog levels.
[CV-21] Concealed Handgun Detection using WiFi Signals
链接: https://arxiv.org/abs/2610.18736
作者: Erik Persson,David Gustafsson,Tomas Wilkinson
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:While conventional camera-based surveillance systems and metal detectors can provide a sufficient level of security when used appropriately, they have fundamental limitations that reduce their applicability. Camera-based systems are unable to detect occluded or concealed threats and metal detectors require each person to enter through a narrow opening one at a time, leading to congestion issues and limited use in large crowds. A WiFi-based security system has the potential to be a complementary modality that is able to handle occlusions or concealment and is not hindered by strict spatial configurations. In this paper, we propose use Channel State Information estimated from WiFi-signals to detect concealed handguns. To this end we collect a dataset of aligned WiFi signals and RGB images that we call the WiFire dataset. We also propose a deep learning model called WiFireNet tailored to detect concealed handguns in WiFi signals using vision-based supervision signals. Our results show that concealed handgun detection is possible in WiFi and has promise as a complementary modality to surveillance cameras and metal detectors. We release the dataset, model and code to facilitate the further research into WiFi-based handgun detection.
[CV-22] Neural Block-wise Auto-exposure for Single-Shot High Dynamic Range Imaging
链接: https://arxiv.org/abs/2610.18733
作者: Zhengguo Li,Chaobing Zheng,Wei Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: High dynamic rang imaging, Single-shot, Block-wise, Auto-exposure, Neural network
Abstract:Block-wise varying exposures are effective for cap- turing appearance of a real-world high dynamic range (HDR) scene via a single-shot. However, brightness order reversal (BOR) artifacts, blocking artifacts and loss of details in highlight/shadow regions are possible issues for block-wise varying exposures. In this paper, a novel neural block-wise auto-exposure algorithm is proposed to address the issues. The proposed algorithm maps a temporary image and its exposure time to the exposure times of all non-overlapping 16 X 16 blocks. The temporary image can be either an sRGB or a raw image. Although the sRGB image is widely adopted in existing auto-exposure algorithms, exposure times of blocks in the highlight and shadow regions might not be accurately learned due to the limited radiance measurement capability of the sRGB image. This paper thus also explores the feasibility of temporary image as a raw image. Due to the better radiance measurement capability of the raw image, the exposure times of blocks are learned more accurately, and more details in the highlight/shadow regions of the real-world HDR scene are captured by the resultant HDR imaging system. Experiment results validate the proposed neural block-wise auto-exposure algorithm.
[CV-23] Segment Where You Look: Attention as the Interface Between Language Models and Segmenters
链接: https://arxiv.org/abs/2610.18729
作者: Eslam Abdelrahman,Hamza Omar Alkaf,Yuheng Li,Siddharth Srinivasan Iyer,Nanxuan Zhao,Eli Shechtman,Krishna Kumar Singh,Mohamed Elhoseiny
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-language models (VLMs) that segment from language must pass the location of the referent to a pixel decoder. The dominant interface compresses it into the hidden state of a single [SEG] token, a d-dimensional vector with no spatial index, and leaves the decoder, which never sees the expression, to recover the referent’s extent. We show that the VLM already computes a better signal: the [SEG] token’s attention over the image tokens, which is defined on the patch grid by construction. Probing a frozen segmentation VLM, we find that this attention yields 18 to 22 more gIoU than the [SEG] hidden state when read by the same probe, that it gains nothing from the image where the hidden state does, and that it is concentrated in the late layers. Building on these findings, we introduce AttnSeg, which replaces the [SEG] embedding with this attention as the decoder’s prompt. A learned, signed combination of the attention heads forms a dense spatial prompt, and an attention-guidance objective supervises it directly. A two-stage recipe first trains the VLM to localize without the decoder, then trains the decoder to refine, within the budget of standard fine-tuning. Applied to Sa2VA, AttnSeg improves ReasonSeg by 6.4 gIoU and GS-Eval by 10.2 at 4B (4.2 and 9.6 over the same model fine-tuned on the same data), and by 8.8 and 11.3 at 14B. The gains reach 13 to 18 gIoU on object parts, multiple objects, and single objects, where the referent must be singled out, and are negligible on stuff regions, which the decoder already recovers unaided. Code and models are available at this https URL
[CV-24] UniW(A)M: A Unified Mixture-of-Experts Framework for World and World-Action Models ICLR2027
链接: https://arxiv.org/abs/2610.18697
作者: Jingbo Xu,Jianjie Fang,Zhuohang Li,Zihao Song,Zhaolu Wang,Xvyuan Liu,Xin Zhang,Haisheng Su,Yinzhou Tang,Chen Gao,Wei Wu,Yong Li
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Under review at ICLR 2027. 34 pages, 19 figures, 15 tables
Abstract:World models (WMs) and world-action models (WAMs) focus on future world prediction and executable action generation, respectively. Although these models learn shared physical dynamics and interaction semantics from embodied trajectories, they are typically trained in isolation, which restricts knowledge reuse between prediction and control. The central challenge is therefore to develop a unified learning framework that shares these dynamics while preserving the distinct modeling capabilities required by each paradigm. We propose UniW(A)M, a mixture-of-experts world-(action) model built upon diffusion Transformers that unifies action-conditioned world modeling and diverse WAM variants. Shared experts capture common dynamics, while atomic capability experts are reused across interfaces so that each interface keeps its own conditional pathway. At deployment, a plug-in adaptive interface router (PAIR) learns from a small subset of target-task chunk scenarios to select a suitable WAM interface for each execution chunk, balancing task success and execution efficiency while keeping the backbone frozen. We evaluate UniW(A)M on LIBERO, RoboTwin 2.0, and three real-world manipulation tasks, where it achieves 88.3% average real-world success versus 76.7% for the strongest baseline. On WorldArena, it leads action-conditioned world modeling on all three tasks. AC-WM co-training contributes to action generation on every action interface and PAIR raises OOD success from 63.3% to 76.7%. Comments: Under review at ICLR 2027. 34 pages, 19 figures, 15 tables Subjects: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) Cite as: arXiv:2610.18697 [cs.RO] (or arXiv:2610.18697v1 [cs.RO] for this version) https://doi.org/10.48550/arXiv.2610.18697 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-25] CryoTracer: A Synthetic-to-Real System for Filament Instance Tracing in Cryo-EM
链接: https://arxiv.org/abs/2610.18696
作者: Radosław Kuczbański,Wojciech Kozłowski,Mateusz Grzesiuk,Konrad Karanowski,Lukas Frey,Andrzej Rzepiela,Maciej Zięba
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Particle picking in cryo-electron microscopy is commonly formulated as an object-detection problem in which individual particles are represented by bounding boxes, a natural representation for globular particles but a poor match for filaments that can extend across an entire micrograph. Existing methods reconstruct filament trajectories by linking local predictions in a separate post-processing stage, decoupling detection from instance formation and creating potential ambiguities at gaps and crossings. Directly preserving filament identity is particularly useful for polymorphic specimens, where segments originating from the same filament should remain associated. However, learning such instance-level representations is challenging because expert centerline annotations are scarce and costly to obtain. We introduce CryoTracer, a synthetic-to-real framework that addresses this limitation through procedural micrograph generation, transformer-based polyline prediction, and specimen-specific fine-tuning. Procedural pretraining exposes the model to diverse filament geometries, intersections, and imaging conditions before adaptation using a limited number of fully annotated experimental micrographs. CryoTracer represents filaments as ordered centerline polylines and optimizes their geometry using an orientation-invariant discrete Fréchet objective, enabling consistent learning regardless of tracing direction. Evaluation on established and newly expert-annotated datasets shows that CryoTracer improves instance-level filament recovery over existing particle-detection approaches. These results establish end-to-end polyline prediction as a practical alternative to detection-and-linking pipelines for recovering complete filament instances with limited micrograph annotation budgets. Code and data are available at this https URL
[CV-26] GeoReader: Learning Latent Geometry for Spatial Reasoning in Vision-Language Models
链接: https://arxiv.org/abs/2610.18694
作者: Rong Li,Shijie Li,Jiangyong Huang,Lingdong Kong,Jiaming Zhou,Dongyue Lu,Zeying Gong,Baoxiong Jia,Junwei Liang,Siyuan Huang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multimedia (cs.MM)
备注: 35 pages, 11 figures, 15 tables; Project Page at this https URL GitHub Repo at this https URL
Abstract:Vision-language models (VLMs) struggle with metric and multi-view relationships that remain implicit in RGB observations. Geometry foundation models recover this structure from RGB, but existing methods either fuse the estimated features with visual tokens and learn to use them from question answering alone, or align VLM latents with geometry features. Capturing geometry, however, does not ensure its use: zeroing the aligned latents of a released feature-alignment model leaves its accuracy unchanged. We introduce GeoReader, a framework that learns geometric representations for the VLM that reads them. First, we train lightweight Geometric Readers to produce continuous input tokens, supervising them with geometry descriptions decoded by a frozen VLM, so that the Readers express geometry in a form the target model can already read. We then freeze the Readers to preserve this interface while fine-tuning the VLM on spatial QA, with geometry injection conditioned on the question. Geometry descriptions serve only as training targets: at inference, the VLM answers directly from the continuous geometric tokens without generating a geometry chain of thought. With geometry predicted from RGB images, our approach reaches 94.3 on MindCube-Tiny with Qwen3-VL-4B and, with a multi-source QA mixture, an average of 72.5 across six additional spatial benchmarks with Qwen3.5-9B. These results suggest that a useful geometric representation is one the receiving model can use, not merely one from which geometry can be reconstructed.
[CV-27] Orientation Estimation of Cryo-EM Images for Molecules with Arbitrary Symmetry by Semidefinite Programming
链接: https://arxiv.org/abs/2610.18691
作者: Yunpeng Shi,Amit Singer,Jinxin Wang,Ruiyi Yang
类目: Computer Vision and Pattern Recognition (cs.CV); Numerical Analysis (math.NA); Optimization and Control (math.OC)
备注:
Abstract:This paper extends the non-unique games framework introduced in Bandeira et al. (2020) to address symmetric structures. Precisely, we consider the problem [ \beginaligned \undersetg_1,\ldots,g_n\in \mathcalG\operatornamemin,, \frac1|S|\sum_i,j=1^n \sum_s\in S f_ij(g_i s g_j^-1), \endaligned ] where \mathcalG is a compact group, S\subset \mathcalG is a finite subgroup of symmetry elements, and f_ij:\mathcalG\rightarrow \mathbbR is a loss function. A notable example is common lines based orientation estimation of cryo-EM images for molecules with point group symmetry S , where \mathcalG is the rotation group \operatornameSO(3) . With a semidefinite programming relaxation, we turn the problem into a convex program that is solved with the alternating direction method of multipliers. Unlike existing approaches which are tailored to specific symmetry groups, ours provides a unified computational framework for addressing all point group symmetries. We demonstrate the applicability of the proposed algorithm on the cryo-EM orientation estimation problem on both simulated and experimental datasets. In particular, our algorithm is effective in generating high-quality ab initio models from only a few class averages computed from experimental images. Subjects: Computer Vision and Pattern Recognition (cs.CV); Numerical Analysis (math.NA); Optimization and Control (math.OC) Cite as: arXiv:2610.18691 [cs.CV] (or arXiv:2610.18691v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2610.18691 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-28] Just Respect the Semantics: Image-Token Compression via Pre-LLM Semantic Variable-Resolution Tokenization and In-LLM Relevance Pruning
链接: https://arxiv.org/abs/2610.18690
作者: Tianhao Fu,Austin Wang,Charles Chen,Roby Aldave-Garza,Victor Xie,Xiyue Jiao,Yucheng Chen,Martin Tin
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Image-token compression must reduce computation without hurting the semantics. We introduce Varest, guided by semantic fungibility: tokens within a coherent semantic region can be summarized coarsely, while object boundaries and literal text require finer distinctions. Object and text localizers guide variable-resolution merging, preserving spatial coverage and finer detail at boundaries and text. Learned [REL] registers then prune visual tokens after early decoder processing. Training across budgets enables one checkpoint per backbone. Across three backbones and 13 benchmarks, Varest outperforms 15 baselines in mean performance retention at every evaluated budget, retaining 86.10-92.33% of full-token performance with only 5.56% of the original layer-averaged image-token count.
[CV-29] RSIgen: Reciprocal Self-Improvement for Agent ic Visual Generation
链接: https://arxiv.org/abs/2610.18688
作者: Yidi Liu,Haozhe Wang,Jie Xiao,Kai Zhu,Xueyang Fu,Zheng-Jun Zha
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Agentic visual generation produces experience that can improve the native understanding (agent) and generation roles of a unified multimodal backbone, but task success alone does not determine which role should learn. Generator updates can also make the agent’s action supervision stale. We introduce RSIgen, a task-centric reciprocal self-improvement framework linking learning allocation, supervision refresh, and recursive curriculum construction. Dynamic Responsibility Routing (DRR) uses matched interventions on evidence, inputs, and supported actions to select learning recipients and construct compatible training targets. After generator learning, RSIgen re-evaluates the relative utility of tested actions with branch inputs held fixed to refresh agent supervision; the retained roles then construct the next curriculum. Across BAGEL, Qwen-Image, and Wan-Image-7B, RSIgen improves AgentGen-Bench Full Overall9 by 3.2–3.3 points over a three-cycle SearchGen adaptation with action costs. Wan controls support allocation gains at matched role-level training exposure, refresh gains at a matched adaptation budget, and benefits from updated curriculum sources while recipients continue learning. In blind evaluation on 200 real-user requests, pooled tie-adjusted preference scores against the same baseline reach 70.0% for task completion and 64.9% for visual quality.
[CV-30] RadarGSOcc: Doppler-Aware 3D Semantic Occupancy Prediction Using 4D Radar Tensor with Geometric-Semantic Feature Extraction
链接: https://arxiv.org/abs/2610.18686
作者: Christofel Rio Goenawan,Dong-Hee Paek,Seung-Hyun Kong
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注:
Abstract:3D semantic occupancy prediction is crucial for autonomous driving, but camera-based methods degrade in low-light conditions and LiDAR-based methods are sensitive to adverse weather. 4D radar is more robust, yet its sparsity and the high cost of radar tensor processing limit its use. We propose Radar Geometric-Semantic Occupancy Prediction (RadarGSOcc), an efficient framework that directly processes 4D radar tensors by separately extracting geometric and semantic features, including Doppler information, with a deformable attention-based geometric refinement module as its key component. We also introduce the K-Radar 3DSemOcc benchmark, the first large-scale benchmark for 3D semantic occupancy prediction using 4D radar tensors, built on the K-Radar dataset with refined LiDAR-based annotations. RadarGSOcc predicts 15 classes within a 51.2m range and achieves 38.2% IoU, 7.6% mIoU, and 25.5% weighted mIoU, outperforming RadarOcc by 17% relative (25.5 vs. 21.9) in weighted mIoU while using fewer parameters and lower computation. It substantially improves truck prediction while trading off car accuracy, outperforms camera-based approaches, and narrows the gap with LiDAR-based methods. To assess robustness without per-frame ground truth in adverse weather, we further propose a novel proxy metric, Environment-Based KL Divergence (EBKL), under which 4D radar proves a reliable and efficient modality for 3D semantic occupancy prediction. The dataset and code will be publicly available.
[CV-31] Perception Beyond the Visible: Incompleteness-Aware Scene Understanding for Robot Autonomy
链接: https://arxiv.org/abs/2610.18670
作者: Rohit Mohan
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: Research statement for RSS Pioneers 2026 (Robotics: Science and Systems). OpenReview: this https URL
Abstract:Reliable robot autonomy requires reasoning about the world from partial and noisy observations. However, many perception systems model only visible scene regions, rely on fixed semantic categories, and assume reliable sensor measurements. These assumptions break under occlusion, novel object encounters, and sensor degradation. In this paper, I present a research vision for incompleteness-aware scene understanding. I organize this vision around three fundamental challenges: incomplete evidence, incomplete knowledge, and incomplete sensor reliability. I discuss my research contributions toward addressing these challenges through amodal perception, semantic occupancy forecasting, open-world scene understanding, label-efficient learning, and uncertainty-guided multi-modal fusion. Building on these contributions, I outline future research directions toward belief-centric 4D scene representations. These representations aim to jointly model geometry, semantics, motion, and uncertainty. I also discuss active sensing and safe adaptation strategies that allow robots to reduce uncertainty through interaction. The goal is to develop calibrated and actionable scene representations that support reliable autonomy in dynamic, open-world environments.
[CV-32] xture Interpolation as Texton Matching ATC
链接: https://arxiv.org/abs/2610.18655
作者: Ido Madar,Gal Oren,Shai Avidan
类目: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR); Machine Learning (cs.LG)
备注: 28 pages. Code: this https URL
Abstract:Texture interpolation asks what should lie between two textures. Standard methods typically blend pixels or features at corresponding spatial locations, implicitly assuming that co-located elements should be mixed. For unrelated textures, however, this correspondence is arbitrary: a crack at one location has no particular relationship to a fiber at the same location in another image. We study whether this hidden pairing choice is itself a major determinant of interpolation quality. We represent textures as collections of local appearance vectors, or textons, and replace positional pairing with appearance-aware matching while keeping the encoder, representation, and decoder fixed. This isolates the effect of correspondence from changes in model capacity or training. Across texture pairs, appearance-aware matching consistently improves the distributional quality of interpolated midpoints over positional feature blending. Soft matching produces the strongest midpoint statistics, whereas one-to-one assignments yield more balanced transitions and closer agreement with the target endpoint. The same intervention also improves a second pretrained texture interpolator without retraining. These results identify correspondence as a first-class design choice in texture interpolation: meaningful transitions depend not only on how strongly two textures are mixed, but also on which local appearances are mixed with one another.
[CV-33] ST-GRU PredRNN-V2: A GRU-Based Reformulation of PredRNN-V2 for Video Prediction
链接: https://arxiv.org/abs/2610.18645
作者: Augusta van Haren,Hanna Hoogen,Luca Pattavina,Quincy Rommen
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Video prediction requires models to capture complex spatiotemporal dependencies and to retain relevant information over long sequences. PredRNN and its successor PredRNN-V2 address this challenge with spatiotemporal Long Short-Term Memory (ST-LSTM) units that maintain two separate memory states, at the cost of a relatively complex recurrent building block. In this work, we investigate whether the simpler gating of Gated Recurrent Units (GRUs) can be transferred to spatiotemporal predictive learning. To this end, we develop a spatiotemporal GRU (ST-GRU) unit as a GRU-based alternative to the ST-LSTM. The ST-GRU replaces the ST-LSTM’s three gates per memory state with two GRU-style gates (update and reset), and merges the temporal memory state into the hidden state, which reduces the number of parameters by about 5%. We incorporate the ST-GRU into PredRNN-V2 and evaluate the resulting model on the Moving MNIST video prediction benchmark, following the original experimental protocol. Averaged over ten predicted frames, the ST-GRU-based PredRNN-V2 achieves an MSE of 55.1, an SSIM of 0.883, an LPIPS of 0.065 and a PSNR of 19.53dB. It does not surpass the original ST-LSTM-based PredRNN-V2 on MSE, SSIM or PSNR, but it achieves a lower LPIPS and performs on par with the earlier ST-LSTM-based PredRNN, while using a simpler recurrent cell. These results suggest that ST-GRUs are a promising building block for spatiotemporal predictive learning and motivate further work on their design and integration.
[CV-34] Grid or Nodes? Probing Object Memory in World Models Under Partial Observability
链接: https://arxiv.org/abs/2610.18637
作者: Justin McMillen,Yasin Yilmaz
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:A robot’s world model must remember objects that leave the camera’s view. FloWM, a learned world model, keeps them in a spatial grid of features that it moves with five built-in velocities. A node memory instead keeps each object’s position and velocity as numbers. We compare them in FloWM’s own block world, with an oracle-fed Kalman tracker as the node memory. How the grid is read changes its score: a single-cell probe finds 66% of out-of-view blocks, while FloWM’s own probe 89%. FloWM also scores 8 points higher in view although no frame is shown, which we attribute to its write-back of its own predictions. Out of view, the tracker reaches FloWM’s range in 14KB against 5.6MB, with no probe. On diagonal motion, outside its five velocities, FloWM finds 11% of blocks even after fine-tuning, while the unchanged tracker keeps 93%. Faster blocks defeat both until one more allowed speed takes the tracker to 94%, where FloWM’s built-in shifts would need 25 copies of its grid. A learned node memory must still replace the oracles and handle velocity under noise, long gaps, and the bounce.
[CV-35] MajorTOM:Elliot Pretrain A Multi-Sensor Multi-Temporal Pretraining Dataset for Earth Observation
链接: https://arxiv.org/abs/2610.18624
作者: Carmen Oliver Huidobro,Mikolaj Czerkawski,Kevin Monsalvez-Pozo,Cesar Aybar,Gustau Camps-Valls,Oscar J. Pellicer-Valero
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Pretraining generalist Earth observation foundation models requires data that combine several sensors, varied temporal sampling and, increasingly, text, yet open datasets rarely provide all three in one consistent, easily accessible form. We introduce MajorTOM:Elliot Pretrain, a collection of 279,166 tiles of 10 km drawn from the Major-TOM grid by clustering AlphaEarth embeddings. Each tile stacks Sentinel-2 L1C, Landsat-8/9, Sentinel-1 RTC, Copernicus DEM and ESA WorldCover on a shared 1056x1056-pixel, 10 m grid, complemented by cloud masks, OpenStreetMap entities and ERA5 weather. The tiles are partitioned into three temporal regimes, Monotemporal (250,000 tiles, one acquisition), Monthly (12,500 tiles, twelve acquisitions spread over the annual cycle) and Burst (16,666 tiles, six acquisitions within 30 days), and are released as 6.3 TB of Cloud Optimized GeoTIFFs indexed by TACO v3 Parquet metadata, a standard that allows cloud streaming, efficient data access, and filtering before any pixel is read. A companion library derives sensor-tagged fact sheets, captions and 16 families of vision-language tasks from these layers at load time. We test our design choices in short pretraining runs of OlmoEarth-Base. First, a 25% subset of Major-TOM Core-S2L2A curated with AlphaEarth embeddings matches or exceeds the full corpus on 8 of 9 benchmarks. Second, on Elliot-Pretrain, adding Landsat-8/9 improves or matches Sentinel-2 only pretraining on every benchmark. Third, using DEM and WorldCover as prediction targets during pretraining further improves 8 of 9 benchmarks, while giving the same maps to the encoder as inputs degrades 5 of 9. Fourth, Sentinel-1 helps on 7 of 9 benchmarks, particularly on benchmarks with radar inputs. Fifth, rule-based captions match a small language model at no cost. The dataset is openly available at this https URL.
[CV-36] VideoJev: System One Model for Video Understanding and Interaction
链接: https://arxiv.org/abs/2610.18615
作者: Yuhan Zhu,Xiangyu Zeng,Qingyi Si,Chenxu Yang,Yifei Li,Luozheng Qin,Junhao Zhou,Dingyu Yao,Jun Zhang,Chuanyu Qin,Zhiqiu Zhang,Yuandong Yang,Yuqi Xu,Tianyu Zhang,Hongcheng Guo,Jiaqi Wang,Nan Duan,Limin Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 23 pages. Project page: this https URL
Abstract:Visual agents must turn video observations into timely decisions to track progress and choose what to do next. Building on the emerging System One paradigm, we introduce VideoJev, a family of models that cast video understanding and interaction as structured decision tasks. We adapt pretrained multimodal large language models (MLLMs) by replacing autoregressive answer decoding with a shared head that scores temporal intervals and textual options against video evidence. These scores yield event spans or answer choices, unifying temporal grounding and multiple-choice video question answering. This combines the semantic flexibility of MLLMs with efficient decisions that agents can use directly. As an interaction unfolds, new goals and intermediate results prompt further queries about the same visual history. VideoJev therefore reuses visual computation through Cached inference for queries as they arrive and Joint inference for multiple queries in a single forward pass. Across three temporal grounding benchmarks, VideoJev-4B and VideoJev-35B-A3B improve mean mIoU over their backbones by 6.7 and 10.8 points, respectively, with measured inference speedups of 2.8 - 3.1\times . On four video QA benchmarks, both achieve performance comparable to their backbones in Thinking mode, with measured inference speedups exceeding 100\times . These results offer a unified, reusable decision interface for composing structured visual judgments into complex, open-ended workflows.
[CV-37] DPOison-VL: Clean Preference Poisoning in Vision-Language Direct Preference Optimization
链接: https://arxiv.org/abs/2610.18609
作者: Chengtao Jian,Kai Yang
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Direct Preference Optimization (DPO) has become a widely adopted approach for aligning vision-language models (VLMs), yet the security of multimodal preference data remains largely unexplored. Existing preference poisoning attacks either corrupt explicit supervision signals or require visible triggers, making them detectable under human inspection. We introduce DPOison-VL, the first clean preference poisoning attack against VLM-DPO, which manipulates model alignment solely through imperceptible perturbations on a small fraction of training images, while leaving all prompts, responses, and preference annotations semantically intact. We formulate poison construction as a bilevel optimization and derive a gradient alignment surrogate that efficiently steers image perturbations toward the attacker-specified preference. We further introduce a preference displacement score for principled poison selection with provable first-order optimality under a fixed poison budget, and derive a sufficient condition for preference inversion. Extensive experiments across diverse attack tasks, different VLM architectures and scales, and baseline comparisons demonstrate that DPOison-VL achieves up to 98% ASR with only 2% poisoned data and exhibits cross-model transferability, while largely preserving clean accuracy and general utility. Our findings reveal a critical vulnerability in VLM-DPO, where imperceptible image perturbations in preference training data can silently redirect alignment.
[CV-38] Recasting DiT for Extreme Image Compression with Hierarchical Visual Priors
链接: https://arxiv.org/abs/2610.18607
作者: Yunuo Chen,Jingnan Gao,Bing He,Jinjing Zhao,Chuqin Zhou,Xiaoyue Ling,Li Song,Guo Lu,Wenjun Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Pretrained diffusion models provide powerful generative priors for extreme image compression, but their stochastic, iterative generation processes are not designed to decode compressed representations directly. We present PrismCodec, which recasts a pretrained Diffusion Transformer (DiT) as a bitstream-grounded one-step generative decoder. PrismCodec uses the decoded representation in two complementary roles: a projected codec state defines the starting state for generation, while hierarchical visual priors derived from the same decoded latent provide semantic and structural guidance. To bridge the gap between stochastic, iterative generation and deterministic one-step decoding, we develop a progressive adaptation strategy. We first learn codec-conditioned transport from a noisy neighborhood centered on the codec state, distill it into a one-step mapping, and finally anneal the source perturbation to zero during end-to-end optimization. The resulting decoder reconstructs deterministically from the received representation without additional conditioning bits. Across Kodak, CLIC, and DIV2K, PrismCodec consistently improves rate-perception performance over prior generative codecs, achieving BD-rate reductions of 70-75% in LPIPS and 75-81% in DISTS over StableCodec. Code will be released upon acceptance.
[CV-39] LLM 2R: From Large Language Models to 3D Reassemblers
链接: https://arxiv.org/abs/2610.18604
作者: Zhouhan Zhong,Qiuhong Shen,Xingyi Yang,Jinyi Xu,Xinchao Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: 20 pages, 5 figures, 7 tables
Abstract:3D part assembly aims to reconstruct a complete shape from a set of unassembled parts by predicting a rigid SE(3) transformation for each part. Because the placement of a part is rarely determined by its own geometry, the task demands reasoning over the entire part set, and existing methods, which rely on pairwise relations or iterative generation and verification, scale poorly to objects with many parts. We present LLM2R, which casts part assembly as sequence modeling with a pretrained decoder-only large language model (LLM): a Point-BERT encoder and a Q-Former compress each part into a block of geometric tokens, the blocks of all parts are concatenated, and a causal LLM adapted with Low-Rank Adaptation (LoRA) processes the sequence in one forward pass. Causal attention, however, lets the hidden states of a part observe only the parts that precede it. LLM2R therefore appends a block of learnable summary tokens after the last part: these tokens follow every part, so their states aggregate the whole assembly at a constant token cost, and a part-aware pose head fuses each part’s states with them before predicting its rotation and translation. One architecture thus handles semantic and geometric assembly and predicts all poses in a single non-autoregressive pass. LLM2R achieves the highest part accuracy and success rate on all three PartNet categories, the best rotation error, translation error, and part accuracy on PartNeXt, and the lowest translation error and fastest inference on Breaking Bad, where only a fracture-specialized method is more accurate. Its advantage grows with the number of parts: at 100 parts it exceeds GARF by 33.5 points in part accuracy on Breaking Bad and Assembler by 29.1 points on PartNet, and it transfers zero-shot to unseen categories with more than 16 points higher part accuracy than the strongest of three baselines.
[CV-40] Detecting Unreliable Classifications from Quantized Models
链接: https://arxiv.org/abs/2610.18575
作者: Ananya Kotla,Jorge T. Gomez
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 1 figure, 5 tables
Abstract:Post-training INT8 quantization is typically validated by checking accuracy retention, which requires ground-truth labels often unavailable in low-resource clinical settings. We show a label-free alternative: comparing a model’s own output before and after quantization. We compute cosine similarity between the FP32 and INT8 models’ output probability vectors for each prediction, which requires only a single forward pass. This output-agreement similarity separates lower-accuracy predictions across four evaluated architectures (ResNet-50, VGG-16-BN, Wide ResNet-50, and ViT-B/16) under fixed-percentile splits, outperforming a confidence-only baseline for all four. These results show that quantized models can be compared to their baselines to assess their drift and determine whether they are well suited for clinical deployment.
[CV-41] Parameter-Efficient Multimodal Brain Tumor Segmentation With Stage-Wise Shared Adapters
链接: https://arxiv.org/abs/2610.18571
作者: Milad Rostamian,Mohammadreza Mohammadi,Ender Konukoglu,Mohsen Soryani
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Different MRI modalities reveal different parts of a brain tumor. Segmentation models combine information from these modalities to outline tumor regions. Many models use a separate encoder for each modality. However, exchanging features among these encoders at several stages can increase model complexity. We introduce the Block-Wise Shared Adapter (BWSA), which combines compressed features from the modality encoders in a shared path and returns adapted features to each stream. In its fixed-blending form, BWSA adds the adapted features to the original features. Its trainable-gated variant (TG-BWSA) instead learns separate weights for the original and adapted features at each modality and encoder stage. We evaluate both variants on BraTS 2020 and BraTS 2023 with four segmentation backbones. On the official BraTS 2023 online validation set, BWSA increases average region-wise Dice by 1.3–7.6 percentage points relative to the corresponding unmodified backbones. TG-BWSA increases it by 2.8–7.7 points. Both variants add approximately 0.85–3.73% parameters and reduce average HD95 relative to the unmodified backbones.
[CV-42] Observation Quality in DSM-Assisted Aerial Localization: A Benchmark and Base-Anchored Uncertainty Calibration
链接: https://arxiv.org/abs/2610.18564
作者: Pengyu Yin,Junxiang Tan,Jie Pan,Mengqi Fu,Kun Cheng,Yifan Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 15 pages, 7 figures, 7 tables; supplementary material integrated into the main text
Abstract:Aerial map matching provides position observations for navigation, but a position error alone does not determine how an observation should be weighted. We define five benchmark tasks: position availability and agreement, grouped error structure, predictive uncertainty, controlled updates, and temporal accumulation. A case study in digital-surface-model-assisted localization evaluates base-anchored uncertainty calibration within this framework. The method combines propagated image and map sensitivities with the second moment of a compact configuration subset relative to the reported position. The reported position remains unchanged, and scale is fitted using separate position references without local surface-height ground truth. Thirteen covariance models are evaluated on two held-out query sets from locations represented in the calibration data. Relative to a constant full-matrix baseline, the method reduces equal-location normalized position-update risk by 7.70% on 728 conditions and 5.46% on a separate 142 conditions under an independent prior, with all localization failures included. On the primary set, equal-location negative log likelihood improves by 2.571 nat. These results show improved reference-based weighting at seen locations. Exploratory base-quality and conditional-addition controls leave the incremental update benefit of configuration disagreement unresolved. Navigation benefit and real temporal accumulation remain untested.
[CV-43] Denoise X: Self-Supervised U-Net for Quantum Mottle Suppression in Chest Radiography
链接: https://arxiv.org/abs/2610.18556
作者: Parth Giramkar,Aryan Patle,Vanita Kshirsagar
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Deep learning for medical image restoration is often constrained by the need for paired training data. Acquiring perfectly registered clean-noisy radiographs in vivo is impractical due to patient motion and radiation-dose limits. To address this, we employ a self-supervised Noise2Noise paradigm that trains on independently corrupted observations without paired clean images. We propose Denoise X, a lightweight self-supervised U-Net model (7.79M parameters) for suppressing quantum mottle and electronic readout noise in low-dose chest radiographs. It is trained on the NIH ChestX-ray14 dataset (112,120 images) using a stochastic Poisson-Gaussian noise simulator. Zero-shot evaluation on the unseen Montgomery County dataset achieved a PSNR of 36.09 dB and SSIM of 0.8806, outperforming the Bilateral filter while preserving high-frequency anatomical details. Using patch-wise inference, Denoise X processes native-resolution chest radiographs in ~2.2 s per image on a Tesla P100 GPU. Cross-domain evaluation on the Shenzhen dataset achieved 36.44 dB PSNR and 0.9099 SSIM, showing strong generalization. For resource-constrained deployment, the model was converted into a 31.16 MB ONNX model for efficient CPU inference on non-DICOM images.
[CV-44] Unbalanced W-Flow: Unbalanced Wasserstein Drifting under Finite-Batch Marginal Fluctuations
链接: https://arxiv.org/abs/2610.18552
作者: Jongwook Jeon,Mingue Park,Jiwon Choi,Jaihoon Kim,Yunhong Min,Minhyuk Sung
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: 31 pages, 9 figures, 9 tables
Abstract:Teacher-free one-step generative models such as W-Flow learn directly from distributional objectives, with the Sinkhorn divergence providing the strongest empirical performance among the energies studied in W-Flow. However, Sinkhorn divergence is based on balanced optimal transport, whose exact marginal constraints can induce undesirable transport under finite minibatch fluctuations. We introduce UW-Flow, which replaces the balanced Sinkhorn energy with a debiased unbalanced Sinkhorn divergence and derives the corresponding Wasserstein velocity. The resulting dynamics retain W-Flow’s cross-minus-self structure, recover the balanced limit, and preserve target identifiability under suitable assumptions. We further characterize the finite-batch velocity error of balanced and unbalanced transport by comparing each minibatch field with its population counterpart. Our analysis identifies conditions under which marginal relaxation yields smaller minibatch-to-population velocity error, and shows that for separated matched modes, balanced sensitivity grows with inter-mode separation while the corresponding Unbalanced Optimal Transport (UOT) sensitivity remains controlled by local transport geometry. Controlled experiments on fixed-support and continuous two-mode distributions support these predictions. On MNIST, CIFAR-10, and ImageNet-100, UW-Flow improves generation under matched training budgets, with the largest gains in unconditional generation and consistent improvements in the class-conditional setting.
[CV-45] Beyond the Encoder: Object-Centric Representations Emerging from Segmentation Foundation Models
链接: https://arxiv.org/abs/2610.18543
作者: Michalis Pistos,Bartłomiej W. Papież
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 20 pages, 7 figures, 3 tables
Abstract:Foundation models are increasingly used as feature extractors for downstream tasks with limited data, typically through representations obtained from their encoders. However, foundation models trained for structured tasks such as segmentation also form object-specific representations during task execution, whose transferability remains largely unexplored. We investigate Segment Anything Model 2 (SAM2) object pointers, compact object-level representations produced through the segmentation pathway, as transferable representations for downstream tumour prediction. Through experiments on two PET/CT tumour prediction tasks, we find that object pointers learned using segmentation supervision alone already encode tumour-discriminative information, despite never being optimised for the downstream labels. We further introduce task-guided object pointers, where an auxiliary objective jointly shapes the object pointer space for the downstream prediction together with segmentation. Task-guided pointers improve cancer-type classification, outperforming conventional encoder-based transfer and radiomics, while segmentation-only pointers perform best on the more challenging recurrence prediction task. Representation analyses reveal associations between object pointers and interpretable radiomics features, class-dependent organisation of the learned representation space, and explicit spatial grounding around the segmented tumour. Together, these findings suggest that useful transferable representations can emerge through the execution of a foundation model’s primary task, providing an alternative to conventional encoder-based transfer.
[CV-46] Meta-Video: Efficient Standardized Evaluation for Video Understanding
链接: https://arxiv.org/abs/2610.18535
作者: Soumya Shamarao Jahagirdar,Anna Kukleva,Vishaal Udandarao,Samuel Thomas,Brian Kingsbury,Rogerio Feris,James R. Glass. Ameya Prabhu,Hilde Kuehne
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Video understanding benchmarks are the primary yardstick for measuring progress in video-language modeling, and a growing body of work relies on them to claim improvements. The reported benchmark scores in these works, however, are almost always produced by frameworks (e.g., lmms eval) used as black boxes: certain evaluation choices like frame sampling, preprocessing, and input formatting are inherited from preset defaults and seldom inspected carefully. We show these unexamined choices are a major source of variance. Even for the same model checkpoint, evaluation configuration alone can shift accuracy by up to 19.5pp. Hence, it is unclear whether reported gains reflect genuine improvements in capability or differences in evaluation setup. Moreover, many benchmark samples offer little evidence of visual capability, since they can be answered without the video, and the full suites are expensive to evaluate. We address these issues in two parts. First, we curate existing suites for discriminability and efficiency, producing Meta-Video, a balanced 1,000-question benchmark over 791 videos spanning diverse tasks and durations, which largely preserves model rankings, at 5 times inference speedup for a full-frame count sweep experiment (4-128 frames). Second, we establish a standardized evaluation protocol by sweeping the configuration space to isolate choices that measurably affect results, distill them into best practices and reporting standards, and release reproducible reference scores across five models and seven benchmarks. We hope that the proposed evaluation framework and the Meta-Video benchmark will support video understanding research by providing a reproducible and efficient foundation to measure progress in a reliable way.
[CV-47] Dual-Domain Sparse-View 4D CBCT Reconstruction with Spatial-Temporal Consistency Learning
链接: https://arxiv.org/abs/2610.18518
作者: Qi Lai,YuTong He
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Four-dimensional cone-beam computed tomography (4D CBCT) provides phase-resolved volumetric images for image-guided radiotherapy. Yet good reconstruction quality usually needs many projections. This increases scan time and radiation dose. We propose STCNet, a sparse-view 4D CBCT reconstruction framework that uses spatial-temporal consistency learning. STCNet reconstructs high-quality images from only 16 projections in each respiratory phase. STCNet integrates physics-driven positional encoding to enforce geometric consistency in the projection domain and introduces temporal encoding in latent space to capture respiratory motion dynamics across phases to capture dynamic respiratory motion across phases. A dual-domain collaborative strategy jointly optimizes both projection and image domains, enhancing structural fidelity and temporal coherence. We tested STCNet on the TCIA-LUNG dataset and a CIRS thorax phantom dataset. STCNet performed better than conventional methods and recent deep learning baselines in image quality, artifact reduction, and structural preservation. On TCIA-LUNG, STCNet reached PSNR values of 33.311, 32.695, and 32.658 dB at phases 2, 5, and 8. The corresponding SSIM values were 0.845, 0.823, and 0.825. On CIRS, STCNet achieved PSNR values up to 36.497 dB and SSIM values up to 0.877. These results support STCNet as a promising framework for low-dose, phase-resolved 4D CBCT reconstruction in image-guided radiotherapy.
[CV-48] EgoOmniWorld: Unifying Egocentric Understanding via Ego-Action-World Interactions
链接: https://arxiv.org/abs/2610.18499
作者: Songruo Chen,Yuqian Fu,Yu Li,Mingkang Dong,Sifan Zhou,Bin Ren,Xu Zheng,Tianwen Qian,Mohamed Elhoseiny,Yanwei Fu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Egocentric video understanding spans diverse capabilities, modalities, and evaluation formats, yet these are often studied in isolation despite their tight coupling in real-world interaction. We propose the Ego-Action-World interaction loop as a unifying principle: the wearer perceives the world, forms goals, acts, changes the world, and uses feedback to guide subsequent behavior. Based on this view, we introduce EgoOmniWorld, a benchmark organized around four complementary interaction structures: Node captures ego, action, and world states; Edge captures directed relations among them; Chain captures linked dependencies; and Loop captures feedback-closed interactions. We develop an interaction-first, evidence-grounded data pipeline that extracts structured interactions from egocentric videos, grounds them in visual and auditory evidence, and generates questions according to their interaction structure. Required modalities and single- or multi-turn formats follow naturally from the supporting evidence and dependencies. EgoOmniWorld contains 24,166 QA groups, including 20,284 training and 3,882 test samples. Experiments on representative open- and closed-source multimodal large language models reveal substantial and uneven limitations across interaction structures. Models trained on EgoOmniWorld improve in-domain interaction understanding and generalize to existing egocentric benchmarks. EgoOmniWorld provides a unified foundation for evaluating and advancing egocentric video understanding. Data, code, and models will be released.
[CV-49] Physics-Aware Deepfake Detection via Distance-Speech Consistency INTERSPEECH2026
链接: https://arxiv.org/abs/2610.18485
作者: Kyeongrae Kim,Kim Sung-Bin,Oh Hyun-Bin,Tae-Hyun Oh
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Interspeech 2026, Project Page: this https URL
Abstract:Recent progress in generative models has made talking-head deepfakes highly realistic and easy to create, jeopardizing public trust. However, most audio-visual deepfake detectors rely on lip-speech synchronization and are designed for static, frontal videos, limiting their reliability in dynamic, in-the-wild recordings where the speaker moves and lip cues are degraded or missing. As an alternative, we propose a physics-aware detector for dynamic speaking videos that leverages an acoustic constraint: in real recordings, speech energy varies predictably with speaker-to-camera distance, whereas deepfake generation can break this coupling. Our method combines distance estimates from video with distance-dependent acoustic measures from speech to detect deepfakes. Experiments on in-the-wild benchmarks show that the proposed physical cue is effective in dynamic scenarios, and that it complements lip-sync-based detectors, with a simple ensemble achieving consistent gains across datasets. Codes are available at this https URL.
[CV-50] Simplifying and Improving On-Policy Distillation for Few-Step Diffusion
链接: https://arxiv.org/abs/2610.18472
作者: Jiawei Zhang,Ziyu Wan,Nanye Ma,Long Zhao,Dong Chen,Hongyu Liu,Qifeng Chen,Dongdong Chen
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Image and Video Processing (eess.IV)
备注: Project Page: this https URL
Abstract:Few-step distillation reduces the sampling cost of diffusion and flow models, but existing approaches often require auxiliary score models or specialized architectures. We show that simple on-policy velocity regression already provides a strong foundation for 4-NFE generation. To represent teacher velocity variation within each coarse transition, we introduce Pivot, a training-only residual proxy that separates local corrections from the student’s interval-mean prediction. We further repel the student from weaker teacher predictions, providing a lightweight mechanism for avoiding undesirable modes. The resulting framework, Simplified Matching Distillation (SMD), remains entirely regression-based, requires neither online score estimation nor adversarial training, and leaves the deployed generator unchanged. We evaluate SMD on image generation and editing, 3D generation, video synthesis, and joint audio–video generation, scaling to 33 billion parameters. At 4-NFE, SMD outperforms strong few-step baselines and surpasses multi-step teachers on several benchmarks. It also supports cross-model, multi-teacher, and causal-video distillation.
[CV-51] MFVSR: A Multi-Style and Fine-Grained Benchmark for Video Subtitle Removal
链接: https://arxiv.org/abs/2610.18461
作者: Xiyan Huang,Qingdong He,Peng Tang,Bo Yin,Chaoyi Wang,Xiaobin Hu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 31 pages, including appendix. Code: this https URL ; Data: this https URL
Abstract:Video subtitle removal (VSR) is an important video restoration task, yet existing evaluation remains limited by low-resolution datasets, fragmented test sets, and generic video-inpainting protocols that fail to adequately capture localized restoration errors, unintended modifications, temporal instability, and incomplete outputs. In this paper, we introduce MFVSR, a large-scale multi-style benchmark for fine-grained and reliability-aware video subtitle removal evaluation. MFVSR distinguishes itself through three key features: (1) Large-scale and diverse data, comprising 3,000 paired videos from Real-HQ, Real-LQ, and AI-Generated sources, with an auxiliary set of 953 in-the-wild real videos for real-world coverage analysis; (2) Controlled subtitle diversity, spanning 16 languages and four factors covering spatial layout, temporal behavior, visual style, and language mode; (3) Fine-grained evaluation, introducing an information-conditioned, spatially decoupled protocol that separates blind from mask-guided removal and measures Restoration Fidelity, Preservation Fidelity, Temporal Fidelity, coverage, and failures. Experiments on six representative methods reveal substantial discrepancies between local restoration, outside-region preservation, temporal consistency, and full-set reliability that are obscured by conventional full-frame metrics. MFVSR therefore provides a more comprehensive and diagnostic foundation for evaluating and advancing video subtitle removal methods.
[CV-52] mathrmS3-PnP: Plug-and-Play Denoising for Scientific Imaging with Self-Supervised Score Priors
链接: https://arxiv.org/abs/2610.18458
作者: Yukun Cui,Jiangshe Zhang,Shuang Xu,Lilun Deng,Haowen Bai,Tao Feng,Haotong Qin,Konrad Schindler,Zixiang Zhao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Modern image restoration methods increasingly rely on large-scale supervised training, yet clean targets are often prohibitively difficult to obtain in scientific imaging modalities such as hyperspectral imaging (HSI) and magnetic resonance imaging (MRI). Plug-and-play ADMM (PnP-ADMM) with pretrained diffusion priors offers a promising alternative, but the learned prior distribution can mismatch both the target scientific image distribution and the distribution of ADMM denoiser inputs induced by dual feedback. To mitigate these mismatches, we propose \mathrmS^3 -PnP, a self-supervised plug-and-play restoration framework that learns a score prior directly from observation-derived ADMM intermediate states, without clean training targets or externally pretrained weights. Specifically, the learned score corrects the input to the analytical prior update, combining observation-driven corrections with analytical regularization. We further analyze the convergence and reconstruction error of the corrected ADMM algorithm assuming an ideal score and quadratic data fidelity. Experiments on reference-guided HSI restoration, single-image HSI denoising, and MRI denoising demonstrate consistent improvements over competing methods.
[CV-53] EmoGenie: Benchmarking and Evaluating Expressed and Evoked Emotion in T2I Generation with MLLM s
链接: https://arxiv.org/abs/2610.18455
作者: Zixuan Xing,Lancheng Gao,Shengyan Li,Huiyu Duan,Zhaolin Cai,Guangtao Zhai,Xiongkuo Min
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Comments: Corresponding author: Xiongkuo Min. Equal contribution: Zixuan Xing and Lancheng Gao
Abstract:Rapid advances in artificial intelligence-generated content (AIGC) have significantly promoted using text-to-image (T2I) models to generate images with specific affective intentions. However, the emotional effects conveyed by generated images frequently deviate from target expectations, and it remains unclear whether multimodal large language models (MLLMs) can reliably interpret the emotion of AI-generated visual content. Existing affective benchmarks are predominantly built on natural images, while affective studies of generated images have largely focused on emotions expressed in portraits, lacking a unified benchmark for studying the emotions expressed and evoked by generated images. We therefore introduce EmoGenie, a benchmark of emotion for AI-generated images that evaluates capabilities in T2I generation and MLLM understanding across expressed and evoked emotions. It comprises 6,400 images generated by eight T2I models using 800 emotion-targeted prompts, annotated with question-answer (QA) pairs, termed EmoGenie-QA, and continuous human ratings, termed EmoGenie-Assess. EmoGenie-QA contains 90,000 closed-set questions in single-image and image-pair settings, as well as 1,057 open-ended questions for emotion description. EmoGenie-Assess benchmarks continuous affective and aesthetic prediction using 13,175 mean opinion scores (MOSs) aggregated from human ratings in terms of valence, arousal, dominance, and aesthetic quality. We further develop EmoProbe, which combines emotion question answering with response-aligned, dimension-specific regression heads for continuous score prediction. Experimental results show that EmoProbe achieves superior results on emotion-related QA tasks, while its affective and aesthetic predictions correlate better with human ratings, demonstrating its state-of-the-art capabilities of unified coarse-to-fine affective understanding for AI-generated images.
[CV-54] hinker-WM: A Unified World Model with Shared Multimodal Representations for Embodied Intelligence
链接: https://arxiv.org/abs/2610.18447
作者: Thinker Team Ubtech: (Alphabetical order),Peng Gao,Jichao Jiao,Chen Liao,Daqin Luo,Yiming Ma,Baiyu Pan,Hailin Shi,Jiyuan Wang,Xiangming Xu,Xiongwei Zhao
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: \href{ this https URL }{Project Page}
Abstract:World models have emerged as a promising paradigm for embodied AI, yet existing systems often remain fragmented in the modalities they model and the functions they support. In particular, visual dynamics, geometric structure, and executable action are commonly coupled through separate interfaces or downstream modules rather than being jointly represented and evolved within a shared dynamics model. In this work, we present Thinker-WM, a unified multimodal world model that jointly models and generates RGB, metric depth, and robot action through a shared intermediate Unified Dynamics Transformer with modality-specific encoders and decoders. The model performs multimodal unification at the representation level: appearance, geometry, and action are transformed into a common dynamic representation and jointly evolved by shared parameters, while their distinct output structures are preserved. The same parameterization supports multiple functions, including joint video-action generation, world prediction, policy inference, action-conditioned generation, video-to-action generation, and inverse dynamics. To examine whether the shared representation captures action-related information, we introduce a lightweight motion probe that shows the frozen visual representation contains recoverable motion information. Thinker-WM is trained on a heterogeneous corpus comprising approximately 20K hours of egocentric video, 10K hours of real-robot data, and 1K hours of simulation data, including approximately 4K hours of real-robot data with aligned depth supervision. Extensive experiments demonstrate strong performance across embodied world-model evaluations. Thinker-WM ranks first on EWMBench, WorldOlympiad and WorldArena, while controlled studies further demonstrate the benefits of action conditioning, geometric interaction, and multi-function training within the shared representation.
[CV-55] CoTL: Compact Token Learning for 3D Scene Representation
链接: https://arxiv.org/abs/2610.18446
作者: Han Liang,Sibo Wu,Anpei Chen,Rüdiger Westermann
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We present CoTL, a framework that learns a compact set of tokens that preserves scene geometry and appearance from unposed multi-view images. Our encoder compresses the input images into small per-view sets of latent tokens that aggregate information across views to form a view-consistent scene representation. Each additional view expands the representation with only a small, resolution-independent set of tokens, allowing its capacity to grow with the available observations. To force this compact bottleneck to retain sufficient scene information, a query-based Gaussian decoder uses scene-independent queries to attend to the learned tokens and recover an explicit, renderable set of 3D Gaussian primitives. Novel-view rendering supervision trains the compact representation end to end through the reconstructed Gaussian scene. Experiments on \retenk and ACID show that the resulting compact, unstructured token set supports competitive novel-view synthesis and can serve as the latent space of a generative model for 3D Gaussian scene generation. Together, these results show that 3D Gaussian reconstruction is an effective objective for compact scene representation learning and the learned token space provides a basis for broader downstream tasks in 3D vision.
[CV-56] Resource-Efficient 3D Shape and Orientation Estimation of Road Users using Roadside RGB Camera and 3D Radar
链接: https://arxiv.org/abs/2610.18417
作者: Shiva Agrawal,Savankumar Bhanderi,Gordon Elger
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted at the 18th IEEE Symposium Sensor Data Fusion: Trends, Solutions and Applications (SDF 2026), Bonn, Germany
Abstract:End-to-end deep learning pipelines for camera-radar 3D object detection are well established in driver assistance systems and autonomous vehicles; however, adapting them to roadside-mounted infrastructure presents significant challenges. On vehicles, sensor heights, pitch angles, and relative geometries remain largely uniform, allowing end-to-end models to generalize effectively. On the contrary, roadside sensor installations vary drastically in height, tilt, and field of view across locations, rendering end-to-end models prone to failure and difficult to adapt without massive location-specific datasets. Hence, instead of designing a complex end-to-end pipeline for direct 3D bounding box estimation, a low-cost, resource-efficient two-stage approach is proposed for real-time 3D shape and orientation estimation. The first stage utilizes any off-the-shelf, state-of-the-art method to extract 2D camera bounding boxes and associated 3D radar points for each road user. The proposed second stage then processes these inputs to construct 3D oriented bounding boxes where radar data determines 3D position, predefined templates define size for each road user, and a tiny neural network predicts a site-specific and optimized yaw angle. The algorithm is successfully validated at a curved road with a junction for pedestrians, bicycles, and cars, where it achieves 25 frames per second on an NVIDIA Jetson ORIN for real-time applications.
[CV-57] Bifocal: Training-Free Mixed Granularity Visual Token Pruning for Multimodal LLM s
链接: https://arxiv.org/abs/2610.18399
作者: Peiran Yin,Hao Wu,Zihao Yang,Fengyu Zhou,Yijuan Liang,Haozhe Hu,Xudong Wang,Wei Zhang,Xiaoyu Shen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Visual token pruning for multimodal large language models (MLLMs) appears highly effective, with existing methods retaining over 95% of unpruned performance on common VQA benchmarks using a quarter of the visual tokens or fewer. However, our extensive evaluation of ten methods across three MLLM families and five ability groups reveals a critical gap: strong general performance often comes at the cost of severe degradation on fine-grained tasks such as OCR and visual grounding. We trace these failures to two limitations of the prevailing paradigm of retaining the highest-ranked native patches. A single importance ranking overlooks the complementary evidence required by different tasks, while sparse patch selection disrupts both image content and spatial integrity. To address these limitations, we propose Bifocal, a training-free, two-stage pruning method built on a mixed-granularity framework. It allocates separate budgets to query-focused and image-intrinsic evidence, complemented by a pooled coarse representation of the entire image. Before the LLM, Bifocal matches question words to visual regions through a soft conjunction balanced against norm-debiased image saliency, while selecting image-intrinsic evidence to maximize representational coverage. Inside the LLM, it further prunes tokens using density-adjusted text-to-image attention. With roughly a quarter of the original visual tokens, Bifocal is the only evaluated method to retain over 90% of unpruned performance in every ability group across all three models, and ranks first on average in all nine model-budget settings. The code is available at: this https URL
[CV-58] ReDiS: Reasoning -Diagnosis Self-Evolution for Multimodal Reasoning in Remote Sensing
链接: https://arxiv.org/abs/2610.18388
作者: Yi Liu,Wentao Jiang,Shuo Ni,Zhiming Luo,Di Wang,Zengmao Wang,Jing Zhang,Bo Du
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal large language models (MLLMs) have shown strong potential for remote sensing understanding, yet reliable reasoning remains challenging when tasks require fine-grained visual grounding and multi-step inference. Existing approaches typically rely on either costly pre-constructed reasoning trajectories or self-generated signals such as response consistency, which can reinforce shared errors and provide unreliable supervision. We introduce \textbfReasoning–Diagnosis Self-Evolution (ReDiS), a framework that separates reasoning exploration from candidate evaluation while allowing the two processes to evolve jointly. ReDiS instantiates two role-specific experts from a shared multimodal backbone: a \textbfReasoner that explores diverse visual evidence–answer candidates and a \textbfDiagnoser that jointly evaluates the complete candidate set and provides credibility-aware feedback. As the Reasoner evolves, it exposes new candidate distributions and failure modes; the Diagnoser adapts to these changes and, in turn, reshapes the learning signal for subsequent exploration. To handle groups in which all sampled answers are incorrect and group-relative advantages vanish, we further introduce \textbfFailure-Triggered Recovery (FTR), which provides minimal outcome-level correction without prescribing intermediate reasoning trajectories. Experiments on LHRS-Bench, MME-RS, and CHOICE show consistent improvements over Qwen3.5 backbones at 0.8B, 2B, and 4B scales. ReDiS improves category-averaged accuracy by 6.47 percentage points at 0.8B, while ReDiS-4B achieves the highest average performance among the compared remote sensing MLLMs.
[CV-59] Noise2Match: Self-Supervised Low-Dose CT Denoising via Adaptive Inter-Slice Matching
链接: https://arxiv.org/abs/2610.18379
作者: Langrui Zhou,Guang Li,Xinyu Huang,Ziteng Zhou,Zuyang Huang,Hongming Shan,Shouhua Luo,Ning Gu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:With the rapid development of deep learning, supervised convolutional neural network methods have greatly improved the performance of low-dose computed tomography (CT) denoising. However, they require paired noisy-clean training data, which is difficult to obtain in practice. Some classic self-supervised denoising methods rely on the assumption of pixel-wise independent noise, making them ineffective for correlated noise in low-dose CT (LDCT) images. Considering the high similarity and weaker noise correlation between adjacent CT slices, several recently proposed self-supervised methods use adjacent LDCT slices for training. Nevertheless, these approaches fail to effectively account for the non-rigid misalignment between slices caused by anatomical variation, hindering them from satisfying the equal-mean condition required by the classic Noise2Noise framework. In this paper, we propose Noise2Match, a self-supervised LDCT denoising method based on adaptive inter-slice matching. Noise2Match introduces an inter-slice matching network to model the label transition process between adjacent slices, and then applies correction to the denoiser’s outputs to better satisfy the equal-mean condition. We theoretically prove that, under idealized conditions, training Noise2Match using adjacent LDCT slices is equivalent to supervised training with paired noisy-clean images. We further find that Noise2Match can also effectively perform low-dose conebeam CT (CBCT) denoising, which could be attributed to its correction operation that reduces inter-slice noise correlation. To the best of our knowledge, Noise2Match is the first effective image-domain self-supervised method for low-dose CBCT. Extensive experiments demonstrate that Noise2Match outperforms existing image-domain selfsupervised methods, offering broader applicability and higher practical value.
[CV-60] SchemaPose: RGB-Based Category-Level Pose Estimation with Parametric Category Schema NEURIPS2026
链接: https://arxiv.org/abs/2610.18374
作者: Shuang Wu,Jiude Wei,Cewu Lu,Jianhua Sun
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at NeurIPS 2026. Project page: this https URL
Abstract:RGB-based category-level 9D pose estimation remains highly challenging because a monocular system must generalize across unseen object instances while jointly inferring 3D translation, 3D rotation, and 3D size. Unlike instance-level pose estimation, the category-level setting requires the network to learn both what is shared across objects within a category and how individual instances vary in structure. This challenge is amplified in the RGB-only setting, where pose cues, category-level commonality, and instance-specific variation are entangled in appearance without direct geometric constraints, making it difficult for existing frameworks to learn a representation that is truly aligned with the category-level estimation objective. To address this problem, we propose SchemaPose, a schema-guided framework for RGB-based category-level pose estimation. SchemaPose equips the estimation pipeline with an explicit parametric schema for modeling category-level regularities together with structured instance variation, and uses the inferred schema variables to guide pose-related prediction in an end-to-end query-based framework. The same formulation also supports template instantiation, synthetic annotation, and pose learning within a unified framework. Experiments on synthetic and real data show that SchemaPose achieves state-of-the-art performance among RGB-based methods for category-level 9D pose estimation.
[CV-61] Hallucination-aware Faithful Reasoning in Remote Sensing
链接: https://arxiv.org/abs/2610.18369
作者: Yi Liu,Jing Zhang,Xiang Feng,Xiaoyu Tian,Haonan Guo,Di Wang,Zengmao Wang,Bo Du
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal large language models (MLLMs) have demonstrated promising capabilities for remote sensing reasoning, yet they frequently generate visually unsupported reasoning that contradicts the visual evidence. Existing post-training methods optimize primarily for final-answer correctness, providing limited supervision for identifying and correcting hallucinations that arise during the reasoning process. To address this limitation, we propose \textbfFaithRS, a hallucination-aware self-reflection framework for faithful reasoning in remote sensing. We first construct \textbfSFHD-43K, a dataset of 43K structured, fine-grained hallucination annotations, and train a detector to identify hallucination types, underlying causes, and supporting visual evidence. FaithRS then incorporates this diagnostic feedback into a Safe Detector-Guided Corrective Rollout framework, where a safety-aware gate selectively determines whether and how to revise the response. Training is driven by a composite reward that jointly optimizes answer accuracy, reasoning faithfulness, hallucination reduction, format compliance, and revision safety. Experiments with Qwen3.5 backbones (0.8B, 4B, and 9B) show consistent improvements across four remote sensing benchmarks, with gains of up to 11.64 points.
[CV-62] Visual Sevoing from Lines
链接: https://arxiv.org/abs/2610.18363
作者: Nicolas Andreff,Bernard Espiau,Radu Horaud
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:In this work, we present a fundamentally new approach to visual servoing using lines. It is based on a theoretical and geometrical study of the main line representations which allowed us to define a new representation, the so-called binormalized Plucker coordinates. They are particularly well suited to visual servoing. Indeed, they allow the definition of a proper image line alignment concept. The control law which realizes such an alignment has moreover several properties: partial decoupling between rotation and translation, analytical inversion of the motion equations and global asymptotic stability conditions. This control law was validated both in simulation and experimentally in the specific case of an orthogonal trihedron.
[CV-63] Urbex: Agent ic Spatial Grounding for City-Scale 3D Scenes NEURIPS2026
链接: https://arxiv.org/abs/2610.18356
作者: Yu Wang,Rui Dai,Wei Chen,Yi Wang,Bo Dang,Kaikui Liu,Xiangxiang Chu,Yansheng Li
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: NeurIPS 2026. 23 pages, 10 figures. Code: this https URL
Abstract:City-scale 3D grounding is commonly formulated as candidate ranking, which requires precomputed instance proposals and becomes inefficient in large outdoor scenes. We propose Urbex, an agentic vision-language framework that instead casts grounding as active spatial search. Urbex represents a city-scale 3D scene as an interactive multi-view environment, allowing a VLM agent to query landmarks, zoom into local regions, render oblique views, and commit a grounded bounding box. We optimize this tool-use policy with reinforcement learning on CityRefer, using smooth localization rewards and lightweight shaping rewards for efficient exploration. Experiments show that Urbex achieves the best Hit@1 and IoU on CityRefer, supports candidate-free grounding without ground-truth boxes, and is substantially faster than the strongest city-scale ranking baseline. Zero-shot results on STPLS3D-Refer further indicate improved cross-scene robustness.
[CV-64] CityCode: Building Agent -Operable 3D Cities from Sparse Aerial Observations
链接: https://arxiv.org/abs/2610.18342
作者: Jiahui Sun,Ran Bi,Yifei Liu,Xiaolong Zhou,Yishuo Xu,Yuanqi Li,Congcong Zhu,Lei Yang,Yanwen Guo,Zhihang Zhong,Hongjie Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 13 pages, 4 figures;
Abstract:Urban spatial intelligence calls for 3D cities that agents can measure, interpret, and modify across successive tasks. Maintaining such cities as persistent, code-based states requires lightweight acquisition and support for consistent state updates without repeated dense reconstruction. We therefore formulate a new setting for grounded urban world construction: generating large-scale, instance-level parameterized 3D cities from one low-altitude aerial RGB image per region and known camera parameters, while recovering metrically scaled buildings and explicit road geometry and connectivity. We introduce CityCode, an agent-operable representation that unifies metric geometry, urban topology, visual semantics, and 3D assets through persistent entity identities, with dependency-aware versioned execution for validated updates and historical restoration. We further develop Air2Code to construct this representation by combining visual structure extraction with adapted metric depth estimation. To address occlusion and incomplete coverage, recovered structure and observed appearance guide generative completion and subsequent structure-aware 3D generation for buildings and backgrounds. Experiments on CitySample-Aerial and UrbanScene3D demonstrate substantially more accurate metric structure recovery and superior visual fidelity to the target cities compared with the evaluated baselines. Across 50 task chains of 50 steps each, CityCode supports continued querying, editing, planning, and historical restoration over evolving city states. Together, CityCode and Air2Code turn sparse aerial observations into persistent, agent-operable urban worlds for long-horizon urban spatial intelligence.
[CV-65] ShoeFit: A New Dataset and Dual-image-stream DiT Framework for Virtual Footwear Try-On NEURIPS2025
链接: https://arxiv.org/abs/2610.18325
作者: Yuhan Li,Zhiyu Jin,Yifan Tong,Wenxiang Shang,Benlei Cui,Xuanhong Chen,Hangcheng Zhu,Bingbing Ni
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted at NeurIPS 2025. Project website: this https URL , including proposed multi-view footwear try-on dataset MVShoes-clean
Abstract:Virtual footwear try-on (VFTON), a critical yet underexplored area in virtual try-on (VTON), aims to synthesize faithful try-on results given diverse footwear and model images while maintaining 3D consistency and texture authenticity. Unlike conventional garment-focused VTON methods, VFTON presents unique challenges due to (1) Data Scarcity, which arises from the difficulty of perfectly matching product shoes with models wearing the identical ones, (2) Viewpoint Misalignment, where the target foot pose and source shoe views are always misaligned, leading to incomplete texture information and detail distortion, and (3) Background-induced Color Distortion, where complex material of footwear interacts with environmental lighting, causing unintended color contamination. To address these challenges, we introduce MVShoes, a multi-view shoe try-on dataset consisting of 7305 well-annotated image triplets, covering diverse footwear categories and challenging try-on scenarios. Furthermore, we propose a dual-stream DiT architecture, ShoeFit, designed to mitigate viewpoint misalignment through Multi-View Conditioning with 3D Rotary Position Embedding, and alleviate background-induced distortion using the LayeredRefAttention which leverages background features to modulate footwear latents. The proposed framework effectively decouples shoe appearance from environmental interferences while preserving high-quality texture detail through decoupled denoising and conditioning branches. Extensive quantitative and qualitative experiments demonstrate that our method substantially improves rendering fidelity and robustness under challenging real-world product shoes, establishing a new benchmark in high-fidelity footwear try-on synthesis.
[CV-66] Omni-RS: Unifying Multi-Task Understanding and Generation in Remote Sensing
链接: https://arxiv.org/abs/2610.18323
作者: Pan Yin. Shuai Yang,Yuanhao Liu,Datao Tang,Jialin Li,Hao Wang,Pei Liu,Kaiyu Li,Cong Zhao,Xiangyong Cao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Practical remote sensing applications often require both understanding and generation capabilities for the same geographic scene. However, most existing methods address either understanding or generation in isolation, while existing unified remote sensing models’ generative ability is limited to text-to-image. To bridge this gap, we propose Omni-RS, a unified framework supporting four remote sensing understanding tasks and five generation tasks. Specifically, we introduce the “Scene-Guided Two-Stage Training” strategy, which first adapts a multimodal large language model to remote sensing understanding and then uses its hidden states to guide a diffusion transformer for image generation. To enable joint learning across diverse generation tasks, we construct UniGen, a unified multi-task generation dataset, by organizing heterogeneous remote sensing data into a unified format. Furthermore, we introduce “Detailed Visual Conditioning”, which uses visual conditions encoded by a VAE to guide the generation process, thereby supporting additional generation tasks that require reference images. Extensive experiments demonstrate that Omni-RS performs exceptionally well in understanding tasks and has achieved outstanding results in a variety of generative tasks.
[CV-67] AffectWorld: Benchmarking Emotion-Driven Future World Prediction
链接: https://arxiv.org/abs/2610.18305
作者: Changhao Song,Yacong Chen,Hui Gao,Yazhou Zhang,Peng Zhang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:The world encompasses both physical environments and human actors. Predicting its evolution requires understanding both physical laws and how emotions shape human actions. However, existing world model benchmarks primarily assess physical plausibility and perceptual quality, leaving emotion-driven world evolution largely this http URL introduce AffectWorld, a benchmark that evaluates the chain Emotion-Action-World State through three tasks: World-State Consistency, Emotion-Action Consistency, and Causal Consistency. It comprises 1,498 everyday video samples annotated with plausible emotions, action tendencies, and atomic world-state constraints. A counterfactual subset of 233 samples tests whether changing an agent’s emotion under fixed scene and event conditions yields corresponding changes in actions and consequences. We define eight metrics across the three tasks and an additional counterfactual metric, combining automatic visual assessment with human-calibrated VLM/LLM judgments. Evaluations of five SoTA world models (e.g., Cosmos3) required world-state change scores of 0.100-0.280 and counterfactual emotion sensitivity scores of 0.027-0.070 on a [0, 1] scale. These results reveal substantial gaps in generating the consequences of emotion-driven actions and adapting future predictions to emotional interventions. AffectWorld provides a testbed for advancing world models that capture how emotions shape future physical and social states.
[CV-68] Robust Logical Anomaly Detection under Capture Shifts via Synthetic Anomalies
链接: https://arxiv.org/abs/2610.18295
作者: Ufuk Efe,Alptekin Temizel
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 42 pages, 10 figures, 15 tables
Abstract:Detectors of logical anomalies, such as missing, extra or wrongly combined parts, usually score a test image by its distance from normal (good) images. They achieve near-perfect image-level AUROC on standard industrial anomaly detection benchmarks such as MVTec LOCO, but their robustness to changes in capture conditions remains largely unexamined. Under a heterogeneous perturbation protocol in which each test image is randomly kept unaltered or subjected to a brightness reduction or a zoom-out transformation, the normal-reference methods we tested lose 9.7 to 34.1 AUROC points, and two of the strongest, CSAD and SALAD, fall from 97.7% and 96.4% to 68.1% and 62.3%. We propose a framework in which a classifier, trained on normal images and on synthetic anomalies, scores a test image without comparing it with normal data. From a short description of each product, a language model derives anomaly types and editing instructions, an image editor applies them to normal images, and the classifier learns to separate the edited images from automatically discovered normal sub-modes. With its full training recipe, the framework reaches 94.9% AUROC on clean images, 1.5 to 2.8 points below SALAD and CSAD, and 94.1% under the perturbation protocol. Trained without brightness or zoom-out augmentation, so that neither test perturbation occurs in training, it still reaches 92.5% (93.2% clean). These results suggest that clean benchmark scores can hide a strong sensitivity to capture conditions, and that learning the decision from both normal images and rule violations may reduce it.
[CV-69] Unsupervised and Supervised Approaches for Infrared Small Target Detection : Overview Challenges and Perspectives (Part 2)
链接: https://arxiv.org/abs/2610.18291
作者: Thierry Bouwmans,Anastasia Zakharova,Belmar Garcia,Marwa Chendeb El Rai,Meghna Kapoor,Badri Narayan Subudhi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Since the rise of supervised deep learning modelet dets, researchers has attempt to solve infrared small targection using various models like conventional neural networks, GANS, Transformers, Mixture of Experts and recent graph/hypergraph neural networks. Moreover, these deep neural networks methods surpassed the traditional unsupervised models and the low-rank and sparse decomposition unsupervised models reviewed in "Traditional and Recent Approaches for Infrared Small Target Detection : Overview, Challenges and Perspectives (Part 1). In this context, the aim of this paper is to provide a complete survey of the deep learning models in infrared small target detection. First, we categorize the different approaches found in the literature in terms of the modeling models of strategies. Then, we have discussed them in terms of the critical situations that they claim to handle. Furthermore, we present the specific performance evaluation. Then, we conclude with several promising directions for future research.
[CV-70] LongForcing: Scaling Temporal Manifold Matching for Long-Horizon World Rollout
链接: https://arxiv.org/abs/2610.18289
作者: Ziyu Zhu,Zhaoxu Sun,Fan Jiang,Mengchao Wang,Junfeng Ni,Ting-Bing Xu,Zhicheng Liu,Mu Xu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Autoregressive video diffusion models have emerged as a foundation for long-horizon world rollout. However, when extended far beyond their training windows, these models suffer from progressive drift: visual quality deteriorates, motion becomes inconsistent, and object identity gradually collapses as generated frames are recursively reused as context. We study a supervision-horizon gap: short-horizon distillation leaves later rollout states without direct supervision. We show that matching short-window marginals need not determine long-horizon behavior, while longer on-policy trajectories expose delayed failures to training. This motivates temporal manifold matching at scale for long-horizon world models. Building on this insight, we introduce LongForcing, a distillation approach that scales teacher supervision to long student self-rollouts. LongForcing preserves the short-horizon teacher-forcing and causal ODE-distillation stages, while replacing the realscore teacher in the final distribution-matching stage with an extended-horizon bidirectional teacher. The long-horizon teacher supervises extended student self-rollouts using the standard distribution-matching formulation, without modifying the student architecture or increasing inference cost. We evaluate LongForcing on text-to-video generation and interactive action-conditioned world models, achieving a VBench Quality score of 85.52 and improving WorldRoamBench 60s action score by 23.9 points in first-person ablations. Qualitative stress tests further sample interactive rollouts up to 24 hours, far beyond the training horizon. These results suggest that scaling supervision along the temporal axis is a practical step toward long-horizon world models. The project website and code are at this https URL.
[CV-71] Efficient Scene-Text Refinement via Pixel-Space Grouped Image Generation
链接: https://arxiv.org/abs/2610.18284
作者: Yaoli Liu,Guo-Hua Wang,Yiyang Chen,Qing-Guo Chen,Zhao Xu,Weihua Luo,Yao-Xiang Ding
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 29 pages, including appendices
Abstract:Generated images often have coherent layouts and visual content, yet local errors in small text and complex characters still limit their readability and practical use. Scene-text refinement requires accuracy, preservation, and efficiency. Existing methods based on variational autoencoders (VAEs) can lose fine strokes during reconstruction, while full-image editing can shift background colors. In this work, we introduce PixText, a model that formulates local text refinement as grouped image generation in pixel space. Conditioned on the original image, PixText generates only the target regions, preserves unedited content, and uses MeanFlow for four-step generation. We further introduce Text Refinement Bench to evaluate text correction and image preservation in generated images. On this benchmark, PixText achieves 83.06% glyph pass and 94.66% exact target-character OCR accuracy, versus 36.77% and 52.31% for the strongest baseline. Its four-step MeanFlow variant retains an 81.84% pass rate with less than one GPU-second of amortized generation cost per image.
[CV-72] raditional and Recent Approaches for Infrared Small Target Detection : Overview Challenges and Perspectives (Part 1)
链接: https://arxiv.org/abs/2610.18276
作者: Thierry Bouwmans,Anastasia Zakharova,Belmar Garcia,Marwa Chendeb El Rai,Meghna Kapoor,Badri Narayan Subudhi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Infrared small target detection is often used in different civilian and military applications to detect static or moving targets in scenes like in ground, air and sea surveillance. First valuable research has emerged since 30 years ago and the two last decades witnessed very significant publications in this field following the rise of low-rank and sparse decomposition (model-based methods) and deep learning models (data-based methods). Furthermore, several surveys can be found in the literature but none of them addresses an overall review in this field. In this context, the aim of this paper is to provide a complete survey of the traditional and recent approaches in infrared small target detection. First, we categorize the different approaches found in the literature in terms of the modeling models, i.e mathematical models, machine learning models and signal processing models. Then, we have discussed them in terms of the critical situations that they claim to handle. Furthermore, we present the available resources, datasets and libraries. Then, we conclude with several promising directions for future research. For the sake of reading, this survey is divided into two parts: 1) “Traditional and Recent Approaches for Infrared Small Target Detection : Overview, Challenges and Perspectives (Part 1)” about conventional and low-rank and sparse decomposition (model-based methods), and 2) “Traditional and Recent Approaches for Infrared Small Target Detection : Overview, Challenges and Perspectives (Part 2)” about deep learning models (data-based methods).
[CV-73] FALLNEST: Dual-Domain Dataset Condensation of Imbalanced IMU Data for Wearable Fall Detection
链接: https://arxiv.org/abs/2610.18260
作者: Rashi Niyas P,Ritika Jain,Hitika Tiwari
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Signal Processing (eess.SP)
备注: 10 pages, 2 figures, for Conference
Abstract:Wearable fall detection is hampered by rare fall events, scarce labeled IMU data, and the cost of training deep models on large, imbalanced corpora. Existing methods either rely on hand-crafted features with shallow classifiers, which limits learning from raw signals, or train deep networks on full datasets that require re-sampling to counter class skew and demand substantial data and compute. To address this, we propose a novel Fall detection framework via Normalized-gradient and Embedding matched Synthetic Time-frequency condensation, dubbed FALLNEST, a data-centric approach that distills large, imbalanced IMU datasets into tiny, class-balanced synthetic training sets. Building on dual-domain time-series condensation (CondTSC), FALLNEST introduces a fall-specific 7-channel input with accelerometer vector magnitude, class-balanced K-Means initialization, and normalized gradient matching with gradient clipping, combined with MMD-based embedding alignment and multi-view Fourier-domain augmentation, so that class balance arises from the condensed set itself rather than from re-sampling. We evaluate on SisFall, KFall, and MobiFall using strict subject-wise splits and naturally imbalanced test sets, reporting accuracy, Macro-F1, and PR-AUC over five seeds. A 102K-parameter TinyCNN trained on only 32 to 146 synthetic windows (about only 1% of the training windows) reaches 93.5-95.8% mean accuracy and 0.89-0.98 PR-AUC, with standard deviations of 1.0-1.7% in accuracy, demonstrating the efficacy of our approach.
[CV-74] LiFUSE: Frequency-Spatial Unified Self-Supervised Pretraining for Liver Disease Analysis on Non-contrast CT
链接: https://arxiv.org/abs/2610.18243
作者: Hao Zhang,Yi Huang,Wei Liu,Yuan Gao,Zeju Li,Xu Han,Yu Shi,Ling Zhang,Ke Yan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted at the 2026 IEEE International Conference on Bioinformatics and Biomedicine (BIBM 2026)
Abstract:Given the growing global burden of liver tumors and diffuse liver diseases, non-contrast computed tomography (NCCT) is increasingly used for opportunistic screening because of its low cost and favorable safety profile. However, downstream tasks such as tumor detection and fine-grained disease diagnosis remain challenging due to NCCT’s inherently low tissue contrast, lack of hemodynamic enhancement, and complex anatomical background. These limitations make subtle, heterogeneous, and multifocal hepatic lesions difficult to identify. Although large-scale self-supervised pretraining can learn robust representations from unlabeled NCCT scans, conventional voxel-wise reconstruction does not fully exploit the complementary information available in NCCT. Under weak NCCT contrast, low-level intensity patterns become less informative for pathology recognition, limiting the capture of lesion-specific discriminative features. In this paper, we propose the Liver Frequency-Spatial Unified Self-Supervised Pretraining (LiFUSE). The frequency branch applies random band-stop masking and spectrum reconstruction to encourage the model to capture subtle differences between lesions and surrounding tissues, while the spatial branch performs hybrid-granularity masked volumetric reconstruction to accommodate lesions of diverse sizes and morphologies. Finally, a cross-domain consistency scheme couples the spatial and frequency branches, yielding more discriminative representations for hepatic structures and lesions. Extensive experiments on three clinically relevant tasks across internal cohorts and a public target-domain transfer setting show that LiFUSE consistently outperforms strong baselines. These results suggest that unified frequency-spatial learning can improve liver disease analysis on NCCT, highlighting its potential for large-scale screening.
[CV-75] Heading Hierarchy Reconstruction for Complex and Long Documents with Level-Wise Multimodal Memory
链接: https://arxiv.org/abs/2610.18238
作者: Qian Chen,Xianyin Zhang,Lifan Guo,Chi Zhang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Vision-language models (VLMs) have been widely adopted for document parsing and optical character recognition (OCR), as they can accurately extract page-level elements. However, downstream applications such as retrieval-augmented generation (RAG) also require coherent document-level structure. Although these models indicate heading levels using different numbers of preceding ``#‘’ symbols, page-by-page prediction often lacks document-level consistency. In practice, input-resolution and context-length constraints make it infeasible for a VLM to jointly process all page images in documents containing tens or hundreds of pages. To address this limitation, we propose a VLM-based framework that incrementally reconstructs heading hierarchies with a \emphlevel-wise multimodal memory. The memory retains a compact subset of all previously processed pages, selected by anchors. Our framework supports flexible anchor selection, and in this work we select the most recent node at each level, allowing the VLM to observe the visual, textual, and layout patterns of headings at every level. For comprehensive evaluation, we also introduce the Complex and Long Document Heading Hierarchy Reconstruction Benchmark (CLDoc-HHRBench), comprising 250 documents and 11,212 pages with expert-verified annotations. Experiments on CLDoc-HHRBench and three existing benchmarks demonstrate the effectiveness of our method.
[CV-76] A Controlled Evaluation of Curvature Inspired Scalar Feature Scaling for Domain Generalization
链接: https://arxiv.org/abs/2610.18227
作者: Sunny Gupta,Amit Sethi
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at TMLR, 9 pages, 1 figures
Abstract:Domain-specific curvature is an appealing idea for domain generalization (DG), but curvature learning can denote substantially different computations. This paper evaluates one deliberately narrow implementation: the code calls it HyperDG, but it rescales Euclidean encoder features during training by learned positive scalars and uses an ordinary linear classifier on unscaled Euclidean features at inference. It does not implement a Lorentz or Poincaré representation, geodesic prediction, or a Sharpness-Aware Minimization inner loop. We audit the executed objective and evaluate six conditions on PACS, VLCS, Office Home, and TerraIncognita. The completed design contains 384 runs: six conditions, four held-out targets, and four run seeds per dataset. Checkpoints are selected using only the three source-validation environments. Inference first averages paired seed differences within each held-out target; targets, rather than 16 seed-by-target observations, are the per-dataset clusters. The full implementation’s equal-dataset macro point estimate is 1.61 percentage points below native DomainBed ERM (95% clustered CI [-3.17, -0.04]). All four dataset point estimates favor ERM, but an exact four-dataset sign-flip test gives p = .125 and Holm-adjusted p = .250; we therefore claim no adjusted evidence of a difference. The cleaner domain-specificity contrast, separate learned source scales without their variance penalty versus one shared learned source scale, is -0.13 pp (95% CI [-0.53, 0.27]). Its 90% CI lies within a revision-stage \pm 1 pp smallest effect of practical interest (TOST p = .003), supporting equivalence only for the source-selected suite macro, not every dataset, target, or final-step endpoint. The study thus supplies a bounded negative result for this scalar training package; it does not test or refute genuine hyperbolic representation learning.
[CV-77] Closing the Estimation Gap: Training-Free Transductive Inference for Cross-Domain Few-Shot Action Recognition BMVC
链接: https://arxiv.org/abs/2610.18209
作者: Changyi Li,Zeyu Yang,Yu Xiao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by BMVC Workshop 2026
Abstract:Cross-domain few-shot action recognition (CD-FSAR) aims to recognise unseen action classes in a target domain from a handful of labelled videos. Despite advances in representation learning, CD-FSAR inference still relies on prototypes estimated from extremely limited support examples. To assess this bottleneck, we perform an oracle analysis with the representation fixed. Constructing oracle prototypes and decision boundaries from ground-truth query labels improves 5-way 1-shot accuracy by 7–20 percentage points across five target datasets. This reveals a substantial estimation gap: the representation contains useful discriminative information, but prototypes and boundaries estimated from scarce supports fail to capture each episode’s true class structure. Motivated by this, we propose TRACE, a training-free transductive inference framework that improves episode-level estimation without modifying the backbone. TRACE comprises two complementary stages. Confidence-Thresholded Transductive Prototype Refinement (CTPR) refines support-based prototypes using confidently predicted queries. The refined prototypes initialise a lightweight per-episode classifier, which Transductive Classifier Refinement (TCR) further adapts by encouraging confident, class-balanced predictions over all queries. Together, the two stages improve prototype and boundary estimation, enabling better use of information in the unlabelled queries. Across five target datasets (HMDB51, UCF101, SSv2, Diving48, and RareAct) in both 1-shot and 5-shot settings, TRACE consistently improves a strong CD-FSAR baseline, including gains of more than six percentage points on HMDB51 1-shot, while recovering much of the oracle-estimated gap. These results suggest that improving test-time estimation is a powerful, broadly applicable complement to representation learning for CD-FSAR under extreme support scarcity.
[CV-78] VARI: Robust Text-to-Video Generation via Group-Structured Intent Alignment
链接: https://arxiv.org/abs/2610.18189
作者: Mingxuan Zhu,Hohei Chan,Shiyuan Liu,Yuanxing Zhang,Lu Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Modern text-to-video (T2V) models generate visually compelling videos, yet meaning-preserving changes in prompt wording can substantially alter semantic realization and overall quality. Existing prompt rewriting modifies the inference input, whereas conventional post-training constructs supervision from one prompt formulation at a time; neither explicitly optimizes reliability across equivalent forms. We introduce VARI (Verified Alignment over Representations of Intent), a group-structured framework that treats automatically verified formulations of the same visual intent as a shared supervision unit. VARI builds matched form-seed rollout pools, cross-scores each video against all forms, and applies a wording-independent structured intent audit to identify valid realizations and hard semantic failures. Group-Consistent Distillation (GCD) aligns every verified form with interchangeable valid realizations, while Form-Contrastive Flow-DPO (FC-DPO) learns group-level preferences from audited positive-negative pairs, prioritizing same-seed, different-form contrasts. On HunyuanVideo-1.5, VARI attains the highest T2V-CompBench, VBench, and StoryEval scores among the evaluated methods. On prompt-form extensions of T2V-CompBench and VBench, it improves weakest-form alignment over the strongest prompt-level baselines by 2.45 and 3.55 points and reduces cross-form drop by 30.5% and 32.2%, respectively, while preserving visual quality. Human evaluation, natural-caption controls, prompt-enhancement experiments, diversity analyses, and transfer to LTX-2.3 further support the robustness and transferability of these gains, while VARI requires no additional inference-time module.
[CV-79] Benchmarking Multimodal Large Language Models for Zero-Shot Human-Object Interaction
链接: https://arxiv.org/abs/2610.18161
作者: Oriol Rabasseda,Zenjie Li,Kamal Nasrollahi,Sergio Escalera
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal Large Language Models (MLLMs) have demonstrated strong zero-shot reasoning across diverse domains, but their potential for Human-Object Interaction (HOI) remains unexplored. Unlike traditional HOI approaches that rely on extensive labeled datasets and fixed interaction taxonomies, MLLMs offer flexible reasoning, making them well-suited for dynamic, real-world scenarios. In this work, we present the first systematic study of MLLMs for HOI recognition and detection within a pipeline that combines an external object detector for entity localization with zero-shot multimodal reasoning for interaction understanding. We introduce a benchmark framework for principled zero-shot evaluation, consisting of (i) a unified taxonomy of zero-shot prompting strategies, and (ii) a methodology that transforms token-level logits into interaction scores, enabling fair comparison with specialized HOI models under standard metrics. To support prompt-based evaluation, we additionally construct new question-answer annotations for standard HOI datasets, transforming them into a format suitable for MLLMs. Building on this, we propose two visual prompting strategies, namely Objects as Context and Black Other Objects, that guide MLLMs to target objects while preserving scene context. Experiments on HICO, HICO-DET, and V-COCO demonstrate state-of-the-art zero-shot performance, achieving 53.50 mAP on HICO and 23.69 mAP on HICO-DET. Overall, our work provides a principled benchmarking framework for MLLMs in HOI, opening the door to new models and datasets and enabling flexible open-world interaction understanding.
[CV-80] FlowCred: Hierarchical Credit Assignment and Functional Coordination for Joint Text-to-Video Post-Training
链接: https://arxiv.org/abs/2610.18142
作者: Mingxuan Zhu,Yanwei Huang,Shiyuan Liu,Xinzhi Zhang,Shujie Li,Yuanxing Zhang,Lu Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Prompt optimization and generator post-training are two complementary ways to improve text-to-video (T2V) generation, and recent systems increasingly train both modules jointly. However, optimizing a prompt policy and a video generator from the same terminal reward creates two ambiguities: prompt quality is confounded with stochastic generation, and the two modules may realize overlapping local improvements. We introduce FlowCred, a two-level framework for joint T2V post-training. First, hierarchical credit assignment uses matched prompt-seed rollout grids to separate prompt-level credit from prompt-conditioned generator residuals under shared randomness. Second, functional coordination probes alternative current-policy prompts at states from actual generation trajectories, providing generator-aware evidence for prompt optimization while selectively attenuating generator updates already supported by conditioning changes. Across multiple T2V benchmarks, FlowCred consistently improves matched joint-training baselines. On LTX-2.5, FlowCred improves a strictly matched joint-training control from 61.47 to 64.46 on T2V-CompBench and from 55.58 to 58.08 on StoryEval, without increasing decoded-video or reward-model-call budgets. Factorial and realized-update analyses show complementary contributions from prompt shaping and generator routing, while human evaluation and transfer across backbones and updater families provide further supporting evidence.
[CV-81] Human-Inspired Asymmetric Collaboration for Dual-Student Semi-Supervised Medical Image Segmentation
链接: https://arxiv.org/abs/2610.18133
作者: Huiyi Ren,Ziming Fang,Jiaming Cai,Rui Xu,Haojie Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Dual-student learning enriches predictive diversity in SSMIS by introducing a strongly augmented view alongside a weakly augmented view. Insufficient strong augmentation may homogenize the students, whereas excessive augmentation can corrupt anatomical evidence and destabilize collaboration. This challenge echoes feedback-regulated human instruction, where learners receive differentiated tasks and the task difficulty is continually adjusted according to their evolving responses. Inspired by these, we propose Human-Inspired Asymmetric Collaboration (HAC), a closed-loop dual-student framework that maintains weak–strong learning through adaptive difficulty calibration and geometry preservation. Specifically, Reinforcement-guided Difficulty Calibration (RDC) formulates strong-augmentation control as a window-level reinforcement learning process, using augmentation history, realized perturbations, and training responses to dynamically adjust the augmentation magnitude, operation number, and magnitude range. Meanwhile, the Gramian Geometry Preservation Constraint (G2PC) preserves the early-stage three-branch geometry through global volume and pairwise area constraints, preventing shared supervision from homogenizing the two students. Comprehensive evaluations across LA, Pancreas, and BraTS-2019 validate both the segmentation effectiveness and capability-aware difficulty calibration of HAC.
[CV-82] H-DDCM: Hadamard Codebooks for Fast Generative Image and Video Compression
链接: https://arxiv.org/abs/2610.18125
作者: Or Ronai,Vladimir Kulikov,Tomer Michaeli
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Project’s webpage at this https URL
Abstract:Training-free diffusion-based compression methods repurpose the generative priors of a diffusion or a flow model to compress data without retraining them as task-specific codecs. Recent codebook-based approaches such as Turbo-DDCM achieve strong perceptual quality at ultra-low bitrates, but their reliance on dense Gaussian codebooks remains costly in both memory and computation, and scales poorly with the large latent dimensions of modern image and video generators. Their combinatorial bitstream encoding further becomes computationally heavy at this scale. We introduce H-DDCM, a scalable generative compression framework that replaces Gaussian codebooks with implicit randomized Hadamard dictionaries. The fast Walsh-Hadamard transform allows for implicitly synthesizing full-dimensional dictionaries with O(d \log d) compute for projection and reconstruction and O(d) memory, without materializing the full-dimensional dictionary. We further introduce an efficient table-free lexicographic coding scheme that makes combinatorial encoding practical at larger dimensions. Beyond images, we adapt the framework to an autoregressive video flow model and develop a streaming compression scheme that periodically re-encodes recently reconstructed frames into a fresh KV-cache prefix, preserving temporal context across segments at no additional transmitted bits. This enables long-form, close to real-time generative video compression on a single GPU. Across image and video benchmarks, H-DDCM achieves competitive rate-distortion-perception performance while substantially reducing the memory and computational overhead of codebook-based training-free generative compression.
[CV-83] SYNPRED: A Synergistic Approach to Multimodal Learning for Clinical Prediction
链接: https://arxiv.org/abs/2610.18110
作者: Ivana Janíčková,Yen Y. Tan,Thomas H. Helbich,Philipp Seeböck,Julieta Di Marco,Zsuzsanna Bago-Horvath,Ulrike Heber,Thomas Spiegel,Georg Langs
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal medical data captures complementary aspects of disease. It provides a comprehensive basis for diagnosis and the identification of biological associations linked to progression or treatment response. Established contrastive learning approaches exploit shared information across modalities, but fall short in exploiting complementary, modality-specific signals relevant for a task. Here, we propose a synergistic multimodal variational autoencoder (VAE) framework that separates task-relevant unimodal and multimodal information from instance-specific variation task-irrelevant, allowing complementary signals across modalities to be leveraged for prediction. The method yields benefits even if modalities are missing at test time. We evaluate the method on a treatment response prediction task using medical imaging and RNA data. Empirical comparison with alignment-based and multimodal VAE state-of-the-art approaches demonstrates improved multimodal prediction performance, and consistent benefits for prediction from a single modality after multimodal training. The full code can be found here: this https URL.
[CV-84] APT: Feedback-Driven Agent ic Post-Training for Text-to-Image Generation
链接: https://arxiv.org/abs/2610.18097
作者: Sirui Zhang,Fangyun Wei,Dong Chen,Jinjing Zhao,Yitong Wang,Yunuo Chen,Baining Guo,Zheng-Jun Zha
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Post-training is crucial for improving text-to-image models, yet existing pipelines typically rely on manually curated training data and fixed training distributions. We introduce APT (Agentic Post-Training), a fully automated framework that adaptively post-trains a base model according to its evolving capability profile. APT coordinates three agents to construct evaluation sets from user requirements, diagnose model-specific strengths and failure modes, and generate reinforcement learning data with an adaptive prompt distribution that emphasizes underperforming capabilities. By repeatedly evaluating, diagnosing, and updating the generator, APT lets the current model’s behavior shape its next training tasks, forming a feedback loop inspired by recursive self-improvement. For general-purpose settings without user-specified evaluation scenarios, we further develop an evaluation harness that enables broad and diverse evaluation set generation. Experiments across multiple base models, benchmarks, and generation architectures demonstrate that APT is model-agnostic, benchmark-agnostic, architecture-agnostic, and fully automated.
[CV-85] LensErase: A Foundation Model for Erase Anything
链接: https://arxiv.org/abs/2610.18072
作者: Yunuo Chen,Xiuyu Wu,Qinhong Yang,Jinjing Zhao,Yanchen Dong,Sirui Zhang,Dong Chen,Fangyun Wei,Lei Shi,Ji Li,Dongdong Chen,Guo Lu,Yan Lu,Baining Guo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: LensFamily Technical Report. Project page: this https URL
Abstract:We introduce LensErase, a general-purpose mask-based Erase Anything foundation model designed to cover a broad spectrum of real-world erase scenarios. LensErase supports diverse objects, visual text, watermarks, associated visual effects, multiple-object removal, and a wide range of object scales, aspect ratios, and image resolutions. First, to enable such broad coverage, we construct LensErase-100M, a large-scale dataset containing 101.07 million paired source-mask-target examples spanning diverse erase scenarios. Second, we pretrain LensErase from the text-to-image model Lens, adapting its strong generative prior to mask-based erasing. To efficiently incorporate the source image and erase mask, we propose Mask-Aware RoPE, a 3D rotary positional encoding that directly injects mask information into the source-image tokens without introducing a separate mask token sequence. Finally, we refine the model through supervised fine-tuning and reinforcement learning with contract-guided, region-aware rewards, and apply TraceFusion at inference to preserve high-fidelity content outside the effective edit region. Across six public benchmarks and the proposed LensErase-Bench, LensErase achieves substantial improvements under both reference-based objective metrics and reference-free VLM evaluation. We hope LensErase can serve as a foundational model for erase-related applications and systems.
[CV-86] Scaling Vision Transformers Without Training: Prune-and-Merge Token Compression NEURIPS2026
链接: https://arxiv.org/abs/2610.18039
作者: Vatsal,Soumil Malhotra,Priyanshu Chahal,Sanjeev Kumar
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted at the GlobalSouthAI Workshop at NeurIPS 2026; 12 pages
Abstract:Vision Transformers achieve strong performance but incur high computational costs from quadratic token interactions. Training-free token reduction methods improve efficiency, but their robustness remains underexplored. We propose Prune-and Merge ViT, a training-free framework for off-the-shelf models. It combines class attention-based pruning with similarity-based merging to reduce redundancy while preserving semantic information. To evaluate robustness we introduce two metrics. Relative Accuracy Drop (RAD) measures relative accuracy degradation under transformations, and Prediction Flip Rate (PFR) quantifies prediction consistency. On pretrained ViT and DeiT, Prune-and-Merge reduces computation competitively while maintaining robustness and better accuracy than existing token reduction methods. Our analysis suggests that hybrid token reduction does not introduce hallucination-like behavior and preserves prediction stability for common image transformations.
[CV-87] Data-Free Continual Learning for World Action Model with Rectified Generative Replay
链接: https://arxiv.org/abs/2610.18038
作者: Kaipeng Fang,Hangxu Liu,Chaojian Shi,Yunsong Zhou,Hongzi Zhu,Minyi Guo,Jingkuan Song,Lianli Gao
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:World action models (WAMs) built on pretrained video generation models adapt quickly to new robotic tasks, but they forget previously learned tasks under sequential fine-tuning. Experience replay (ER) mitigates such forgetting by rehearsing data of previous tasks, yet this data is often unavailable in robotics, where a policy is deployed as weights only and its training data is not shared along with it. In this work, we ask whether a WAM can achieve data-free continual learning by regenerating its own past. Our key observation is that a WAM can indeed autoregressively regenerate complete rollouts of previously learned tasks and replay them to mitigate forgetting. However, replaying such raw rollouts still lags far behind replaying real data, which we attribute to two issues. First, the WAM is unable to recognize task completion and keeps generating purposeless motion after the task is done. Second, under open-loop generation, the generated actions gradually drift away from the generated observations. To address both issues, we present rectified generative replay (R ^2 -WAM), which truncates generated rollouts at task completion with an agentic evaluator and filters out rollouts whose actions are physically inconsistent with the observations. On the four LIBERO suites and five real-world tasks, R ^2 -WAM reduces forgetting by over 90% relative to sequential fine-tuning and nearly matches ER without accessing any data from previous tasks. Our analysis further reveals that forgetting in WAMs concentrates in inverse dynamics modeling: a forgotten WAM still produces visually plausible observations of previous tasks while its generated actions fail.
[CV-88] COMPASS: Contextual Moment Prior Alignment for Generalizable AIGI Detection
链接: https://arxiv.org/abs/2610.18034
作者: Zhan Yang,Tiancheng Shen,Weijie Lyu,Shuo Chen,Bernard Ghanem,Yibo Yang,Ming-Hsuan Yang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Generalizable AI-generated image (AIGI) and deepfake detection remain challenging due to the asymmetry phenomenon, where adaptation to limited training distributions can degrade the transferable representations of Vision Foundation Models (VFMs). Recent subspace adaptation methods mitigate this issue by freezing high-magnitude principal components while adapting the residual subspace. However, such magnitude-based decomposition is data-agnostic and does not explicitly account for target-relevant semantics, potentially leading to semantic misalignment. We propose COMPASS (COntextual Moment Prior Alignment for Subspace Separation), a data-aware subspace adaptation framework that uses a small unlabeled Semantic Proxy Set to estimate target activation statistics. COMPASS incorporates activation covariance into the decomposition of pre-trained weights, preserving target-relevant representations as a stable semantic reference while the residual adapter learns forgery-related variations. The proxy calibration requires no gradient-based optimization. Extensive experiments demonstrate strong generalization across cross-dataset, cross-method, joint cross-dataset-method, and open-world cross-generator evaluations, achieving state-of-the-art performance on both Deepfake and AIGI detection benchmarks.
[CV-89] FlowSim: Rendering-Free Closed-Loop Simulation for Evaluating End-to-End Driving Systems
链接: https://arxiv.org/abs/2610.18030
作者: Hanfeng Wu,Alvaro Marcos-Ramiro,Michael Schmidt,Christoph Stiller
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Closed-loop evaluation of end-to-end (E2E) driving systems typically relies on generating sensor observations and re-processing them through the perception encoder, introducing substantial computational cost and a sim2real gap between simulated and real sensor observations. In this paper, we ask a more fundamental question: is image-space simulation necessary for closed-loop evaluation of E2E driving systems? We propose FlowSim, a rendering-free simulation paradigm that predicts the internal latent representation directly from a previous latent and target-frame symbolic scene conditions. The predicted latent is then processed by the original frozen downstream modules, eliminating image rendering and perception re-encoding. Explicit cross-frame agent association improves latent transition prediction and rollout stability, while variable-horizon simulation enables long-horizon prediction from a common real anchor without accumulating intermediate rollout errors. Experiments on nuScenes and NavSim demonstrate that FlowSim preserves downstream behavior competitively with, and often exceeds image-space simulation in source-to-reference agreement across variable simulation horizons and both dense and sparse latent representations. By predicting the latent directly, both instantiations of FlowSim simulate each step in 0.04–0.06s, substantially faster than the image-space simulation methods they are compared against. These results establish FlowSim as an efficient and practical alternative to conventional image-space simulation methods.
[CV-90] Dynamic Interpolation for Training-Time Prediction Reshaping in Low-Light Image Enhancement
链接: https://arxiv.org/abs/2610.18018
作者: Peirui Zhu,Shijie Hao,Jingxi Liao,Richang Hong,Meng Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 18 pages, 9 figures. Published in IEEE Transactions on Circuits and Systems for Video Technology (Early Access)
Abstract:Low-light image enhancement (LLIE) is an important image/video processing problem for reliable visual systems under poor illumination, yet it remains challenging due to the strong coupling among insufficient illumination, noise amplification, and structural degradation. Existing methods typically rely on sophisticated network architectures or engineered objective functions. While effective, the former increases model complexity and deployment burden, whereas the latter relies on empirically tuned loss weights. We revisit supervised LLIE from a training-dynamics perspective and observe that brightness-related errors tend to converge early, while residual noise and structural artifacts often remain insufficiently optimized in later stages. To address this issue, we propose a Dynamic Interpolation Training-only Module (DITM), which introduces no additional learnable parameters and adaptively reshapes the model prediction for supervised feed-forward LLIE training. Specifically, DITM employs brightness-aware interpolation, in which the model prediction is combined with a mean brightness-aligned image constructed by rescaling the model prediction according to the mean brightness of the ground-truth image. By reducing the relative dominance of global brightness discrepancy, DITM makes residual local errors more influential during backpropagation after global luminance has largely stabilized, thereby facilitating artifact suppression and fine-detail restoration in later training stages. DITM is used only during training and is fully removed at inference, introducing no additional inference cost. Extensive experiments on multiple benchmark datasets and CNN-, Transformer-, and Mamba-based LLIE backbones show that DITM consistently improves fidelity and perceptual quality while preserving the inference-time architecture and introducing no additional inference-time computation.
[CV-91] Surface Reconstruction from Range Scans Under Visibility Constraints
链接: https://arxiv.org/abs/2610.18005
作者: Gabriel Lenain,Edmond Boyer,Jean Ponce
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Surface reconstruction from range scans is challenging in the presence of noise and outliers, as fidelity to observations must be balanced against sensitivity to measurement errors. Yet, most neural approaches discard the sensor positions and corresponding visibility information, compensating for this loss with geometric priors or regularization terms. We instead formulate surface reconstruction as robust regression of the predicted first surface intersection along each acquisition ray against its measured depth, which penalizes predicted surface intersections within the segment between the sensor and the observed point. We represent the surface as the zero set of a 1-Lipschitz multi-layer perceptron, allowing the use of linearly convergent sphere tracing to compute the first hit, and use a robust Huber loss for regression. Our formulation requires neither oriented normals, sampled off-surface points, nor an eikonal loss. Comparisons with the state of the art on synthetic and real data with varying levels of noise and outlier contamination demonstrate competitive performance on clean data and much improved fidelity and robustness under even moderate corruption. Our analysis further confirms a trade-off between fidelity and robustness: greater fitting capacity improves data fidelity in the clean case but increases sensitivity to the measurement errors inherent in real acquisitions, with a sweet spot typically emerging at intermediate capacity.
[CV-92] Saliency-Guided Replay for Continual Learning in Deep Spiking Neural Networks KDD2026 ECML
链接: https://arxiv.org/abs/2610.17998
作者: Francesco Panetta,Federico Di Valerio,Michela Proietti
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Neural and Evolutionary Computing (cs.NE)
备注: Streaming Continual Learning Workshop. In post-proceeding of International Workshops of ECML PKDD 2026
Abstract:Spiking Neural Networks (SNNs) are promising for energy-efficient neuromorphic computing, but they still suffer from catastrophic forgetting when learning tasks sequentially. This work investigates the integration of Experience Packing and Replay (EPR) as a memory-efficient continual learning strategy for deep SNNs under low-latency constraints. The proposed framework combines Batch Normalization Through Time stabilized VGG9 SNNs with Spike Activation Maps (SAM), which provide forward-only explanations based on spike activity. These explanations are temporally aggregated into spatial attention maps and used to select the most informative image patches for replay, instead of storing full images. Experiments on Split CIFAR10 and Split CIFAR100 show that SAM-guided EPR improves the accuracy-memory trade-off under tight memory budgets, and remains competitive while requiring substantially less storage.
[CV-93] Uncertainty-Aware Semi-Supervised Framework with Supervised Contrastive Learning for Sugarcane Leaf Disease Detection
链接: https://arxiv.org/abs/2610.17996
作者: Shalini Dangi,Divya Teja,Omkar Kabde,Shahid Shafi Dar,Nagendra Kumar
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Sugarcane leaf diseases severely impact crop yield and quality, making accurate and timely detection critical for sustainable agriculture. Conventional deep learning models for plant disease detection depend on large annotated datasets, which are expensive to obtain and challenging to prepare due to challenges commonly observed across crop disease detection tasks, including (i) high intra-class variation caused by diverse symptom expressions across growth stages, (ii) subtle inter-class similarities between diseased and healthy tissues, and (iii) the risk of noisy labels degrading model reliability. To handle these challenges, we presented an Uncertainty-aware Mean-Teacher based Semi-Supervised Learning Disease Detection Network (U-SSDNet) that leverages unlabeled data while maintaining robustness. First, discrete wavelet transform preprocessing is applied to enhance lesion visibility and reduce intra- class variation. Subsequently, the KL divergence is employed to refine the pseudo-labels and reduce the impact of confirmation bias. At the same time, supervised contrastive learning is applied to make the features more distin- guishable and improve the separation among different classes. Experimental evaluations on sugarcane leaf disease datasets, considered a representative case study, demonstrate that the proposed approach achieves performance close to that of fully supervised models while requiring far fewer annotated samples and consistently outperforms state-of-the-art supervised and semi-supervised approaches.
[CV-94] PolUR: Polarization Aware Multimodal Degradation Perception for Underwater Image Restoration
链接: https://arxiv.org/abs/2610.17991
作者: Wenshuo Zhang,Xiangyue Zhang,Junlin Li,Wenjie Tian,Yubo Zheng,Hongli Xu,Jingyu Ru,Xinde Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages,10 figures
Abstract:Recovering high-quality images from degraded underwater observations is crucial for underwater exploration. Polarization observations show promise as an effective source of complementary physical cues for recovering target radiance under clear-water conditions. However, existing methods have yet to jointly model raw intensity information and polarization cues to enable image restoration driven by an understanding of mixed degradations (color casts, haze, and low light). To overcome these limitations, we propose PolUR, a unified multimodal language-driven framework that jointly uses polarization and RGB information to identify degradation types and severity levels through structured semantic instructions and enhances joint restoration representations under the guidance of degradation-aware semantic features to improve restoration robustness under mixed degradations. We further construct a new benchmark named MixPolar, which contains underwater images exhibiting real-world degradations and their corresponding degradation descriptions. MixPolar supports systematic evaluation of degradation perception, image restoration, and performance on downstream tasks. Extensive experiments demonstrate that PolUR effectively integrates polarization and RGB information, achieving competitive performance in degradation perception and image restoration, while also demonstrating generalization capability and utility for downstream tasks.
[CV-95] dMRI-F: A Foundation Model for Generalizable Diffusion MRI Analysis from Sparse Acquisitions
链接: https://arxiv.org/abs/2610.17968
作者: Xi Zhu,Yijie Li,Xiaofan Wang,Wei Zhang,Huafu Chen,Ce Zhu,Fan Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Diffusion MRI (dMRI) probes water diffusion within the brain, providing unique information about tissue microstructure and white matter connectivity. High-quality dMRI enables advanced and accurate analyses but demands dense acquisitions, resulting in long scan times and hardware constraints that are often infeasible. Further, dMRI requires dedicated computational methods to support diverse analysis tasks (e.g., parcellation, registration, and synthesis) for both neuroscientific and clinical applications. Here, we present dMRI-F, a foundation model that unifies dMRI data enhancement and downstream tasks across diverse acquisitions and populations, while requiring only minimal acquisition input. dMRI-F is pre-trained with a q-space masked reconstruction task: predicting densely sampled multi-shell data from only six volumes. It learns generalizable latent feature representations of dMRI data, which serve as a core foundation for adapting to diverse downstream tasks, including brain parcellation, brain registration, and cross-modal synthesis. Across nine independently acquired datasets, we demonstrate that dMRI-F outperforms state-of-the-art methods with robust generalization across acquisitions, populations, and tasks. Our results demonstrate dMRI-F’s versatility as a general-purpose framework that enables diverse dMRI applications from low-cost, sparse acquisitions.
[CV-96] FETALFusion: Resolution-Aware State-Space Modeling for Multi-Domain Fetal Head Segmentation
链接: https://arxiv.org/abs/2610.17951
作者: Ammar Bhilwarawala,Mainak Bandyopadhyay
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted at the 15th International Conference on Image Processing Theory, Tools and Applications, 2026
Abstract:Head circumference (HC) measured from 2D ultrasound is a routine marker of fetal growth, and automating it needs segmentations whose boundaries hold up on individual scans, not only on average. We present FETALFusion, a dual-path CNN-Mamba encoder-decoder for fetal head segmentation across acquisition types. It applies four-directional state-space scans only at the two highest-resolution stages, routes bottleneck features through domain-specific channel gates that are averaged at test time, and adds a weak landmark shape prior. We train jointly on transabdominal (HC-18) and transperineal (PSFHS) images and compare ten methods on a held-out test set of 471 images, each trained under one schedule and scored by surface distances at native resolution. FETALFusion reaches a Dice score of 0.963 and an HD95 of 2.10 mm without test-time augmentation or a domain label. It improves on U-Mamba on all five reported metrics (HD95 reduced by 0.31 mm; Holm-corrected ( p 10^-4 ), a lead that holds under test-time augmentation and across two independent training runs, and in the mean on U-Net, Attention U-Net and TransUNet. Residual encoders close the gap: a ResNet-34 UNet++ exceeds FETALFusion, and an ImageNet initialised ResNet-34 U-Net is best overall (HD95 1.75 mm). The ablation agrees. Calibrated against run-to-run variation, it attributes FETALFusion’s measurable gains to the landmark shape prior and, on transperineal images, to the domain gates, while removing the state-space branch does not measurably change accuracy. HC from the predicted masks agrees with the clinical reference at an intraclass correlation of 0.9986.
[CV-97] AI-Enhanced Hyperspectral Imaging for Intraoperative Detection of Low-Grade Glioma: A Clinical Study
链接: https://arxiv.org/abs/2610.17944
作者: Siri Willems,Ishika Jain,Jens De Winne,Siri Luthman,Hamed Zivari Adab,Roeland Vandebriel,Anna Alperovich,Xiaohan Zhang,Ben Stoffelen,Tom De Schepper,Lien Solie,Steven De Vleeschouwer
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Current image-guided neuronavigation techniques, such as intraoperative MRI and ultrasound, are limited by high costs, limited sensitivity and operational complexity. There is a clear need for more efficient systems to enhance intra-operative tumor detection and improve patient outcomes for intrinsic brain tumor resection. Whereas previous intraoperative brain HSI studies have predominantly investigated high-grade gliomas, glioblastomas, or mixed-grade cohorts, this clinical study focuses specifically on low-grade gliomas (LGGs). LGGs strongly resemble normal brain under white light conditions on the operative microscope. This study characterizes spectral differences between healthy and tumor tissue and provides a paired comparison of HSI- and RGB-based classification at matched tissue locations and under identical patient-level data splits. Tumor regions showed consistently lower reflectance, which may be associated with differences in hemoglobin concentration and oxygenation. A range of classification models was evaluated. The 3D CNN, which integrates spectral and spatial features, achieved the highest median ROC-AUC of 74.5%, compared with 67.9% for the spectral-only XGBoost model and 64.0% for the RGB-based model. These results support further investigation of HSI for low-grade glioma classification and quantify its performance relative to conventional RGB imaging in this cohort.
[CV-98] Sea-Net: A Small and Inherently Interpretable Neural Network for Time Series Classification
链接: https://arxiv.org/abs/2610.17939
作者: Ubaid Ur Rehman,Tanmoy Mondal
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Traditional time-series classification methods often operate as black boxes, providing a class prediction without indicating which time point or time interval within the series led to that prediction. To improve temporal interpretability, MILLET (Multiple Instance Learning with Temporal Explanation), which combines an InceptionTime encoder with Conjunctive Pooling, provides both classification and temporal explanations. However, MILLET contains approximately 424 K parameters, making it unsuitable for deployment on resource-constrained micro-controllers, and it achieves an overall time-series accuracy of 0.888 and an explanation score of 0.681 NDCG@n. We therefore propose SEA-Net, a lightweight architecture using small multi-scale depthwise-separable convolutional TSC encoders and improved Multiple Instance Learning (MIL) pooling heads. Our SEA-Net bottleneck encoder with Top-k pooling achieves 0.931 accuracy and 0.765 NDCG@n with only 41 K parameters. Another variant, using a gated encoder with Top-k pooling, achieves 0.955 accuracy and 0.750 NDCG@n with 62 K parameters. These results demonstrate that lightweight architectures can maintain strong classification performance and temporal explanations while substantially reducing model size. The proposed variants address specific limitations of the pooling strategy used by MILLET, providing a practical direction for interpretable time-series classification on resource-constrained edge devices.
[CV-99] MERIT: Multimodal Evidence Reasoning over Image Tables
链接: https://arxiv.org/abs/2610.17937
作者: Haoyang Jiang,Zhuotong Fan,Mingyang Yi,Ju Fan,Lanqing Hu,Xiao Zan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Table reasoning with language models typically assumes access to clean textual tables. In practice, tables are often serialized into text, which can corrupt cell contents, distort structural relationships, and discard visual cues, leaving the model to reason over incomplete evidence. In contrast, table images retain this information, but using it requires both accurate perception and reasoning grounded in the relevant cells. To address this, we propose MERIT (Multimodal Evidence Reasoning over Image Tables), a three-stage framework for learning to reason over table images, with serialized tables serving as optional auxiliary inputs (if provided). Our analysis identifies visual evidence acquisition as a bottleneck, motivating table-specific encoder adaptation through reconstruction, localization, and color-conditioned supervision. Building on the adapted representations, supervised fine-tuning teaches explicit evidence selection and reasoning using verified trajectories. Finally, reinforcement learning refines the policy with separate rewards for evidence correctness, calculation, final answers, and output format. By providing feedback on intermediate steps, the framework targets retrieval and computational errors that final-answer supervision alone does not distinguish. We evaluate the approach on table question answering and fact verification, analyze the contributions of visual inputs and training components, and assess generalization to additional datasets with different table structures and visual presentations.
[CV-100] PUGS: Physics-Guided Uncertainty Modeling for Underwater Dynamic 3D Gaussian Splatting
链接: https://arxiv.org/abs/2610.17936
作者: Tingjun Li,Yuxi Wang,Guoqiang Zhong,Zhaoxiang Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Dynamic 3D Gaussian Splatting (3DGS) provides an efficient framework for dynamic scene reconstruction. Most existing approaches model scene dynamics with deterministic deformation fields, assuming each Gaussian follows a distinct geometric trajectory over time. This assumption breaks down underwater, where light scattering, refraction, marine snow and low visibility create substantial physical ambiguity in captured images. We present PUGS, a physics-guided uncertainty model for underwater dynamic 3DGS. PUGS learns inherent uncertainty induced by underwater image formation and leverages it as a physical signal for dynamic scene modeling. Rather than estimating a single deterministic displacement, it parameterizes each Gaussian’s motion as a probability distribution with expected deformation and physical uncertainty. We adopt stochastic deformation optimization to sample Gaussian trajectories via local reparameterization, exploring multiple physically plausible solutions during training. We further propose uncertainty stochastic annealing and a dual-path self-consistency distillation objective, converting stochastic exploration to deterministic reconstruction while preserving temporal coherence. Extensive experiments on underwater dynamic datasets and physics-aware benchmarks demonstrate superior rendering quality, temporal consistency and reconstruction robustness. Our method achieves 1.16 dB PSNR gain over the strongest baseline on UVEB and 3.99 dB on DRUVA.
[CV-101] HierarchiRetina: A Diabetic Retinopathy Screening and Grading Cascade with Heterogeneous Mixture-of-Experts Lesion Segmentation ALT
链接: https://arxiv.org/abs/2610.17933
作者: Md Asif Uddin,Niloy Farhan,Rafeed Rahman,Md. Golam Rabiul Alam
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 8 figures, 8 tables. Submitted to the IEEE Journal of Biomedical and Health Informatics. Code, trained weights and per-image predictions: this https URL
Abstract:Clinical screening for diabetic retinopathy proceeds in stages, yet most automated systems grade fundus images with a single black-box network. Three gaps remain: (i) lesions differ widely in size and contrast, but segmenters typically apply the same blocks to all; (ii) ungradable images are usually excluded or treated as an extra class, and lesion-aware graders seldom test whether lesions drive the grade; (iii) staged systems often report each stage only on the images reaching it, leaving propagated errors unmeasured. HierarchiRetina, a screening, segmentation and grading cascade with inspectable stage outputs, addresses each gap. First, HSMoE-AUNet is a U-Net whose skip, bottleneck and decoder blocks are sparse mixtures of heterogeneous experts, with lesion-matched types, counts and losses; SwinHRUNetPP, a dual-encoder vessel segmenter, uses a field-of-view-restricted loss. Second, LG-DRG separates gradability from ordinal severity and lets lesion masks query the image through gated cross-attention. Third, an end-to-end evaluation counts screening misses as errors and traces each error to its stage. Pooled Dice ranged from 0.184 (microaneurysms) to 0.579 (exudates); vessel Dice was 0.731. On 58,689 held-out images from six datasets, end-to-end quadratic weighted kappa was 0.804, with referral sensitivity 0.832 at specificity 0.947, and the screening gate caused 66% of errors. In ablations, the grade ignored the masks, even with the cross-attention gate started open. On all 4,105 official DDR test images, six-class accuracy was 85.6%, above the best published six-class result we found (83.8%). Staged screening systems should be evaluated end-to-end, with stage attribution and mask ablation, before clinical deployment.
[CV-102] CineMem: Attention-Aligned Fine-grained Memory Retrieval for Multi-Shot Long Video Generation
链接: https://arxiv.org/abs/2610.17926
作者: Xiaoxue Wu,Xinyu Gao,Fang Yu,Yu Qiao,Xinyuan Chen
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Project Page: this https URL
Abstract:Autoregressive video generation relies on memory retrieval to model long-range dependencies. This capability becomes particularly critical for multi-shot generation, where models must reconcile substantial visual changes across shot transitions with persistent semantic continuity. Existing methods tend to perform memory retrieval at the frame level and share the same historical context across all query tokens, overlooking the heterogeneous information needs and leading to redundant retrieval and inefficient use of historical memory. To bridge this gap, we introduce CineMem, an attention-aligned fine-grained memory retrieval framework that combines online memory organization with query-adaptive retrieval for autoregressive multi-shot video generation. Specifically, guided by intrinsic memory-access patterns revealed through attention analysis, we propose an online-clustered KV cache as a memory organization strategy over all historical tokens, enabling semantic-aware memory grouping and more precise retrieval. In addition, we introduce a query-level retrieval mechanism that provides broader access to historical information under a fixed retrieval budget, thereby improving long-term memory utilization. Together, these designs enable the multi-shot video generator to preserve long-range consistency across shot transitions, particularly in revisitation scenarios, while approximating full-memory access under a constrained memory budget. To facilitate evaluation, we further introduce a suite of task-specific, VLM-assisted metrics for faithful inter-shot consistency assessment. Extensive experiments demonstrate the effectiveness of our framework.
[CV-103] LinMorph: Learning to Morph Weights for Efficient Video DiT Linearization
链接: https://arxiv.org/abs/2610.17914
作者: Lingyi Long,Congyue Zhao,Xiaoou Li,Yanmin Qian,Tao Huang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Post-hoc linearization is an attractive way to accelerate pretrained Video Diffusion Transformers (Video DiTs), yet existing methods largely treat the quality loss induced by attention replacement as a recovery problem. We show that a more fundamental bottleneck lies upstream: \emphinitialization. Random initialization discards pretrained knowledge, whereas manual weight reuse transfers it through rigid and incomplete source–target correspondences. We introduce \textbfLinMorph, which recasts Video DiT linearization as a learnable softmax-to-linear weight transformation problem. Instead of copying predefined counterparts, LinMorph learns a cross-parameter morphism that lets each target Gated DeltaNet (GDN) parameter draw jointly from pretrained query, key, value, and output projections, including for parameters with no direct softmax counterpart. A single morphism is shared across layers and learned through local teacher alignment, while producing layer-specific initializations from their respective pretrained weights. The resulting weights are then materialized as independent GDN modules for progressive distillation and joint this http URL the morphism operates in weight space rather than token space, it can be learned on short, low-resolution clips and directly reused at larger video scales. Across pretrained Video DiTs, LinMorph enables markedly faster recovery and better final generation quality than random initialization and heuristic weight reuse. Results show that efficient linearization depends not only on how a converted model is recovered, but critically on how pretrained knowledge is transferred before recovery. Project page: this https URL.
[CV-104] DriveWLA: Unifying Future Dynamics and Scene Semantics through World-Language-Action Modeling for Autonomous Driving
链接: https://arxiv.org/abs/2610.17903
作者: Shuzheng Wang,Haoxuan Li,Shengzhen Li,Chenxi Wu,Yajing Sun,Lu Hou,Wenyu Liu,Xinggang Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: 17 pages, 4 figures
Abstract:Vision-Language Models (VLMs) and world models have demonstrated strong capabilities in semantic understanding and future scene evolution for autonomous driving. However, effectively integrating both capabilities for trajectory generation remains challenging. In this work, we propose DriveWLA, a unified World-Language-Action model based on Mixture-of-Transformers (MoT) that jointly models semantic understanding, future scene dynamics, and trajectory generation. DriveWLA consists of three experts: a World Expert for future prediction, a Language Expert for semantic understanding, and an Action Expert for trajectory generation. We further introduce Cross-Expert Attention, which enables controlled information flow across experts while aligning semantic and dynamic representations. To improve trajectory generation beyond imitation learning, we apply Flow-GRPO reinforcement learning post-training to refine the Action Expert. Extensive experiments show that DriveWLA achieves superior performance in both open-loop and closed-loop evaluation. DriveWLA achieves 92.3 PDMS on NAVSIM v1 and 90.1 EPDMS on NAVSIM v2, demonstrating strong performance among VLM-based and world-model-based planners. Without specific fine-tuning, DriveWLA further demonstrates strong zero-shot generalization to HUGSIM, nuScenes, and Bench2Drive. These results validate the effectiveness of our unified World-Language-Action modeling, establishing DriveWLA as a generalizable framework for autonomous driving.
[CV-105] Visual Dependence Does Not Imply Answerability: A Systematic Analysis of Answerability Signals in Frozen InstructBLIP
链接: https://arxiv.org/abs/2610.17902
作者: Minjun Kim,Hyunmin Song,Changwon Lim
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 10 pages, 3 figures, 5 tables
Abstract:Visual question answering models often produce plausible responses even when the available visual evidence is insufficient to support a valid answer. Although prior work has introduced benchmarks and abstention mechanisms for unanswerable visual questions, it remains unclear which signals derived from a frozen vision-language model reliably indicate answerability on unseen image-question pairs. We systematically evaluate answerability signals on a stratified 10K subset of UNK-VQA using frozen InstructBLIP, covering predictive uncertainty, cross-modal alignment, visual grounding, visual dependence, internal representations, and learned probes. Answerability discrimination remains weak, with test AUROC largely confined to 0.50-0.54. CLIP alignment yields the highest point estimate but does not significantly outperform generation confidence under paired bootstrap analysis. Targeted removal of question-relevant visual evidence causes substantially larger changes in answer confidence than same-size random masking, yet intervention sensitivity provides essentially no answerability discrimination. Increasing probe capacity, combining multiple signals, and adding reason-aware supervision do not provide robust improvements, while post-freeze cross-validation yields similarly weak discrimination. Selective-answering analysis shows limited usable coverage under low-risk operating criteria. Failure-mode analysis shows that the relative utility of individual signals varies across perturbation mechanisms. Taken together, these results show that, on our UNK-VQA subset with frozen InstructBLIP, visual dependence, uncertainty, semantic alignment, and static representations are not robust standalone proxies for answerability. Our findings emphasize the need to distinguish a model’s reliance on visual evidence from whether that evidence is sufficient to support an answer.
[CV-106] Diagnose-to-Generate: Eliminating Structural Distortions in Flow Models via Patch-Level Grounded Alignment
链接: https://arxiv.org/abs/2610.17882
作者: Yifan Li,Lishuai Gao,Tong Zhang,Qilong Huang,Chen Zhao,Zhuoliang Kang,Jie Hu,Yujie Zhong,Fenghe Tang,S Kevin Zhou,Xiaoming Wei
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Flow-based image generators achieve strong semantic synthesis but still produce localized structural distortions, such as malformed hands and limbs. Existing post-training methods typically optimize global scalar rewards, which provide weak spatial credit assignment and may alter well-formed regions while correcting local defects. We present \textitDiagnose-to-Generate (D2G), a grounded alignment framework that separates visual diagnosis from generative optimization. We construct \textbfAchilles-600K, a region-annotated dataset for structural anomalies, and train \textbfARGUS through supervised cold-start and Region-Aware Group Relative Policy Optimization to predict anomaly categories and bounding boxes. These diagnoses are converted into confidence-weighted spatial rewards for our \textbfPatch-level Flow-GRPO. We further identify that global stochastic exploration can shift semantic regions across group rollouts, making patch-wise relative advantages spatially inconsistent. Our \textbfTrajectory-Preserving Stochastic Sampling (TPSS) anchors early denoising with deterministic ODE trajectories and introduces stochasticity only at later steps, thereby limiting intra-group spatial drift while retaining local exploration. Experiments on FLUX.1-dev and SD3.5 show improved structural fidelity with preserved FID and CLIP scores. Consistent gains under HumanRefiner, zero-shot GPT-4.1, and blinded human evaluation further demonstrate that the improvements extend beyond the training reward model.
[CV-107] LoopREG: Representation Supervision Enables Test-Time Recurrent Depth in Diffusion Transformers
链接: https://arxiv.org/abs/2610.17877
作者: Zipeng Wang,Anping Xiong,Jing Zhang,Xiaoya Zhu,Yi Jiang,Yanbing Xiao,Zige Zeng,Zezhou Chen,Xiang Liu,Huan Hu,Ping Chen,Kai Wang,Zhaoxiang Liu,Shiguo Lian,Ye LV,Fang Zhao,Huanlin Gao,Zhengrui Huang,Bo Shen,Yuren You
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 23 pages; including appendix
Abstract:Reusing Transformer blocks offers a way to increase diffusion-model computation at test time, but naive recurrence can degrade generation beyond its training depth. We introduce LoopREG, a representation-supervised recurrent diffusion Transformer that repeatedly applies a shared block at the state supervised by a pretrained visual encoder. Alignment is applied after the final recurrent visit during training; inference extends the same learned operator to additional visits. On ImageNet 256 x 256, LoopREG reduces SiT-B/2 FID from the local REG baseline’s 15.08 to 4.91 at 400K training steps, a 67.41% reduction. Within a checkpoint trained with two visits, one extra inference visit improves FID from 14.71 to 5.34 with unchanged weights. The benefit persists on SiT-L/2 and SiT-XL/2. Controlled interventions show that useful recurrence follows the location of representation supervision, while REPA alignment produces a smaller extrapolation benefit. These results make representation-supervised recurrence a practical route to improving diffusion generation through additional inference depth.
[CV-108] MicroCLIP: Connecting Text and Microscopy Images via Contrastive Language–Image Pretraining
链接: https://arxiv.org/abs/2610.17866
作者: Yanran Ruan,Zhixiang Wei,Shiyao Hong,Xiaoxiao Ma,Yuxuan Gu,Liming Ye,Huaian Chen,Yi Jin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Understanding fluorescence microscopy images is fundamental to automating biological research. However, existing methods typically learn predefined concepts from closed-set data, limiting transfer across tasks and biological contexts. In this work, we introduce MicroCLIP, a vision-language model that learns generalizable cellular and subcellular representations from natural language supervision for open-set image classification and retrieval. To pretrain MicroCLIP, we construct MicroPool-2M, a dataset of approximately 2.5 million fluorescence microscopy image-text pairs spanning diverse cellular structures, cell lines, and imaging conditions. Aligning microscopy images with biological concepts in natural language is challenging because most images lack descriptive captions and biological knowledge annotations. We therefore progressively establish image-text associations at three levels to provide MicroCLIP with richer semantic supervision: (1) category-level alignment links images to standardized labels defined by subcellular structure-cell-line combinations; (2) knowledge-level alignment incorporates biological priors derived from these labels to capture relationships across species and cell types; and (3) detail-level alignment uses a multimodal large language model to recaption and verify image-text pairs at scale, emphasizing fine-grained visual characteristics. Extensive experiments demonstrate that MicroCLIP achieves leading performance in zero-shot classification and text-image retrieval on microscopy images. It retrieves images with specific cellular characteristics from thousands of candidates and generalizes to unseen biological concepts without additional supervision. These results highlight MicroCLIP’s potential as a foundation model for fluorescence microscopy.
[CV-109] Frequency Routing and Identity Conditioning for Thermal-to-Visible Face Translation
链接: https://arxiv.org/abs/2610.17859
作者: Haojie Li,Xinlong Sun,Xiaohaoyang Lei,Kai Cheng,Yiguo Qiao,Xu Yang,Cunjian Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Thermal-to-visible (T2V) face translation has important applications in nighttime biometric recognition, low-light surveillance, and human interpretation of thermal observations. The task remains challenging because thermal images do not directly record visible appearance, while the cross-spectral cues useful for reconstruction vary across frequency components and semantic roles. Existing spatially conditioned models tend to process these cues through a homogeneous adaptation path and do not explicitly separate local structure from global identity. We propose FreqID-Control, a parameter-efficient framework that organizes thermal guidance according to spectral reliability and semantic role while preserving the pretrained generative prior. It combines frequency-routed low-rank adaptation, which allocates separate transformation subspaces to low-frequency structure and high-frequency detail, with shared cross-spectral identity conditioning, which supplies compact global identity tokens independently of spatial thermal tokens. A controlled analysis on MCXFace and SpeakingFaces provides the empirical basis for this design, revealing stable low-frequency correspondence and edge-localized detail correspondence. Experiments on both datasets demonstrate strong image fidelity and biometric identity preservation, while controlled ablations on MCXFace confirm the complementary contributions of frequency routing and shared identity conditioning.
[CV-110] DVTR-KD: Distilling Vision-Language Models via Intermediate Visual Token Reconstruction
链接: https://arxiv.org/abs/2610.17853
作者: Haolan Wei,Jiahui Wang,Junli Zha,Xinkai Lu,Yunxiao Zou,Chao Zhou,Xiang Song
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal large language models (MLLMs) have achieved strong performance on diverse vision-language tasks, but their computational demands limit deployment under constrained budgets, particularly on edge devices. Many works focus on Knowledge Distillation (KD) and try to transfer knowledge from powerful teacher models to students. In addition to reducing model size, visual token compression provides a complementary route to efficiency, but introduces mismatched visual token counts that complicate direct token-wise knowledge transfer. To mitigate this, we propose DVTR-KD, a knowledge distillation framework based on \textbfDense \textbfVisual \textbfToken \textbfReconstruction, which reconstructs the teacher’s dense visual tokens from the student’s compressed tokens. We reconstruct visual states after the final LLM layer and at the input to an intermediate layer identified as important for visual-information readout. At the intermediate layer, reconstruction is supervised through the teacher’s response predictions rather than by direct token-wise KL or MSE matching. An auxiliary attention-readout loss complements this response-level supervision by aligning how the student and the original teacher integrate contextual information when predicting responses. Evaluations on diverse benchmarks demonstrate that DVTR-KD improves the performance of token-compressed students while retaining the inference efficiency afforded by visual token compression.
[CV-111] What Makes a Good Visual Representation for Spatial Understanding in MLLM s? NEURIPS2025
链接: https://arxiv.org/abs/2610.17844
作者: Xiaohu Huang,Yumin Guo,Kai Han
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Extended version of Neurips 2025 paper: https://arxiv.org/abs/2506.01946
Abstract:Spatial understanding is a necessary step in moving multimodal large language models (MLLMs) toward reasoning about the 3D physical world. Their visual pretraining data, however, is almost entirely 2D, so the visual tokens processed by MLLMs often lack inherent 3D structural information. A growing number of methods inject such structure by transferring representations from external vision or geometry models, and collectively establish that such transfer helps. Behind these methods lies a largely underexplored question: what makes a visual representation good for spatial understanding? To answer it, we first separate what is transferred from how it is transferred, and then trace the resulting differences back to properties of the representations themselves. What we transfer comes from 3D geometry models, self-supervised image and video models, vision-language models, and generative models, in the form of latent features or explicit depth and point maps. We transfer it either by distillation, using it as a supervision target, or by fusion, feeding it as an additional input. We find that a representation’s 3D structure correlates strongly with how much it helps, measured by cross-view consistency for latent features and by depth and reconstruction accuracy for explicit geometry. We further show that the form of a representation dictates the mechanism it needs. Latent features gain as much from fusion as from distillation; explicit geometry gains little from fusion, and most of its distillation gain depends on a geometry decoding head. Guided by these findings, we distill both latent features and explicit geometry into a single MLLM. Taking only RGB video as input, it consistently improves over its base model on both static 3D and dynamic 4D spatial reasoning benchmarks.
[CV-112] QK-Edit: Revisiting Attention-based Injection in MM-DiT for Image and Video Editing
链接: https://arxiv.org/abs/2610.17829
作者: Tiancheng Shen,Zilong Huang,Xiangtai Li,Zhijie Lin,Jiyang Liu,Yitong Wang,Jiashi Feng,Ming-Hsuan Yang,Jun Hao Liew
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Multimodal Diffusion Transformers (MM-DiTs) have recently emerged as a powerful framework for unified text-vision synthesis, surpassing traditional U-Net architectures in generative tasks. One key innovation lies in its Multimodal Self-Attention (MM-SA) interaction, where image and text tokens are concatenated and processed via self-attention. However, this mechanism poses significant challenges for editing, rendering conventional U-Net-based attention manipulation methods ineffective. To address this limitation, we propose QK-Edit, a training-free framework that exploits the unique attention dynamics of MM-DiTs for precise text-guided image and video editing. By introducing a novel query-key manipulation strategy, our method isolates and adjusts critical attention components to achieve an optimal balance between prompt fidelity and structural consistency. This enables seamless edits across various tasks, including object addition, object removal, object replacement, changing background, changing material, changing color, and style transformation. Notably, it can be easily implemented with feature replacement in inference. QK-Edit demonstrates superior editing performance on state-of-the-art models, such as FLUX and HunyuanVideo, effectively bridging the gap between generative power and editable flexibility in MM-DiTs, and paving the way for scalable multimodal content creation.
[CV-113] SPARCL: Spatially Adaptive Regional Coverag e for Active Learning in Chest X-ray Analysis
链接: https://arxiv.org/abs/2610.17808
作者: Juho Jung,Seongjae Kang,Mijoo Kim,Sungduk Cho,Yunkyoung Jun,Sunghoon Joo,Yoona Hwang
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 10 pages, 7 figures
Abstract:Pixel-level annotation for chest X-ray (CXR) lesion segmentation is expensive, making active learning (AL) essential. However, existing AL strategies fail to account for the spatial distribution of selected samples, resulting in inconsistent performance across focal (e.g., aortic calcification) and diffuse (e.g., nodules) pathologies. In this paper, we propose SPARCL (Spatially Adaptive Regional Coverage for Active Learning) to dynamically balance predictive uncertainty with cumulative spatial coverage. By employing a concave coverage function, SPARCL imposes diminishing returns on previously annotated regions, naturally directing the annotation budget toward under-labeled areas. Furthermore, an entropy gate reverts the algorithm to standard uncertainty sampling when spatial variation is absent. This coverage objective is monotone submodular, guaranteeing near-optimal greedy batch selection. To address how aggregate metrics like Dice obscure regional failures, we further introduce two metrics—Spatial Inconsistency (SI) and Performance Coverage (PC)—to quantify anatomical fairness across AL rounds. Evaluated on an in-house calcification cohort and public CXR benchmarks, SPARCL reaches 95% of fully supervised performance using only 24% of the data, compared to 29% for the best baseline. Ultimately, SPARCL ensures robust spatial reliability, surfacing regional gains that aggregate metrics miss.
[CV-114] Beyond the End-Effector: Learning Whole-Arm Contact Manipulation from Human Demonstrations ACL
链接: https://arxiv.org/abs/2610.17802
作者: Bo-Ruei Huang,Daniel Seita,Erdem Bıyık
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Accepted to the 10th Conference on Robot Learning (CoRL 2026). 23 pages, 8 figures, 5 tables. Project page: this https URL
Abstract:Robot manipulation has traditionally relied on end-effector-centric interactions. In contrast, humans routinely exploit whole-arm contacts to improve efficiency, robustness, and adaptability, especially when the hands are occupied or when precise grasping is unnecessary. Enabling similar capabilities in robots remains challenging: the robot must select where on its arm to make contact, generate effective non-prehensile interactions, and satisfy task-specific pose or safety constraints on the end-effector. We propose Whole-Arm Contact Learning (WACL), a hierarchical framework that describes each whole-arm contact task with three robot-agnostic representations and hands them to a quasi-static surface-contact optimizer: semantic keypoints that summarize the scene, a reference contact trajectory that specifies where and how the arm presses on the object, and safety-aware path constraints that a vision-language model generates from the task description to protect the end-effector’s task. The robot’s kinematic chain enters only in this optimizer, which grounds these representations onto the arm by jointly searching over which link makes contact, where on that link, and the joint configuration that realizes it. This decomposition lets WACL learn the contact trajectory from 10 bare-hand human demonstrations per task collected without teleoperation, while new end-effector constraints change only the optimizer’s input rather than any trained component. Real-world experiments on a 7-DoF arm across four whole-arm tasks show that WACL sustains contact under end-effector constraints where a strong link-pose baseline fails (78% vs. 33% success) and generalizes to new constraints without retraining. Website: this https URL
[CV-115] Multimodal Fusion of RGB and Complementary Modalities for Semantic Segmentation
链接: https://arxiv.org/abs/2610.17800
作者: Hu Cao,Yifeng Cheng,Mu He,Mengyu Li,Yinlong Liu,Alois Knoll
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: This work is submitted to the IJCV journal
Abstract:Multi-modal semantic segmentation adds an auxiliary sensing stream to RGB images (RGB-X), such as thermal, LiDAR, depth, event, polarization, or light field. This can improve robustness in poor lighting and under motion blur. But different sensors often produce features that do not match well, and fusion can become unstable. To alleviate these issues, we propose a bidirectional polarity-aware cross-modality fusion (BPCF) module that effectively captures complementary cues while enhancing feature alignment. We evaluate the framework on six modality pairings: RGB-Thermal, RGB-LiDAR, RGB-Depth, RGB-Event, RGB-Polarization, and RGB-Light Field. Our model achieve state-of-the-art (SOTA) results on eight public datasets, including MFNet, KITTI-360, DELIVER, DDD17, DSEC, MCubeS, ZJU, and UrbanLF. The codes and trained models will be released upon acceptance.
[CV-116] MHVF: Training-Free GUI Grounding via Verification-Guided Multi-Hypothesis Search
链接: https://arxiv.org/abs/2610.17792
作者: Minghua Chen,Xingming Liao,Weiqiang Zhong,Zhuowei Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 11 pages, 8 figures, 5 tables
Abstract:GUI grounding enables web agents to translate natural-language instructions into actionable interface locations. However, dense layouts and small, visually similar controls can lead models toward plausible but incorrect elements, while refinement around a single prediction may leave alternative targets unexplored. We propose the Multi-Hypothesis Verification Framework (MHVF) for training-free GUI grounding. It combines multi-hypothesis exploration with local visual verification using a single pretrained vision-language model. To explore alternative targets, MHVF constructs spatial hypotheses from multiple full-screen predictions and examines the corresponding regions through zoomed re-localization. To resolve ambiguity among the resulting candidates, it verifies their agreement with the instruction in marked local views. Verification scores then guide a second round of local search and rank candidates retained from both search levels for final selection. Experiments with five general-purpose and GUI-specific models across eight benchmark settings demonstrate improvements in grounding accuracy in most settings. MHVF improves UI-Venus-Ground-7B by 4.3 and 2.1 percentage points on Mind2Web and VisualWebBench, respectively, and raises Qwen2.5-VL-7B’s accuracy on ScreenSpot-Pro from 23.6% to 53.6%.
[CV-117] When Where and Why: Hierarchical Causal Explainability for Video Face Understanding with MLLM s
链接: https://arxiv.org/abs/2610.17791
作者: Anubhav Kataria,Shreya Ghosh,Jingwen Ye,Abhinav Dhall
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Understanding facial behaviour in video requires answering three fundamental questions: when a key facial expression occurs, where the model attends on the face at that moment, and why that evidence is necessary rather than merely correlated. Today’s multimodal large language models (MLLMs) can label and describe video, but they rarely provide faithful explanations, especially for small facial movements that change over time. We present VideoFace-X-LLM, a framework for explainable facial video understanding comprising two key components. First, we introduce ExplainFace, a weakly supervised video-text corpus for facial behaviour understanding, automatically constructed from CelebV-HQ videos and validated through human evaluation and a two-model audit (at most 2.3% contradictions), demonstrating consistent gains across two backbone architectures. Second, H-TCC (Hierarchical Temporal Spatial Causal Concepts), which finds the key temporal moments (When), the facial regions (Where), and verifies its causal necessity (Why) through targeted occlusion. Fine-tuning on ExplainFace raises appearance Macro F1 from 19.48% to 32.87% (+13.39), with gains of +4.38 on actions, +6.77 on temporal emotion, and +11.84 F1 on deepfake detection. We also measure the explanations pointing lift, face/edge lift, stability, an Adebayo sanity check, and a necessity test rather than only showing examples. Our models, code, and corpus are available on request for research use.
[CV-118] DexStitch: Learning and Composing Dexterous Skills from Human Videos via Contact-Aware Robotic Inpainting
链接: https://arxiv.org/abs/2610.17789
作者: Byeonggyeol Choi,Seongmin Jung,Unghui Nam,Youngbo Song,Sunhong Min,MinHyung Lee,Byonghyo Shim,Yongseok Lee
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: 9 pages
Abstract:A long-horizon manipulation task consists of primitive skills and the transitions connecting them. Learning skills that can be executed in sequence requires supervision for both, which is costly to collect on a robot. Human videos offer a scalable source of demonstrations, but transferring them to robots requires extracting robot-compatible observations and actions together with their skill structure. We present DexStitch, a framework that cuts long-horizon human videos into skills and transitions, converts them into aligned robot observation-action-skill data, and stitches the learned skills into a requested sequence. A skill classifier first segments the videos into manipulation skills and transitions. DexStitch then converts the human demonstrations into robot observations and actions while retaining their corresponding skill labels. During conversion, per-finger human-object contacts guide dexterous hand retargeting and robotic inpainting, preserving contact geometry and finger-object occlusion ordering. To stitch the learned skills into a sequence, a shared skill-conditioned policy is pretrained on the resulting synthetic data and finetuned on a mixture of synthetic data and a small set of single-skill teleoperation episodes. Conditioning on the current and next skills guides each transition toward the following skill, while predicted skill progression enables closed-loop switching. On a held-out session, the classifier agrees with manual annotations on 92.4% of frames. Across four household skills, DexStitch achieves 80.0% single-skill success and 77.5% sequence success.
[CV-119] Autonomous 3D Visual-Thermal Inspection for Utility-Scale Solar Farms
链接: https://arxiv.org/abs/2610.17762
作者: Lachlan Tychsen-Smith,Mark Cox,Brett Wood,David Haddon,Kenrick Anderson,Benjamin Duck,Jim O’Malley,Noel W Duffy,Peyman Moghadam
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: Accepted at the 2026 Asia-Pacific Solar Research Conference (APSRC 2026), Canberra, Australia, 1-3 December 2026
Abstract:Utility-scale photovoltaic (PV) inspection is dominated by aerial surveys, which are effective for identifying module-level hot cells but provide limited visibility of critical components located under the array, including connectors, cabling, fasteners and structural elements. These assets are difficult to inspect consistently at scale, and their defects are often spatially small, partially occluded and difficult to localise from isolated images. We present an automated 3D visual-thermal inspection system (Tricorder) for under-panel PV asset inspection. The approach combines LiDAR-inertial Simultaneous Localization and Mapping (SLAM) with geometrically calibrated RGB and thermal imagery and machine-learning-based object detection and segmentation. Detected assets are associated with the reconstructed 3D environment, allowing visual and thermal observations from multiple viewpoints to be spatially fused and localised. We demonstrate the approach on utility-scale PV data for MC4 connector detection and connection-state recognition, thermal anomaly identification, and detection of structural components including bolts, nuts and torque tubes. The results demonstrate a pathway from image-based inspection toward spatially grounded autonomous inspection, where defects are detected and linked to specific assets and their 3D locations within the solar farm.
[CV-120] AttenIR: Aligning Dense Visual Attention with Mesh Warping for Unsupervised Image Retargeting
链接: https://arxiv.org/abs/2610.17743
作者: Yun Zhang,Zhiruo Zhang,Lifei Zhu,Zhe Zhu,Tong Li,Fang-Lue Zhang,Shengjuan Liu,Xinyuan Zheng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Image retargeting aims to resize images while preserving salient content free from distortion and stretching. Recent learning-based solutions reduce reliance on traditional hand-crafted features and optimization procedures, leading to more efficient and generalizable performance. Nevertheless, existing approaches still struggle to preserve global structure and content fidelity. To address these limitations, we propose AttenIR, an unsupervised image retargeting framework requiring no manual labels, based on content-aware mesh warping, which ef- fectively couples dense visual attention with sparse mesh constraints. We for- mulate the mesh deformation as a vertex displacement regression task. Guided by saliency priors, our method jointly optimizes scale, structure, and geometry preservation losses. By combining dense visual attention with semantic features, it achieves robust, unsupervised retargeting without manual annotations. Our primary contributions include (1) a lightweight unsupervised method for image retargeting that ensures content fidelity and preserves global structure, and (2) a coarse-to-fine projection strategy to align dense visual attention priors with sparse mesh constraints, enabling content-aware deformation at the pixel level. Experiments demonstrate that our method achieves competitive and superior performance compared with both traditional and state-of-the-art learning-based methods.
[CV-121] DT-LoRA: Reaching Beyond the Tangent Space of Low-Rank Adaptation
链接: https://arxiv.org/abs/2610.17730
作者: Yangming Zhang,Ling Yi,Datong Li,Jinliang Ding,Xiaojie Wang,Tiecheng Song,Zhaolong Ning
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 16 pages, 2 figures, 11 tables. Code: this https URL
Abstract:Low-rank adaptation (LoRA) often underperforms full fine-tuning. We investigate this performance gap through a tangent-space lens: factorized updates are confined to the tangent space of the current low-rank factors, leaving gradient components outside this space unexploited. An analysis of ten representative LoRA variants reveals that these residual components account for most of the gradient energy and exhibit strong spectral concentration. Motivated by this finding, we propose Dynamic Tangent LoRA (DT-LoRA), which partitions a fixed rank budget into a stable branch and a dynamic branch to capture the principal residual gradient directions. We show that these directions define the optimal update of a local quadratic model under orthogonal-complement and rank constraints. Matrix-free randomized SVD recovers them without materializing the residual gradient, while controlled write-back preserves accumulated updates and stabilizes training within the allocated rank. DT-LoRA adds no trainable parameters or inference cost over LoRA. Experiments across computer vision and natural language processing tasks demonstrate consistent improvements over LoRA and its variants. Moreover, despite being a general-purpose adaptation method, DT-LoRA outperforms task-specific approaches to open-vocabulary domain generalization for semantic segmentation, achieving state-of-the-art results. Code is available at this https URL.
[CV-122] From Token Correction to Claim Admission: Training-Free Hallucination Mitigation for Large Vision-Language Models
链接: https://arxiv.org/abs/2610.17706
作者: Hyeonggeun Han,Taehyun Cho,Sangwoo Hong
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Large vision-language models often exhibit object hallucination, generating claims that are not supported by the input image. To mitigate this issue, we propose Evidence-Guided Visual Claim Admission (EGVCA), a training-free framework that admits a semantic claim only when it receives sufficient image-dependent support. EGVCA measures this support by comparing the claim likelihood under the observed image and a visual-null input. For captioning, it controls semantic expansion without altering token generation. For binary visual question answering, it combines atomic visual support with the image-induced change in the model’s native Yes–No margin. Experiments on CHAIR, POPE, and MME across three LVLM backbones show that EGVCA substantially reduces object hallucination while preserving caption quality. It achieves the highest POPE F1 in eight of nine backbone–sampling settings and improves the native MME total score on all three backbones. These results demonstrate the effectiveness of claim-level admission as a simple alternative to token-level distribution correction.
[CV-123] A Differentiable 3D Scene Graph Metric via Contextual Hellinger Triplet Geometry
链接: https://arxiv.org/abs/2610.17705
作者: U-Chae Jun,Yoonhyuk Choi,Jiwoo Kang
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:3D scene graph prediction is commonly supervised with local object and predicate classification losses. Although effective for slot-wise label prediction, this formulation does not provide a graph-level semantic distance between a predicted scene graph and its ground truth: semantically mild label errors and structurally disruptive relational errors can be treated similarly, and subject-predicate-object facts are not compared as scene-level units. We propose Contextual Hellinger Triplet Geometry, a graph-level supervision objective that measures the semantic and structural discrepancy between 3D scene graphs. Our key idea is to reinterpret the conventional classifier outputs of a 3D scene graph predictor as an aligned probabilistic scene graph field, where object and predicate predictions jointly define contextual subject-predicate-object facts. This formulation yields a differentiable metric-based training objective with an efficient factorized implementation and can be applied to existing 3D scene graph prediction architectures without modifying their network design. Experiments on 3DSSG demonstrate that our supervision improves downstream graph-conditioned 3D scene generation, while ablation studies confirm the complementary roles of the proposed graph-level terms and user studies show better graph descriptiveness, relation plausibility, and structural preservation.
[CV-124] SWORD: Scene-Grounded Wasserstein Object-Relation Distance for 3D Scene Graph Prediction
链接: https://arxiv.org/abs/2610.17692
作者: U-Chae Jun,Jaeeun Ko,Yoonhyuk Choi,Jiwoo Kang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:3D scene graph prediction is commonly optimized with object-node and relation-edge classification losses. These objectives encourage exact object and relation label recovery, but they do not directly optimize semantic proximity to the ground-truth scene graph: two wrong graphs can receive comparable penalties even when one is much closer in object roles and relation contexts. A direct graph-distance loss is difficult in general, since graph matching and graph edit distance require correspondence search and edit-cost design. Our key observation is that prior supervised 3D scene graph prediction protocols are conducted on fixed object-node indices and supervised edge slots. This can turn a scene graph comparison into a well-defined aligned graph-distance problem. To this end, we propose SWORD, a Scene-Grounded Wasserstein Object-Relation Distance that builds block-normalized object and relation ground distance metrics from training scene graph annotations through a deterministic base-to-context order, without external semantic priors. SWORD lifts these metrics to Wasserstein distances over aligned node and null-augmented edge marginals, providing semantic graph-distance supervision beyond exact-label penalties while preserving the ground-truth graph as the unique zero-distance solution in aligned marginal probability space. We prove that SWORD is a strict metric on aligned marginal probabilistic scene graphs, preserves scene-graph-grounded substitution ordering, controls additive graph-grounded semantic observables, and yields a surrogate upper-bounding decoded semantic error.
[CV-125] MedEvo: Medical Reasoning Evolution through Knowledge Navigation
链接: https://arxiv.org/abs/2610.17690
作者: Xiaoliang Wang,Zhuo Chen,Chuanyang Zheng,Quanbin Wang,Jiangwei Lao,Yingao Liu,Jiajia Liu,Jinming Cao,Honglin Lin,Zhihong Lu,Hao Jin,Jian Wang,Lijun Wu
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Quantitative Methods (q-bio.QM)
备注: Project: this https URL GitHub: this https URL
Abstract:Medical multimodal reasoning models are typically trained on teacher-generated rationales, discarding the valid reasoning in students’ own failed trajectories, while general-domain correction methods offer little guidance for errors in image reading, anatomical context, or differential diagnosis. We propose MedEvo, which uses medical knowledge, namely a tag taxonomy and error-type diagnosis, to navigate both trajectory repair and data selection. For each failed rollout, a gold-blind auditor locates the first consequential error and provides a minimal, answer-free hint; the same student retains the valid prefix and regenerates the erroneous suffix. After each round, MedEvo identifies capability gaps across medical tags on a held-out feedback set and prioritizes related questions for new rollouts and repairs. With supervised fine-tuning (SFT) alone, MedEvo raises Qwen3.5-9B mean accuracy across seven benchmarks from 65.57% to 69.12% (47.40% to 54.50% on the multimodal track of MedXpertQA (MedXpertQA-MM)), surpassing teacher-distillation SFT (66.93%). It also raises Qwen3.5-35B and Qwen3.5-397B mean accuracy from 67.91% to 73.04% and from 71.83% to 74.95%, respectively. Our analyses show that failed trajectories contain reasoning worth preserving, that repair success does not imply learning value, and that independent feedback is key to sustained improvement.
[CV-126] Hierarchical Curriculum Learning for Surgical Video Generation: Focusing on Bilateral Axillo-Breast Approach Robotic Thyroidectomy Datasets
链接: https://arxiv.org/abs/2610.17684
作者: Youngbin Kong,Seungho Lee,Jae Kyung Shin,Jong-hyuk Ahn,Hyoun-Joong Kong,Kyu Eun Lee
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:High-fidelity surgical video generation is increasingly important for surgical education, simulation, and downstream AI training, as conventional simulation pipelines are limited by cost, realism, and scenario diversity. Adapting generic text-to-video diffusion models to surgical procedures remains difficult, however, due to complex anatomical structures, intricate tool-tissue interactions, the inherently hierarchical nature of surgical workflows, and the scarcity of curated training data outside a few well-studied procedures. We propose a surgical video generation framework that combines hierarchical procedural annotation with curriculum transfer learning. Surgical procedures are decomposed into three levels (phase, task, and action) and a pretrained text-to-video diffusion model is progressively adapted from coarse to fine-grained surgical context, supported by a dynamic sampling strategy that handles variable-resolution and variable-length surgical clips. We instantiate the framework on Bilateral Axillo-Breast Approach robotic thyroidectomy, a procedure for which no curated benchmark currently exists. Relative to the preceding task-level model, the final action-level model reduces FVD by 20.1% and FID by 15.7%, and a matched-budget ablation indicates that these gains arise from the curriculum design rather than from additional training. A user study with six clinicians rates visual realism highly while highlighting remaining gaps in anatomical accuracy and clinical reliability. The proposed framework offers a step toward AI-driven surgical simulation in procedures where curated data are scarce.
[CV-127] Integrating 3D Posture Neutralization with Statistical Shape Modeling for Head-Neck-Shoulder Ergonomic Design
链接: https://arxiv.org/abs/2610.17671
作者: Jinbo Luo,Hang Ren,Yadie Yang,Jie Zhang
类目: Graphics (cs.GR); Computer Vision and Pattern Recognition (cs.CV)
备注: 16 pages, 15 figures, 1 table
Abstract:The design of ergonomic products and workspaces requires a deep understanding of human anatomy and posture. In particular, the head-neck-shoulder region is critical for many applications such as cervical orthoses, helmets, and other head-worn or neck-worn devices. This paper presents a framework that integrates posture neutralization with statistical shape modeling (SSM) for the head-neck-shoulder (HNS) region. We acquire a single three-dimensional (3D) scan of the head-neck-shoulder region from each of 200 Chinese adults in a natural posture for each subject. Every scan is then posture-neutralized before SSM. We fit a parametric head-neck model to each scan to separate cervical pose from anatomy, and reset the cervical pose to a common neutral configuration, so that the resulting model captures anatomical shape variation rather than posture-induced deformation. The resulting shape space yields a standardized template and population-wide modes of variation, providing baselines for parametric ergonomic design. We further discuss how the model supports population coverage analysis and the design of products that accommodate anatomical diversity while maintaining neutral cervical posture.
[CV-128] ScenePercept: Scenario-Conditioned Perceptual Assessment of Real-World User-Generated Content Images
链接: https://arxiv.org/abs/2610.17661
作者: Shuhan Xia,Yu Chen,Xiang Wang,Shuai Li,Peipie Li,Zekun Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The rapid development of mobile imaging and social platforms has led to explosive growth in user-generated content images, making their systematic and comprehensive perceptual assessment increasingly important. Existing image perception benchmarks are typically task-oriented, curating specific scenarios for quality or aesthetic assessment and applying a uniform perceptual criterion across diverse scenarios. This limits their ability to capture the scene diversity and perceptual complexity of real-world UGC images. To address these limitations, we introduce ScenePercept-Bench, a benchmark designed to reflect the distribution of real-world UGC images, comprising 23,100 images across 21 categories and 105 fine-grained scenes. To account for the influence of scene context on human perception, we design a hierarchical, scene-aware annotation protocol, covering visual quality, aesthetic perception, and scene semantics, with 15 attributes in total. Each image is annotated with an overall score, 15 attribute-level scores, and structured explanatory reasoning. Building on this benchmark, we further introduce Scenario-Conditioned Perceptual Alignment, a training framework that tackles the heterogeneity of perceptual criteria across scenarios. SCPA jointly optimizes absolute score prediction, preference ranking, and cross-scenario calibration, preserving the model general perceptual assessment ability while encouraging it to explore and adapt perceptual criteria across different scenarios. Extensive experiments demonstrate that SCPA achieves state-of-the-art performance in UGC image perception and generalizes effectively to established IQA and IAA benchmarks.
[CV-129] Open-o3-Video-v2: Seek Ground and Reason with Verifiable Spatio-Temporal Evidence
链接: https://arxiv.org/abs/2610.17656
作者: Jiahao Meng,Zhongyuan Bao,Yue Tan,Haiquan Wen,Renrui Zhang,Anna Wang,Jason Li,Yanwei Fu,Yunhai Tong,Ming-Hsuan Yang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Extension of Open-o3-Video ( https://arxiv.org/abs/2510.20579 ). V2 extends single-pass grounded reasoning to agentic multi-turn evidence seeking and grounding, enabling models to actively acquire and reason over missing visual evidence. We further introduce STGR-v2-CoT-135K and CS-GRPO to improve evidence acquisition and grounded reasoning
Abstract:Reliable video reasoning requires models to produce correct answers and expose supporting visual evidence. Open-o3-Video enables verifiable reasoning by embedding timestamps and object-level bounding boxes into reasoning traces, but is restricted to frames from an initial sparse sampler. We introduce Open-o3-Video-v2, which formulates video reasoning as a unified agentic evidencegrounded process. The model identifies missing information, acquires additional observations, grounds the moments and objects supporting its reasoning, and writes this evidence into the trajectory before continuing or answering. To support this behavior, we construct STGR-v2-CoT135K, including 52K multi-turn agentic grounded reasoning trajectories, and STGR-v2-RL-13K for reinforcement learning. We further propose Clue-Seek GRPO (CS-GRPO), which tracks how tool calls change accumulated evidence, rewards clue coverage and step-wise gains, and penalizes excessive, repetitive, or uninformative retrieval. On V-STaR, Open-o3-Video-v2-4B achieves 40.4% mAM and 56.3% mLGM, outperforming its Qwen3-VL-4B backbone by 8.5% and 12.6%, respectively. It also delivers consistent gains across general video understanding, long-video reasoning, and temporal grounding benchmarks. These results show that active evidence acquisition and verifiable object-level grounding are complementary for reliable video reasoning. The code and models are released at this https URL.
[CV-130] Code2Worlds-2: Physically Plausible Agent Harness for Embodied World Simulation
链接: https://arxiv.org/abs/2610.17651
作者: Haoyu Zhang,Siming Li,Zeyu Zhang,Yi Zhang,Tat-Seng Chua,Hao Tang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: Code: this https URL , Website: this https URL
Abstract:Recent coding large language models (LLMs) can synthesize dynamic 4D worlds from natural language, but the resulting content remains a visual artifact. It lacks physical semantics such as the collision geometry, mass, friction, inertia, and joint structure, required to enable it to simulate the physical world, and its dynamics are scripted to look plausible rather than designed to execute. What is needed is an executable world: one that a physics engine can load and step, resolving contact physically. We present Code2Worlds-2, an extension of Code2Worlds that turns 4D world generation into an embodied world simulation. We keep the original dual-stream generation pipeline intact and append two modules. First, a SimReady Scene Compiler that turns a generated visual scene into a simulator-compatible representation: it assigns collision meshes via convex decomposition, estimates physical attributes with a vision-language model, infers a lightweight support graph, instantiates the engine’s contact, friction, and dynamics constraints, and applies a gravity settling pass, before exporting the scene as MuJoCo MJCF. Second, a Simulator-in-the-Loop 4D Refiner that implements an automated agent harness-a compile, simulate, critique and refine loop-that converts visual dynamics into executable rigid and articulated dynamics and evolves the physical settings rather than only the code, replacing the purely visual motion critic with a simulator critic that inspects penetration depth, collision pairs, stability, contact forces, joint limits, and robot reachability. We further propose SimCode4D, a state-based benchmark with five task templates for evaluating embodied simulation. Experiments on SimCode4D and across natural, indoor, industrial, urban, and service scenes show that Code2Worlds-2 produces physically plausible, simulator-compatible, and robot-interactable scenes.
[CV-131] MemAct: Streaming Spatial Understanding with Memory and Active Exploration
链接: https://arxiv.org/abs/2610.17640
作者: Yuxi Wei,Wei Huang,Qirui Chen,Lu Hou,Xiaojuan Qi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Spatial understanding is essential for embodied agents navigating and interacting with the physical world. As observations arrive incrementally, agents must reason in a streaming manner. Yet streaming understanding must also be closed-loop: task-relevant evidence may require both recalling past observations and actively acquiring new views. Existing streaming approaches largely emphasize processing incoming observations, leaving task-driven active exploration underexplored. We introduce MemAct, a framework for streaming spatial understanding with memory and action. A shared vision–language model incrementally writes structured historical memory, retrieves records and past frames on demand, and acquires additional evidence through movement and turning. Retrieved evidence and action observations inform subsequent decisions, enabling the agent to continue exploring or answer within a unified reasoning loop. We develop the MemAct suite, a data generation pipeline and collection across real and simulated environments, with approximately 1M training examples for spatial reasoning, memory writing, retrieval, and exploration. Its MemAct-bench comprises 20,000 main QA pairs evaluating current-observation understanding, historical evidence use, and active exploration, plus 2,000 separate Timing questions. MemAct combines supervised fine-tuning for memory and interaction with counterfactual action optimization using cumulative evidence gains, action costs, and terminal answer quality to improve exploration and stopping decisions. Experiments show that MemAct outperforms existing spatial understanding and streaming baselines and approaches frontier multimodal models on MemAct-bench while retaining streaming inference efficiency.
[CV-132] FrameLite: Decision-Oriented Aesthetic Cropping with a Compact Vision-Language Model
链接: https://arxiv.org/abs/2610.17639
作者: Jianze Li,Kai Liu,Ziqing Zhang,Weihang Zhang,Yong Guo,Linghe Kong,Yulun Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Aesthetic image cropping improves composition by balancing subjects, context, and image boundaries. Large vision-language models bring semantic knowledge to this task, but their scale limits lightweight photography assistance. We propose FrameLite, a compact cropping model built on Qwen3.5-0.8B. Our approach is motivated by the observation that a small model can generate better crops than its first answer suggests. We develop three-stage training to turn this potential into stronger direct predictions. Spatial and composition training establishes localization and framing preferences. Quality-Guided Online Supervision (QOS) then learns the highest-quality complete answers among the current policy’s generated candidates. Greedy-Referenced Optimization (GRO) further refines the policy using positive and negative quality differences relative to its current greedy answer. Both online stages use annotation-derived feedback, while inference returns one coordinate answer without search or an external scorer. Experiments show that FrameLite outperforms the compared large-VLM croppers across public benchmarks with a substantially smaller backbone. Controlled studies confirm that selecting high-quality training answers improves the final model beyond online sampling alone. We will release our code and model weights at this https URL .
[CV-133] Calibrated Disambiguation for Partial Multi-label Learning AAAI2025
链接: https://arxiv.org/abs/2610.17635
作者: Zhuoming Li,Yuheng Jia,Mi Yu,Zicong Miao
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: The paper is accepted by AAAI 2025
Abstract:Partial multi-label learning (PML) aims to train a classifier on datasets whose instances are over-annotated with not only relevant labels but also irrelevant labels, which is common when datasets are collected from crowd-sourcing platform. Existing works primarily approach it from a curriculum learning perspective, leveraging the memorization effect to disambiguate noisy labels and produce robust predictions. However, these methods are based on non-adaptive weighting functions and lack theoretical guidance for optimal weighting. To overcome these issues, a calibrated disambiguation model named PML-CD is proposed. We first formulate the optimal weighting function for curriculum-based disambiguation, which is equivalent to the calibration of the model’s predicted confidences, thus providing guidance for curriculum designing. To obtain the optimal weighting function from PML dataset during the training, a transferable calibrator is designed, which takes the histogram of positive samples’ confidences as input, and outputs the optimal curriculum weighting for training. Prototype alignment regularization is also proposed to promote the model’s performance. Experiments conducted on Pascal VOC, MS-COCO, NUS-WIDE and CUB have verified that our method outperforms existing state-of-the-art PML methods.
[CV-134] From Pixels to Executable Physics: Behavior-Grounded World Models for Embodied Action Consequence Prediction
链接: https://arxiv.org/abs/2610.17630
作者: Ruo Chen,Jiahua Pang,Yingjie Niu,Hongyi Lin,Jun Tian,Yang Zhu,Chao Li
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:In outdoor environments, varying terrain and contact conditions can cause identical action commands to produce different motion outcomes for quadruped robots. Predicting these outcomes is essential for reliable navigation and informed path selection. Existing video world models can generate visually plausible futures, yet often struggle to capture robot dynamics and terrain interactions, leading to overly optimistic predictions of action outcomes. We present DR-WM, a physically consistent world model that incorporates robot-terrain contact dynamics to predict the consequences of quadruped actions. Starting from an initial RGB image, our method reconstructs scene geometry and integrates dense, dimensionless priors over terrain contact properties, calibrated using real-world action-response data. The resulting terrain model supports the simulation of physics with spatially varying contact properties such as friction. Given the initial state of the robot and candidate action sequences, we simulate its future motion and use the resulting camera trajectories and rendered geometry to guide the training of a video world model. The generated future observations are then used to update the terrain reconstruction, enabling iterative model rollouts without additional real-world images. Extensive experiments demonstrate that DR-WM substantially outperforms geometry-only baselines and baselines with uniform physical parameters on key physical metrics characterizing action outcomes, while achieving the best performance among the evaluated baselines. These physically grounded predictions support more informed action selection: across 90 challenging scenarios, an 8B vision-language model augmented with DR-WM outperforms an unassisted 32B vision-language model on five of six evaluation metrics.
[CV-135] 4LC: 4D Latent Cache for Efficient Conditioning in Dynamic World Modeling
链接: https://arxiv.org/abs/2610.17625
作者: Jianhui Zhang,Zuyao Lin,Pengxin Xu,Jingdong Wang,Xingyu Chen
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Project page: this https URL
Abstract:For video generation in dynamic world modeling, observed static and dynamic content should be efficiently reused to provide conditions for repeated generation at different times and viewpoints. We introduce 4LC, a 4D latent cache with composite visual representations and a flexible spatiotemporal access interface. As a latent cache, 4LC directly reuses visual representations of static and dynamic content to provide generation conditions, avoiding per-request RGB rendering and re-encoding. As a composite cache, 4LC binds visual representations from multiple visual encoders (e.g., VAE, DINO, V-JEPA, VGGT) to shared geometry and motion, allowing the composite representations to describe the dynamic world with rich information. Besides, we design a flexible spatiotemporal interface, which directly reads the cache at independently specified times and viewpoints and combines readouts across observations, providing a common conditioning format for diverse downstream tasks. Building on 4LC, we develop 4LC-Video to address multiple video generation tasks through different configurations of source observations, spatiotemporal requests, and historical reuse, including camera control, temporal control, novel-view synthesis, scene revisits, and long-term generation. Our experiments show more efficient condition acquisition, effective spatiotemporal controllability, and scene consistency across tasks using a single 4LC-Video model without task-specific fine-tuning.
[CV-136] Smartphone-Based Field-Deployed AI Inspection System for Power Distribution Infrastructure
链接: https://arxiv.org/abs/2610.17620
作者: Sampath Balaji Masabathula(1 and 2),Srija Vanapalli(1 and 2),Gangadhar Kanchi(1),Sreenivasa Rao V.A.N(1),Vijaya Lalitha V(1),Sasank Chilamkurthy(3) ((1) Eastern Power Distribution Company of Andhra Pradesh Limited (APEPDCL), Visakhapatnam, India, (2) Andhra University College of Engineering, Visakhapatnam, India, (3) Von Neumann AI (JOHNAIC), Bengaluru, India)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Routine inspection of power distribution infrastructure remains challenging for utilities operating geographically dispersed networks, where conventional manual inspection can be difficult to perform consistently and at scale. This work presents a smartphone-based AI inspection system for assessing targeted distribution-infrastructure conditions from images captured by field personnel. The proposed workflow integrates image-quality and viewing-perspective validation with AI-assisted assessment of pole lean, crossarm and top-cleat tilt, voltage level, vegetation-clearance risk, and distribution-transformer (DTR)-ground clearance. The system integrates deep-learning models with monocular depth estimation, geometric analysis, and rule-based processing within a common multi-stage inspection workflow designed for practical field operation. The system was pilot deployed across five selected distribution feeders within the Eastern Power Distribution Company of Andhra Pradesh Ltd. (APEPDCL), where 1,222 defect instances were identified across the assessed categories and subsequently tracked through a rectification and verification workflow, with 84 defect instances recorded as rectified during the study period. Feeder-level interruption records were also examined over the study period; however, the available observations do not provide sufficient evidence to establish a reliability improvement attributable to the proposed system. A curated multi-module field-captured dataset, PD-Defect, is also publicly released to support reproducible research in distribution-infrastructure inspection. The field deployment demonstrates the practical integration of smartphone-based AI-assisted inspection into field inspection and maintenance activities, providing a mechanism for systematic identification, documentation, rectification tracking, and verification of targeted distribution-infrastructure defects.
[CV-137] PLEDGE: Progressive Lossy Error-Bounded Compression of 3D Gaussian Splatting
链接: https://arxiv.org/abs/2610.17605
作者: Mingze Xia,Yuxiao Li,Baixi Sun,Wenbo Li,Pu Jiao,Sheng Di,Franck Cappello,Hanqi Guo,Xin Liang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Existing 3D Gaussian Splatting (3DGS) codecs report reconstruction quality after compression, and give users little control over the distortion of individual primitives. We present PLEDGE, a training-free, progressive, error-bounded codec that translates four user-declared tolerances into per-primitive quantization bounds: screen-space position, footprint deformation due to rotation and scale, view-dependent color, and opacity. The bounds apply throughout a declared camera domain, including previously unseen viewpoints. Position and opacity are bounded directly; footprint deformation is bounded under the rasterizer’s affine projection model, and color is bounded over all viewing directions. PLEDGE organizes quantized residuals into a progressive bitstream with explicit error bounds for every valid prefix. Experiments across thirteen scenes verify all declared per-mechanism bounds and demonstrate competitive rate-distortion performance. The four tolerances incur substantially different rate costs, so users can spend rate where their application needs it.
[CV-138] Morphology-aware deep learning predicts colorectal cancer survival from whole-slide images
链接: https://arxiv.org/abs/2610.17603
作者: Hikmat Khan,Metin N. Gurcan,Anil Parwani,Wendy L. Frankel,Wei Chen,Muhammad Khalid Khan Niazi
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Supplementary material included: one figure and two tables
Abstract:Background: Five-year survival prediction from hematoxylin-and-eosin whole-slide images could support risk stratification, but the task is weakly supervised, and the tissue morphologies carrying prognostic information remain incompletely characterized. Methods: We conducted a retrospective secondary analysis using TCGA-COAD/READ (n=562) for morphology-specific analysis, Alliance/ CALGB 89803 (n=424) for model development, and SurGen (n=218) for external evaluation. We propose MorphContrast, a morphology-aware contrastive multiple-instance learning framework that uses tissue labels as structural supervision for five-year survival prediction. MorphContrast combines gated-attention pooling with three auxiliary objectives: patch-level morphology labels define within-patient contrasts, outcome concordance or discordance defines cross-patient contrasts within matched morphologies, and diversity regularization discourages representational collapse. Morphology-matched patches from patients with discordant outcomes serve as hard negatives, encouraging prognostic discrimination within comparable histologic contexts. Results: Morphology-specific models show weak but distributed prognostic discrimination: adipose tissue achieves an area under the receiver operating characteristic curve (AUC) of 0.622, stroma achieves 0.604, and both perform comparably to tumor epithelium (0.601-0.605). On Alliance/CALGB 89803, the single model achieves an AUC of 0.642, while the post hoc ensemble reaches a pooled out-of-fold AUC of 0.70 and a risk-group hazard ratio of 3.79 (95% confidence interval: 2.55-5.63). On the external SurGen colon cancer cohort (n=218), the single model and ensemble achieve AUCs of 0.669 and 0.678, respectively. Conclusions: These findings support tissue morphology as structural supervision for outcome-aware representation learning in colorectal cancer prognosis. Comments: Supplementary material included: one figure and two tables Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) Cite as: arXiv:2610.17603 [cs.CV] (or arXiv:2610.17603v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2610.17603 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Hikmat Khan Ph.D [view email] [v1] Tue, 13 Oct 2026 02:22:42 UTC (4,643 KB)
[CV-139] Physical-Field-Bridged Material Parameter Inference from Multi-View Deformation Videos
链接: https://arxiv.org/abs/2610.17591
作者: Linrui Shou,Jie Li,Yu Liu,Tao Huang
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注:
Abstract:Estimating simulation-ready material parameters from visual observations is crucial for deformable-object manipulation but challenging because material properties are only indirectly revealed through physical responses. We introduce physical-field-bridged material inference, a single-pass framework that uses dense physical responses to connect multi-view RGB deformation videos with constitutive-type classification and continuous parameter regression. Unlike inverse-physics methods requiring per-scene reconstruction, online tuning, or simulation-in-the-loop optimization, our model predicts force/contact masks, projected motion-flow fields, and projected stress-response maps, then conditions material inference on these fields to reduce reliance on appearance or category shortcuts. We construct DeformFieldBench, a 3DGS MPM learning environment comprising 5,000 simulations across 100 object shapes and five mechanically informative loading modes, with dense pixel-aligned physical supervision. Our method achieves the best median errors on all four continuous parameters, improves stress-hotspot recall from 0.0762 to 0.5121, and attains 0.8676 SSIM/36.82 dB PSNR in same-simulator replay on unseen objects. Under a protocol-matched GSO evaluation, inference and future-state prediction take approximately 30 seconds, compared with 15-120 minutes for per-scene optimization. The method also yields measurable replay agreement on five real-world sequences. These results support dense physical fields as an effective bridge for fast material inference, while cross-simulator and real-world tests reveal remaining domain and identifiability gaps.
[CV-140] Benchmarking Self-Supervised 3D Medical Vision Foundation Models and Transferability Estimation
链接: https://arxiv.org/abs/2610.17573
作者: Jiaqi Tang,Yu Wang,Yang Liu,Jing Tao,Fandong Zhang,Zhou Zhen,Xiahai Zhuang,Qingchao Chen,Yining Yang,Yizhou Yu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Self-supervised pretraining has expanded the pool of 3D medical foundation models, but published evaluations use different datasets, endpoints and fine-tuning protocols. Models also differ in architecture, pretraining objective and data, while downstream targets range from voxel-level masks to patient-level diagnoses and outcomes. As a result, published findings provide little directly comparable evidence for choosing a model for a new task. Direct comparison of adaptation performance requires fine-tuning candidates on labelled target data, at substantial computational cost. We benchmark 23 candidates across the same 14 segmentation, classification and regression tasks spanning multiple anatomies and modalities. No candidate ranked among the top three across all segmentation tasks or across all classification and regression tasks; a convolutional encoder could outperform a transformer trained with the same objective on one task but underperform it on another. We benchmark transferability estimators against the fine-tuned rankings and find that their agreement varies across dense segmentation, classification and regression. InsightRank uses task-appropriate frozen-feature statistics and architecture-aware pairwise comparisons to rank candidates before fine-tuning. Across the 23-model pool, mean Kendall tau-b was 0.431 for segmentation and 0.352 for classification and regression. The segmentation Top 3 contained a task winner on four of nine tasks; the image-level Top 3 contained none on the five tasks, despite better full-ranking agreement than the matched conventional estimators. A three-model shortlist reduced recorded extraction-plus-training job time by 85.8% across seven segmentation tasks. Together, the benchmark provides a common reference for model selection and InsightRank offers a lower-cost way to prioritize fine-tuning, with shortlist performance that differs by target type.
[CV-141] FlashVSR: Efficient Video Super-Resolution via Hybrid Attention and Reward-Yield Distribution Matching
链接: https://arxiv.org/abs/2610.17552
作者: Ruchang Yao,Shi Guo,Junhao Zhuang,Mingde Yao,Xin Cai,Jiahao Zhan,Bin Fu,Xinyuan Chen,Yu Qiao,Tianfan Xue
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Recent video super-resolution (VSR) models achieve efficient one-step generation through diffusion distillation. However, high-resolution inference remains costly, and distilled models are difficult to align with human preferences. These challenges call for improvements in both model architecture and training strategy. We present \textbfFlashVSR+ to address both limitations. On the architecture side, we introduce hybrid attention based on the strong locality observed in pretrained VSR models, while retaining sparse global interaction for long-range consistency. It achieves up to 2\times DiT speedup and 45% lower peak memory, enabling spatially untiled 4K VSR on a single 24GB RTX 4090. On the training side, we propose Reward-Yield Distribution Matching Distillation (Ry-DMD) for preference-aligned one-step generation. Ry-DMD uses reward-weighted teacher rollouts to incorporate black-box rewards into distribution matching, without backpropagating through the reward model. We further show that Ry-DMD generalizes beyond VSR to T2I generation. Together, FlashVSR+ delivers stronger restoration quality and substantially improves high-resolution efficiency, with consistent gains in user preference. We release our code, models, and training data at \hrefthis https URL\textcolorFlashBlue\textbfthis http URL.
[CV-142] CrossPortrait: Cross-Species Portrait Animation via Adversarial Motion Alignment
链接: https://arxiv.org/abs/2610.17532
作者: Gezhong Pan,Zhe Li,Yunfan Tao,Yongheng Zhao,Shiqi Zhou,Juyong Zhang,Xiaosong Jia,Yanyong Zhang,Shuquan Ye,Shaofeng Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Cross-species portrait animation, which aims to animate a source portrait of species A using a driving video of species B, is particularly challenging due to substantial anatomical differences in facial structures across species. Previous works typically fine-tune human portrait animation models on animal video datasets but often suffer from driving-species leakage. To address this, we propose CrossPortrait, a novel framework for expressive cross-species portrait animation, supporting diverse scenarios including human-to-animal, animal-to-human and cross-species animal-to-animal (e.g. cat-to-dog) animation. Specifically, we decompose the portrait motion into rigid head poses and expression keypoints. To accommodate species-specific facial structures while facilitating cross-species expression alignment, we introduce a species-routed expression encoder that incorporates species-specific LoRAs into a unified expression encoder. Furthermore, we introduce expression-level adversarial domain adaptation (Ex-ADA) to suppress driving-species-specific information in the expression latents, thereby mitigating species leakage and promoting a shared expression space across species. Extensive experiments show that our method enables expressive and identity-preserving portrait animation across diverse species and outperforms other state-of-the-art approaches.
[CV-143] winPrune: Density-Aware Two-Phase Token Pruning for Vision-Language Models
链接: https://arxiv.org/abs/2610.17530
作者: Kai Liu,Anqi Li,Junxian Li,Zhixin Wang,Zhikai Chen,Renjing Pei,Yulun Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Our code is availabel at this https URL
Abstract:We propose a unified spatial-density signal to score token importance for both inference phases: bipartite merge for prefill and KV eviction for decode, yielding measurable gains in each stage. To isolate the two phases, we split benchmarks by scoring logic. Prefill probes (MMBench, POPE, ScienceQA) evaluate only the first decoded token; their accuracy reflects prefill logits argmax before any decode KV eviction. Full-pipeline probes (TextVQA, DocVQA, MM-Vet) assess the complete generated sequence and capture both phases. We identify a critical evaluation pitfall: decode pruning claims validated solely on prefill probes measure prefill quality, not decode performance. Experiments show discarding all visual KV during decode barely affects prefill-probe accuracy but degrades TextVQA and MM-Vet drastically; this insensitivity stems from the scoring rule, not compression this http URL on seven model variants (InternVL-3.5, Qwen3-VL, LLaVA-1.5) and three full-pipeline benchmarks, our method achieves a three-way Pareto frontier across compute, memory and accuracy. Density-based prefill compression maintains strong accuracy, scales linearly with visual token count, and delivers 100x faster token selection at high resolution. For decode, density top-k KV eviction consistently outperforms strong baselines H2O, SnapKV and StreamingLLM on TextVQA and MM-Vet across three retention ratios. At 0.125 keep ratio on TextVQA, it surpasses H2O by 9.6 points; InternVL-3.5 models (2B to 38B) maintain a 7 to 10 point advantage. Compression scales favorably with model size: every measured 8B to 38B model transition reduces accuracy loss instead of worsening it. We suggest MM-Vet or DocVQA as mandatory full-pipeline validation for any decode-phase claims. Our code is available at this https URL
[CV-144] Gaussian Splatting for Real-Time Scene Reconstruction: Methods Systems and Trade-offs
链接: https://arxiv.org/abs/2610.17513
作者: Ashim Dahal,Rabab Abdelfattah,Nick Rahimi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted version to Els Pattern Recognition
Abstract:3D Gaussian Splatting (3DGS) allows for high-quality real-time novel-view synthesis. Since its inception in 2023, its explicit and highly configurable scene representation has been extended toward sparse-view generation, dynamic reconstruction, efficiency and compression, streaming, and simultaneous localization and mapping (SLAM). Although Gaussian Splatting can provide real-time rendering, rendering an existing representation is distinct from constructing, updating, or maintaining the scene itself. In this survey, we synthesize current literature, categorize them and provide a connection of current approaches and tools towards an end-to-end pipeline. Explicitly, we categorize GS scene reconstruction into five overlapping families: single-shot generation, dynamic representations, efficiency and compression, streaming and immersive systems, and Gaussian Splatting SLAM. We distinguish render-only throughput from task-specific scene generation, update, tracking, mapping, and optimization costs, and combine controlled sparse-view benchmarking with native evaluations to identify cross-family quality, latency, memory, and update-cost trade-offs.
[CV-145] CÆNL: Collapse-Based Active Neural Learning with Objective-Aligned Representation Control
链接: https://arxiv.org/abs/2610.17509
作者: Saad Alqithami
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: Code and numerical reproduction: this https URL
Abstract:Active learning changes the representations used to acquire subsequent labels. We study this interaction through Collapse-Based Active Neural Learning (CÆNL), which separates acquisition, Dynamic Collapse Regularization (DCR), and its coefficient policy. DCR penalizes a normalized within-class to between-class scatter residual above a frozen reference, with a separate class-separation safeguard. Fixed, scheduled, deterministic-feedback, and learned-feedback policies apply this objective. The analysis establishes scatter invariances, conditional local descent, and coefficient bounds, and shows why clean geometry alone cannot guarantee input robustness. Six-seed classification comparisons are complemented by a 104-run study with separate control selection and ten-seed evaluations on each of two backbones. At 60% annotations on ImageNet-100, entropy acquisition with DCR and MACC-Lite reaches 73.420 +/- 0.512% accuracy on ResNet-50 and 73.086 +/- 0.397% on ResNet-18 (mean +/- seed SD), exceeding entropy alone by 1.742 and 1.834 percentage points. Both gains survive Holm correction under paired-t and exact sign-flip tests, with unchanged feedback settings across backbones. Independently tuned constants attain slightly higher mean accuracy; the paired comparisons establish neither adaptive superiority nor equivalence. Separate six-seed spectral studies yield a small fixed-regularization C4 loss reduction, worse CIFAR-10 diffusion FID, and inconclusive AudioCaps captioning effects. Earlier classification studies find near-zero AutoAttack accuracy at L-infinity radius 8/255 and no added benefit from learned control. The results support the evaluated combined classification treatment and identify the task dependence of geometry-informed regularization.
[CV-146] Efficient Fréchet Distributional Decoder Alignment for Latent Generative Models
链接: https://arxiv.org/abs/2610.17466
作者: Xianghong Fang,Jiajun Zhu,Esma Aimeur,Dehan Kong,Tim G. J. Rudner
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 18 pages, 17 figures, and 6 tables
Abstract:Fréchet Distributional (FD) objectives provide an effective distribution-level signal for post-training generative models. For latent generative models, however, existing FD-based post-training methods optimize only the latent generator while keeping the decoder frozen: the decoder is trained on reconstruction latents obtained from real images but deployed on generator-produced latents at generation time. When these latent distributions differ, this \textbfdecoder train-generation mismatch means that the decoder is deployed on inputs drawn from a distribution it was not directly trained on. We introduce \textbfFréchet Distributional Decoder Alignment (FDDA), an efficient post-training method that freezes the latent generator and directly optimizes the decoder on generation-time latents using an FD objective. A single epoch of FDDA consistently improves gFID and gFDr ^6 across diverse flow-based and autoregressive latent generative models. Across all 11 evaluated models, FDDA reduces gFDr ^6 by 34.0-59.6% from the original pretrained models and provides a further 10.2-57.2% reduction after generator-side FD post-training. FDDA also reduces total wall-clock time by factors of 3.6\times - 328.7\times relative to generator-side FD post-training across these models. These results show that generator-side post-training alone does not exhaust the benefits of FD optimization, and that further gains can be obtained by adapting the decoder at substantially lower cost.
[CV-147] ableVision: Recovering Verified Structure from Historical Tables
链接: https://arxiv.org/abs/2610.17448
作者: Todd Y. Zhou,James M. Snyder Jr
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Gaps, skips, and mistakes are the normal state of a historical register: rows go missing, tables cross page breaks, columns slip under wrong names, and column sums disagree with the corresponding row entries. Readers read digits correctly but lose the table structure. On seven clean panels of election returns, the best open-source reader reproduces 99.0% of the digits but files only 83.7% of the figures in the correct cells and gets 11 of 30 columns correct. TableVision treats table structure recovery as the problem. On three degraded 1888 pages, Amazon Textract files only 1 of 1494 figures correctly; TableVision correctly files 99.2%, and all 51 columns its exact-total check releases are correct. A dynamic program cuts a volume into blocks by how many columns sum to their last line; witnesses read the cells; the gate classifies each column as exactly reconciling, flagged as faithful to a page whose arithmetic is wrong, or withheld. A theory of the sum check says which errors it cannot see and why its optimal block is about \theta^-1/2 rows at defect rate \theta . On 2201 columns of Wisconsin electoral data, the gate accepts 2042, flags 118, and withholds 41; failure rises with column length R as 1-(1-\theta)^R at 1.1 defects per 1000 cells in general elections; official returns confirm 933 of 934 testable accepted sums and all 35 testable flagged totals; and the deployment added 491,359 gated precinct rows to a 9.3-million-row dataset. One implication is that the design of the verification block is critical.
[CV-148] Know-Thy-Not: Training-Free Targeted Negation in Dual-Encoder Vision-Language Models
链接: https://arxiv.org/abs/2610.17440
作者: DoYoung Kim,Changkyu Choi,Adín Ramírez Rivera,YunSeok Choi,Jee-Hyong Lee
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 9 pages
Abstract:Dual-encoder vision-language models systematically fail at negation, scoring ‘‘a cat without a dog’’ nearly identically to ‘‘a cat with a dog.’’ We show that this failure is inherent to the single-embedding framework. Under contrastive training, single-embedding methods can at best ignore the negated concept (dilution) but never penalize its presence (rejection). We propose Know-Thy-Not (KTN), a training-free method that introduces an image-side rejection dimension to penalize the negated concept directly in the visual representation. We also introduce COCO-Confusion, the first benchmark requiring genuine rejection to succeed. Across 9 backbones, KTN improves COCO-Confusion mAP by up to +25.6 over vanilla VLMs and +9.6 over the strongest existing method.
[CV-149] World-T3: Test-Time Training as Persistent Memory for Real-Time Video World Models
链接: https://arxiv.org/abs/2610.17434
作者: Junyi Wu,Zekun Li,Chenhongyi Yang,Chen Guo,Lingchen Yang,Yongan Zhang,Xinyi Zhang,Hongbing Hu,Bartek Rymkowski,Srinath Sridhar,Yan Yan,Xijie Huang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Multimedia (cs.MM)
备注: Project Page: this https URL
Abstract:Recent advances in video world models have enabled plausible interactive environments, yet long-horizon generation places a fundamental demand on visual memory: the previously observed world must remain consistent as the model continues to generate. Existing autoregressive approaches often implicitly rely on the generative model itself to retain such information, leading to spatial memory decay beyond the active context. We introduce World-T3, a camera-controlled video world model with a test-time training memory that consolidates past observations into dedicated model parameters. To robustly update this memory, we propose a recursive least-squares rule that incrementally absorbs novel visual features via online pose-keyed memory updates. Our TTT memory reads out previously observed scene features from the fast-weight state by camera pose, enabling consistent generation with constant storage and per-step computation regardless of rollout length. To facilitate real-time interactive generation with this robust memory, we further distill the autoregressive model into a few-step generator via dual-teacher distribution matching. On a long-horizon revisit benchmark, World-T3 raises revisit PSNR from 13.52 to 18.31 dB and reduces absolute rotation error from 3.344 to 1.075 relative to the strongest baseline on both metrics, while sustaining 20.6 FPS at 1280 x 704 on a single NVIDIA B200 GPU.
[CV-150] Bottlenecked semantic compression by representation reconstruction
链接: https://arxiv.org/abs/2610.17412
作者: Xiaoya Tang,Shilong Liu,Daniel Smolyak,Xiwen Li,Quynh Nguyen,Tolga Tasdizen
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Vision Transformers incur high computational costs when processing long visual token sequences. Existing reduction methods typically rank tokens independently and discard those deemed unimportant or redundant. We propose a reconstruction-guided framework for visual token condensation (ReConV), which summarizes the complete visual sequence with a compact Summary Bank (SMB) and retains tokens whose information is not adequately captured by the SMB, as indicated by their reconstruction residuals. At each compression layer, a lightweight position-conditioned decoder reconstructs the current tokens, leaving shared information compressed in the SMB while explicitly preserving poorly reconstructed details. Experiments on image classification and multimodal understanding demonstrate favorable performance-efficiency trade-offs: ReConV surpasses the uncompressed baseline with 48% fewer GMACs on ImageNet and enables reusable image-level representations for efficient multi-turn inference. Code is available at TokenDrop.
[CV-151] What Makes a Latent Space Shared? Direct and Recoverable Compatibility in Motion VAEs
链接: https://arxiv.org/abs/2610.17376
作者: Iago Brito,Walcy Rios,Lucas Fernandes,Kaiki de Oliveira,Arlindo Galvão Filho
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:A shared latent space is often assumed to make representations from different schemas directly interchangeable, yet sharing parameters, dimensionality, or a prior does not determine how corresponding instances are expressed in latent coordinates. We show that this assumption conflates recoverable correspondence, direct compatibility, and decoder reuse. In a controlled multi-schema motion VAE on HumanML3D, even identical inputs yield only 5.9 – 17.7% raw Recall@1 and 0.221 – 0.484 paired-token cosine, despite correspondence being almost perfectly recoverable after linear calibration. Cycle consistency strengthens recoverability but still leaves direct retrieval and token agreement near zero. In contrast, only 64 paired clips raise raw Recall@1 to 88.4 – 90.3% and token cosine to approximately 0.82 , making most post-hoc alignment unnecessary, although crossed decoding remains substantially worse than native reconstruction. We further find that reducing measured marginal discrepancy does not imply schema invariance. These results show that a shared backbone or prior does not ensure directly compatible latent coordinates, while limited paired correspondence can substantially improve direct compatibility in the tested setting.
[CV-152] WIRE: Weight Space Implicit Representation Encoding of Frozen Video LLM s
链接: https://arxiv.org/abs/2610.17355
作者: Ameen Ali,Tamim Zoabi,Lior Wolf,Chaim Baskin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Video language models represent a video as a long sequence of visual embeddings in a decoder’s input. Every visual embedding is processed by every decoder layer during prefill, stored in the key value cache, and attended by every generated token, so a long video raises prefill, memory, and generation cost at once. We present WIRE (Weight Space Implicit Representation Encoding), which reduces this recurring cost by moving part of the video conditioned computation of a frozen video language model from its input activations into a temporary low rank update of its weights. WIRE keeps a short explicit visual context, so the update acts as a correction rather than a replacement of the video. WIRE first compresses the full resolution embeddings of each frame into a small set of carrier tokens using a question conditioned bidirectional selective scan, and physically contracts the decoder input so every decoder layer, cache entry, and decoding step operates on the shorter sequence. Hypernetworks at selected decoder layers read the compressed forward pass together with a readout of the full resolution features removed by contraction and generate LoRA updates to the attention and feedforward projections. The updates are merged after prefill, so autoregressive decoding runs the unmodified frozen backbone. On Qwen3-VL-8B, WIRE with 16 carrier tokens per frame contracts the visual stream by 3.5\times , reduces decoder prefill time by 3.35\times at 256 frames and 4.14\times at 1024 frames, shrinks the key value cache proportionally, and keeps per token decoding latency approximately constant. Across nine benchmark settings at 256 frames, it obtains the highest average accuracy among efficient video language models we evaluate under a common 256 frame budget.
[CV-153] MID: Mask-Image Distributional Divergence for Evaluating Medical Image Segmentation
链接: https://arxiv.org/abs/2610.17354
作者: Vincenzo Marcianò,Xiaoming Zhang,Gianluca Guglielmo,Sebastien Ourselin,Michela Antonelli,Maria A. Zuluaga
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Current evaluation protocols for medical image segmentation remain largely anchored in sample-wise comparisons, e.g., Dice and Hausdorff distance, which require each prediction to be paired with a ground-truth mask. Yet two limitations arise in practice: they are inapplicable when predictions are made on unlabeled data; and even when annotations are available, they cannot assess whether a collection of predictions matches the distributional structure of the ground-truth population. We introduce MID (Mask-Image Distributional Divergence), a distribution-level evaluation metric for medical image segmentation. MID builds upon the observation that segmentation quality is a property of the joint image-mask distribution: a mask may be anatomically plausible in isolation yet incorrect for the image it accompanies. MID compares the joint distribution of images and masks between a reference set and a set of predictions, without requiring sample-wise correspondence. This captures both mask realism and image-mask consistency. We evaluate MID across 13 publicly available datasets spanning CT, MRI, and MRA. Under anatomically realistic corruptions across 56 organ groups, MID tracks mean Dice and Hausdorff distance without per-sample ground-truth pairing. MID also detects distributional shifts across modalities and anatomies, identifies image-mask misalignment through controlled shuffling experiments, and produces reliable model rankings across real segmentation models. Official implementation available at this http URL
[CV-154] HairFlow: Modeling 3D Hair Strands with Multi-Modal Conditioned Diffusion
链接: https://arxiv.org/abs/2610.17348
作者: Yating Wang,Kui Wu,Ran Yi,Zixuan Lu,Tongtong Wang,Zhongtian Zheng,Lizhuang Ma,Yuefan Shen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Modeling 3D hair strands is challenging due to complex hairstyle geometry, scarce high-quality training data, and incomplete or noisy input observations. Foundation models offer broader hairstyle priors, but their outputs are not directly usable for strand-based rendering and simulation. Existing approaches require costly per-instance optimization to fit hair priors to meshes from foundation models and 2D orientations from specialized predictors trained on limited hair data, restricting efficiency and generalization. We present HairFlow, a multimodal conditioned diffusion framework for efficient and controllable hair strands modeling. HairFlow conditions strand generation on two complementary representations: a coarse mesh that constrains global shape and multi-view sketches that specify local strand flow. HairFlow combines an Orientation VAE with a conditioned diffusion transformer to generate volumetric fields that are traced into explicit strands. To overcome the lack of paired sketch-strand training data, we introduce an automatic synthesis pipeline that generates geometrically aligned multi-view sketches with diverse styles representative of inference-time inputs. Experiments show that HairFlow generates plausible 3D strands that faithfully follow the input conditions across diverse hairstyles, substantially reduces inference time, and supports flexible multimodal control.
[CV-155] SSM-Attack: Learning Where to Perturb for Black-Box Face Privacy
链接: https://arxiv.org/abs/2610.17323
作者: Tunahan Parlayıcı,Erchan Aptoula,Yücel Saygın
类目: Computer Vision and Pattern Recognition (cs.CV); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
备注: 25 pages, 3 figures, 6 tables. Preprint submitted to Pattern Recognition
Abstract:Face recognition systems are deployed widely enough that a photograph shared online can be matched without its subject’s consent. Generator-based protection methods rely on heavy pretrained backbones, multi-stage inference or visible semantic edits. We introduce SSM-Attack, a black-box face privacy framework built on Selective State Space Models. Rather than synthesizing the perturbation, the state-space model predicts in one linear-time pass a spatial allocation of the perturbation budget, marking which facial locations carry identity evidence; a short projected refinement then sets the values within this per-pixel budget. Against published benchmark results, SSM-Attack attains the highest average black-box attack success rate on all three evaluation sets: 87.10 percent on CelebA-HQ, 86.90 on LADN and 81.15 on FFHQ, the last from a network that never saw that dataset. The allocation network holds 5.73 million parameters, or 16.13 million with the frozen networks used by its quality terms. Under the same cap of 0.10 on the maximum per-pixel change, allocation improves structural similarity by 0.044, peak signal-to-noise ratio by 1.62 decibels and Frechet Inception Distance by 11.2 points over a uniform map while spending 37 percent of its capacity. Strong facial privacy protection therefore does not require a large generative prior; it requires knowing where to spend a small one.
[CV-156] HybridDualPathNet: Cross-Fusion Attention of EfficientNetV2 and Swin Transformer for Brain MRI Classification of Alzheimers and Parkinsons Disease
链接: https://arxiv.org/abs/2610.17251
作者: Jarpula Nirjala,Rahul Patel
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Software Engineering (cs.SE)
备注: Two experimental settings: Kaggle-only image-level evaluation and combined Kaggle and ADNI evaluation with patient-wise separation for ADNI subjects
Abstract:Separating Alzheimer’s Disease (AD), Parkinson’s Disease (PD), and Healthy Controls (HC) using structural MRI remains challenging because disease signatures overlap and model performance can be strongly affected by data partitioning. We present HybridDualPathNet, a hybrid architecture combining EfficientNetV2-S for local features and Swin-Tiny for global context through a Cross-Fusion Attention (CFA) module. We evaluate the model in two experiments. First, on the Kaggle APD-3-Class dataset using an image-level split, the model achieves 98.25% accuracy. Second, on a combined Kaggle and Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset containing 17,859 slices, we use a per-source split with patient-wise separation for ADNI subjects. In this setting, the model achieves 73.37% standard accuracy and a macro AUC of 0.880. Source-stratified evaluation reveals a substantial domain gap, with 91.2% accuracy on Kaggle test images and 59.2% on ADNI test images. Temperature scaling reduces expected calibration error from 0.130 to 0.015, while split-conformal prediction achieves 89.5% coverage. Saliency analyses using Grad-CAM++, Integrated Gradients, and occlusion indicate that model activations concentrate on brain tissue. These results highlight the importance of patient-level separation, cross-dataset evaluation, and uncertainty estimation when assessing MRI classification systems. The findings also show that strong performance on a single dataset does not necessarily translate to reliable generalization across imaging cohorts.
[CV-157] When Left Is Right? Camera-Motion Understanding beyond Screen
链接: https://arxiv.org/abs/2610.17242
作者: Chong Cheng,Ran Zhang,Xinyuan Li,Alan Zhao,Hao Wang,Xi Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Can a vision-language model tell left from right? Modern VLMs solve visual puzzles and follow hour-long videos, yet when asked whether the camera moved left or right, most do no better than a coin flip. Adding camera-motion data or geometric representations has been tried, separately, and neither closes the gap. We argue that camera-motion understanding needs two conditions at once: the representation must couple off-screen camera motion with on-screen flow dynamics, and it must be interpreted by the language model rather than trained as a standalone pose or depth estimator. We propose CameraVLM, which pairs a motion-state representation with state-language alignment. Camera tokens capture how the camera moves off screen, flow tokens capture how the subject and background move on screen, and a two-stage training strategy keeps language in the loop so that both are learned together with camera-motion semantics. With only 9B parameters, CameraVLM outperforms all open-source and camera-motion-specific models by a wide margin across three benchmarks and more than 20 baselines, and matches or surpasses GPT-6 Astra as well as human viewers in a 48-participant study. Project page: this https URL
[CV-158] IsoVRBench: Video Reward Modeling Without Human Feedback
链接: https://arxiv.org/abs/2610.17234
作者: Shuyue Stella Li,Jacqueline He,Jiahui Chen,Sreya Dutta Roy,Yulia Tsvetkov,Luke Zettlemoyer,Marjan Ghazvininejad,Xiaochuang Han
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: 20 pages, 18 tables, 6 figures
Abstract:Video reward models are trained and evaluated on human preferences between videos that often differ simultaneously on properties such as prompt alignment, visual fidelity, motion, and generator identity. As a result, aggregate accuracy does not reveal which video properties a reward model can reliably distinguish. We construct preference pairs by controlled interventions that change one property of a generated video while holding the rest fixed. The resulting pairs, by construction, provide preferences for how the videos differ along a specific dimension, without requiring any human annotation. We propose IsoVRBench, a video generation reward modeling benchmark of 8,000 such pairs across eight interventions, and show that state-of-the-art models separate broad quality differences between generators but miss important features such as temporal order. The best reward model reaches only 60.8% on reordered frames against 100% for humans, and rankings on existing benchmarks do not predict this failure. Using the synthetic isolated intervation pairs as training data leads to 1) IsoVRM, a reward model outperforming its counterpart trained with a larger human labeled preference dataset by 3.8 points on VBench-2.0, 2) IsoWan, the video generator post-trained with IsoVRM, outperforming its human-data counterpart in human preferences, and 3) gains that stack when combined with human data. Because intervention pairs are automatically generated rather than collected, the benchmark and training data can be easily extended as generators evolve.
[CV-159] Beyond Sparse Captions: Aligning Slide-Level Text and Patch-Level Vision in Pathology NEURIPS2026
链接: https://arxiv.org/abs/2610.17233
作者: Daniel Shao,Dongmin Bang,Luca L. Weishaupt,Sophia J. Wagner,Nic G. Reitsam,Ming Y. Lu,Long P. Le,Richard J. Chen,Faisal Mahmood
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Neurips 2026 Spotlight
Abstract:Contrastive vision-language (VL) pretraining in computational pathology has generally relied on curated image-caption datasets sourced from social media, educational videos, and research publications. These corpora are sparse relative to the diversity of tissue morphology and difficult to scale further, as pathologists do not routinely produce descriptive captions in clinical practice. We propose to instead leverage slide-level pathology reports which can be generated during routine diagnostic workflows in a scalable manner. A key challenge in this weakly-supervised setting is that pathology reports describe macroscopic, diagnostic findings while patch-level visual features describe microscopic findings. We introduce Locked Image Multiple Instance PreTraining (LIMIT), which addresses this gap by freezing a pretrained vision encoder and fine-tuning only the text encoder: report embeddings serve as cross-attention queries over bags of patch features, producing report-conditioned slide representations optimized via InfoNCE. Using LIMIT, we establish CONCH-Z, which aligns the text encoder of CONCH v1.5 to 335k clinical reports. Evaluated across 22 tasks spanning patch-level classification, tumor detection, and slide-level subtyping, CONCH-Z establishes state-of-the-art zero-shot performance over VL encoders trained on curated caption datasets, while simultaneously closing the gap with slide-level foundation models trained with substantially more parameters and compute. The code and model is available at this https URL.
[CV-160] AReA: Adaptive Receptive-Field Attention for Efficient High-Resolution Diffusion Transformers
链接: https://arxiv.org/abs/2610.17227
作者: Javad Rajabi,Koorosh Roohi,David B. Lindell,Babak Taati
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 27 pages, 14 figures. Project page: this https URL
Abstract:Generating images beyond the training resolution of Diffusion Transformers (DiTs) remains challenging, as the number of tokens grows rapidly with resolution, full attention becomes expensive, and outputs often suffer from structural distortions and degraded detail. We identify uniform full attention as the common source of both problems. Attending to all tokens at every step causes quadratic cost and exposes the model to more tokens and relative distances beyond the training range, pushing the model out of distribution. We observe that attention’s effective receptive field narrows during denoising, from global in early steps that form structure to local in later steps that refine details. We propose AReA, a training-free attention mechanism whose receptive field evolves from coarse to fine, following the denoising progression. In early steps, AReA applies cross-scale coarse attention over downsampled keys and values, capturing global context, while in later steps it applies local attention within progressively shrinking windows. Without additional parameters or training, experiments show that AReA consistently outperforms state-of-the-art training-free baselines across target resolutions and different models, while also being faster and more efficient.
[CV-161] DexNinja: Learning Robust Dexterous Cutting Policy with a Real-to-Sim-to-Real Data Engine
链接: https://arxiv.org/abs/2610.17208
作者: Haozhe Lou,Runyi Yang,Wanqi Zhong,Caiyun Liu,Yicheng Liu,Wenhao Liu,Wenlong Ma,Jiawei Xia,Kanzhi Wu,Danda Pani Paudel,Luc Van Gool,Hang Zhao,Yiming Li
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Cutting a piece of food looks effortless in human hands. For a robot, however, every cut is a high-contact interaction with a deformable, elasto-plastic object whose behavior changes with pose, contact mode, and cutting stage. Collecting real-world data is expensive and wasteful: each trial typically contaminates or damages the food, so obtaining scalable real world data for dexterous cutting is expensive and difficult. Simulation is an appealing alternative, but a large sim-to-real gap exists in both physics and perception. This gap is especially severe for topology-changing interactions like cutting, where small errors in contact, friction, or material and elasticity responses quickly compound. We propose \textscDexNinja, a real2sim2real framework that turns a handful of real demonstrations into a dense and realistic training distribution. \textscDexNinja (i) reconstructs object instances from real captures, (ii) randomizes physically meaningful parameters under category-level constraints, and (iii) produces augmented dexterous manipulation episodes using a custom differentiable simulator that couples robot dynamics, tactile contact, and deformable cutting. We evaluate \textscDexNinja on a dexterous food-slicing task and show that (1) with tens of real demonstrations and hundreds of augmented simulator data, the success rate could reach more than 60%. (2) the augmented data consistently improves success rates when only using real data with more than 20% via sim-to-real, and (3) the resulting policies generalize to out-of-distribution objects with varying shapes and sizes.
[CV-162] Multics: Generative Multi-Player World Model with Hierarchical Communication
链接: https://arxiv.org/abs/2610.17201
作者: Kaiyi Huang,Wangbo Yu,Yukun Huang,Jiwen Yu,Kunhao Liu,Donglin Yang,Wenbo Hu,Ying Shan,Xihui Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Multi-player interactive world models aim to simulate shared environments from multiple players’ perspectives, but achieving robust generation while scaling to more players remains challenging. Existing approaches that rely on fully-connected communication (e.g., Solaris) incur quadratic attention costs as the number of players grows, while centralized approaches (e.g., \gamma -World) suffer from interference between unrelated events in a shared global representation. To address these challenges, we present Multics, a multi-player video generation framework with hierarchical communication that structures information exchange across player, group, and global levels. This hierarchy introduces group tokens into self-attention as intermediate group-level representations that aggregate locally relevant player information. These tokens connect to global tokens for cross-group communication, while the hierarchical topology limits direct mixing between unrelated local events. Group-token parameters are shared across group slots, allowing the same communication module to accommodate varying numbers of players without introducing new group-specific parameters. We further propose budget-constrained dynamic grouping based on recent camera trajectories, which adaptively groups relevant players while limiting unnecessary communication. Experiments demonstrate that Multics achieves state-of-the-art performance in 4- and 6-player settings, with improved cross-player consistency, player-specific action and camera controllability, and interference robustness. Moreover, Multics generalizes from 4-player training to 6-, 8-, and 10-player inference without retraining, while reducing self-attention computation compared with fully-connected communication.
[CV-163] ChemExplain: A Multimodal Benchmark for Explanatory Chemical Reasoning
链接: https://arxiv.org/abs/2610.17183
作者: Aotian Chen,Frazier N. Baker,Reza Averly,Shan Zhang,Tianao Chen,Daniel Adu-Ampratwum,Xia Ning,Yuan Xue
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Current chemistry benchmarks primarily evaluate outcome prediction or reaction pathway generation, but rarely test whether a model can explain why an observed chemical effect occurs. We introduce ChemExplain, a benchmark of 1,180 multiple-choice chemistry questions drawn from 383 peer-reviewed papers. Each question pairs a molecular or reaction image with an observed chemical effect and four candidate explanations, each associated with a different structural feature visible in the image. Answering requires the model to perceive the depicted structure, apply chemical knowledge, and identify which feature most directly explains the effect, a capability we term explanatory attribution. The correct feature-mechanism relation is derived from supporting evidence in the source paper rather than invented by the question-generation LLM. We construct the benchmark automatically with a multi-agent framework that applies progressive quality control at three stages and validate it with eight chemistry domain experts. The strongest evaluated model reaches 78.6% accuracy, leaving considerable room for improvement. On a 600-question expert-reviewed subset, reasoning-trace analysis classifies 71.4% of the strongest model’s errors as attribution failures. Attribution is also the largest error category for the open-source and chemistry-specific models included in our analysis.
[CV-164] MARS: Systematic Multi-Level Acceleration for World Action Models
链接: https://arxiv.org/abs/2610.17141
作者: Zengyuan Zuo,Yongchang Zhang,Guangyu Zhao,Tianze Zhang,Biao Wu,Meng Fang,Yiwu Zhong
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:World Action Models (WAMs) have emerged as a promising alternative for embodied intelligence by jointly predicting future world states and robot actions. However, their heavy visual encoding and iterative generative inference incur substantial latency, limiting their applicability to high-frequency closed-loop control. Existing training-free acceleration methods typically optimize a single stage, yielding limited speedups or degraded control under aggressive reuse. In this work, we systematically profile the inference overhead of WAMs and observe abundant redundancy across Transformer computation, diffusion denoising, and consecutive policy queries. However, the safe removal of such redundancy depends strongly on its impact on action prediction and control sensitivity. Based on these observations, we introduce \textbfMARS, a training-free and plug-and-play framework for multi-level WAM acceleration. MARS selectively removes low-impact attention computation, adaptively accelerates diffusion inference according to motion granularity, and speculatively reuses predicted future visual representations across consecutive queries with online verification. Together, these complementary mechanisms provide remarkable acceleration while limiting approximation error. Extensive experiments across LIBERO, LIBERO-PRO, and RoboCasa show that MARS achieves up to 4.00\times , 3.94\times , and 2.10\times end-to-end speedups, respectively, while maintaining task success rate close to the original policy. Moreover, MARS generalizes across different model backbones and across simulation and real-robot experiments, substantially improving the efficiency–performance trade-off of WAM inference. Our code is available at this https URL.
[CV-165] EgoProbe: Benchmarking Long-Range and Exact Memory in Streaming Egocentric Videos
链接: https://arxiv.org/abs/2610.17128
作者: Haokai Pang,Yuqian Fu,Yifei Wang,Yujiao Shen,Tianwen Qian,Mohamed Elhoseiny,Hongbo Fu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 22 pages, 9 figures, 8 tables
Abstract:Egocentric video provides a continuous first-person record of everyday activities, making it a natural foundation for wearable AI assistants. Unlike offline video understanding, streaming systems must process observations as they arrive, while queries may refer to experiences from the distant past. Recent work has begun to explore whether information can be recalled across long temporal gaps. Yet remembering how far is only part of the challenge: recurring objects, actions, and locations also require models to remember exactly which detail belongs to which event. We introduce EgoProbe, a benchmark that evaluates streaming egocentric memory along these two complementary dimensions: long-range retention, i.e., how far, and event-specific recall under interference, i.e., how exact. Built on long-form egocentric videos, EgoProbe uses an event-grounded data engine to extend recall intervals, identify related events, and construct event-grounded distractors, followed by human verification. It contains two subsets: EgoProbe-Base, which evaluates long-range recall without intervening related events, and EgoProbe-Echo, which challenges models with similar intervening events and event-grounded distractors. EgoProbe-Echo adopts coarse-to-fine multi-turn probing: the first turn evaluates coarse-grained event recall, while the second tests whether the model can retrieve the target-specific detail associated with the designated event. Extensive evaluation across diverse models reveals substantial limitations in long-range recall, with further challenges in recovering target-event details among related experiences. Even when models correctly answer coarse-grained questions, they often fail to recall the specific details of the designated event. EgoProbe exposes this underexplored failure mode and provides a testbed for advancing long-range and exact memory in streaming egocentric video understanding.
[CV-166] Contrastive learning for video reuse detection
链接: https://arxiv.org/abs/2610.17098
作者: Tristan J.B. Cann,Arjun Biswas
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The incorporation of footage from other sources has long been a concern from a copyright perspective, as such usage can harm the interests of rights holders. Recent trends in the ``remixing’’ of social media content and the development of bespoke stock libraries for specialist causes mean that methods to identify content reuse will have a range of applications. In this paper, we develop a framework based on a Siamese network fine tuned using contrastive learning to identify potential sources for key frames in a given video. This framework makes two contributions to analysis of large video datasets. Firstly, we demonstrate its utility at reducing complex video data to a set of key visuals typically less than 10% of the size of the full set. Secondly, we show that these reduced sets allow for efficient and effective identification of component footage. We test this method over a set of videos compiled from a stock footage library and demonstrate its effectiveness for constituent video identification with average accuracy greater than 0.95.
[CV-167] Scaling Unified Multimodal Models to Long-Horizon Interleaved Generation
链接: https://arxiv.org/abs/2610.17080
作者: Ziqiao Peng,Yutao Cui,Wang Lin,Zhong-Yu Li,Qingwen Gu,Tianhui Song,Chengyou Jia,Guozhen Zhang,Xuerui Qiu,Qixun Wang,Miles Yang,Zhao Zhong,Yonglong Tian,Hongyan Liu,Jun He
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 29 pages, 11 figures
Abstract:Unified multimodal models (UMMs) provide an interface for interleaved text-image generation, where generated visual states can directly inform subsequent reasoning and synthesis. However, existing systems are still primarily demonstrated on short trajectories or simplified visual content. We develop a large UMM natively trained on interleaved multimodal data, with strong capabilities in image synthesis, text rendering, complex composition, and multimodal planning, and study how to extend these native capabilities to substantially longer autoregressive trajectories. The main challenge is that generated images continuously accumulate in context: irrelevant or imperfect visual states interfere with future synthesis, while previously established characters and scenes must still be recovered after long intervals. We introduce a visual-memory framework that maintains persistent visual references, selectively retrieves scene-relevant historical states while preserving the complete semantic history, and uses few-step self-rollout to adapt the model to imperfect generated histories. We further introduce LongMIG-Bench for evaluating long-horizon text-to-multi-image and image-to-multi-image generation. Our model achieves high-fidelity long-horizon interleaved generation with strong character and scene consistency, demonstrating how visual memory can scale native interleaved generation to substantially longer trajectories.
[CV-168] Rethinking LLM -Guided Visual Question Answering: Does Deeper Knowledge Integration Really Help?
链接: https://arxiv.org/abs/2610.17079
作者: Aneerban Saha,Vijay Kumar,Sameer Madhav D.,Priyam Srivastava,Dhruv Bansal
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 13 pages, 6 figures. Accepted at ADCIS 2026
Abstract:Knowledge-based Visual Question Answering (VQA) requires reasoning that extends beyond visible image content, demanding external commonsense and factual knowledge. While recent work leverages Large Language Models (LLMs) as knowledge sources, the question of how deeply such guidance should be integrated into a vision language architecture remains underexplored. In this work, we generate structured reasoning guidance, visual focus concepts, distractor elements and external knowledge using Llama~3.3~70B and systematically compare two integration strategies within a unified frozen ViT + FLAN-T5 framework on the OK-VQA benchmark: decision level late fusion and attention-matrix injection. Contrary to the prevailing assumption that deeper integration is superior, we find that shallow late fusion achieves the strongest results (39.4% exact match, +10.0 over the no guidance baseline), outperforming deep attention modification (31.8%) under data scarce fine-tuning. We further contribute an LLM based augmentation pipeline that expands training data more than tenfold via question paraphrasing. Our findings indicate that integration depth must be matched to the data regime.
[CV-169] A unified three-branch deep learning framework for automated detection and counting of wildlife in aerial imagery
链接: https://arxiv.org/abs/2610.17067
作者: Zeyu Xu,Tiejun Wang,Andrew Skidmore,Megan Claase,Harry Wells,Isla Duporge,Ramiro D. Crego,Richard Lamprey
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Automating aerial wildlife surveys requires reliable animal detection, consistent delineation of observation strips, and correction for repeated coverage between successive images. We present a unified three-branch deep learning framework that integrates these tasks through WildSeg, WildMatch and WildDet. WildSeg combines rod segmentation with geometric refinement to delineate counting strips from aircraft-mounted markers. WildMatch uses teacher-student distillation to learn dense image correspondences and identify overlapping survey areas without relying on geolocation metadata. WildDet combines dual-scale detection with feature fusion, density supervision and empty-image gating to support animal detection and counting. We evaluated the framework using oblique aerial imagery of four antelope species from the Great Nile Migration Landscape in South Sudan. WildSeg achieved reliable rod extraction, while WildMatch approached teacher-level accuracy with a substantially lighter and faster model. Both modules maintained high accuracy on resized images, improving processing efficiency. WildDet improved detection performance across all four species and reduced detection-based count MAE, RMSE and mean bias by 21.0%, 4.6% and 27.8%, respectively, relative to YOLO11. By connecting spatial survey constraints with animal detection, the framework provides a practical and computationally efficient approach to processing large aerial image collections for wildlife monitoring and population surveys.
[CV-170] PerturbBridge: Direct Phenotypic Supervision for Cellular Perturbation Generation
链接: https://arxiv.org/abs/2610.17066
作者: Yucheng Yang,Xiaoxiao Sun,Huangqingbo Sun,Elaine Sui,Emma Lundberg,Serena Yeung-Levy
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 25 pages, including appendix
Abstract:Virtual cell models should capture phenotypes induced by specific perturbations. However, image-based models are typically trained and evaluated to match real-image distributions, which alone does not explicitly prioritize perturbation-specific biological responses. Recent approaches add biological signals through reinforcement-learning post-training or inference-time steering, raising a central question: where should biological response information enter the generative pipeline? We introduce PerturbBridge, a stochastic data-to-data bridge model that directly supervises perturbation-specific phenotypes in predicted perturbed images during training. PerturbBridge jointly optimizes bridge regression and differentiable biological response losses, with biological gradients passing through a frozen image encoder. We further introduce perturbation-response evaluations based on downstream classification and retrieval. Across three microscopy benchmarks for perturbation profiling and subcellular imaging, PerturbBridge surpasses CellFlux in biological fidelity, image quality, and sampling efficiency. On BBBC021, mechanism-of-action classification improves by 5.66 and 7.5 percentage points with morphological and learned visual features, respectively. On JUMP and AllenCell, retrieval Hit@10 and Hit@1 improve by 1.77 and 7.15 percentage points, respectively. On BBBC021 and JUMP, PerturbBridge achieves lower FID with 80% fewer model evaluations than CellFlux. Overall, these results show that directly supervising perturbation responses during training improves phenotypic fidelity while enabling efficient cellular image generation. Code is available at this https URL.
[CV-171] VL-DocIR: A Benchmark for Vision-Based Long Document Information Retrieval NEURIPS2026
链接: https://arxiv.org/abs/2610.17057
作者: Simon Ott,Yufan Chen,Ruiping Liu,Junwei Zheng,Jiale Wei,Di Wen,Kunyu Peng,Jiaming Zhang,Rainer Stiefelhagen
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by NeurIPS 2026
Abstract:Vision-based document retrieval has improved rapidly, but current evaluation settings still emphasize short documents and single relevant pages. This leaves an important gap between leaderboard performance and realistic long-document retrieval, where the evidence needed to answer a query may span multiple pages or even multiple documents. We present VL-DocIR, a page-level benchmark for evidence-complete visual long-document retrieval. VL-DocIR is built from 29,641 HTML-born documents rendered into 388,548 page images from Wikipedia, arXiv, PubMed, and SEC proxy statements. It contains 271,760 queries across 23 domains and six evidence structures, covering single-page, within-document multi-page, and cross-document evidence configurations. Evidence annotations are grounded to rendered pages and HTML element identifiers, and queries are filtered with a cleaning pipeline targeting clarity, correctness, closedness, and the absence of explicit layout references. We evaluate leading single-vector and multi-vector visual retrievers together with diagnostic retrieval settings such as document-context scoring, query splitting, reranking, and round-robin ensembles. The strongest evaluated configuration reaches 69.3 nDCGAll@10 and 80.6 RecallAll@10 under structure-balanced Macro evaluation, but source- and evidence-structure breakdowns show that cross-document retrieval remains substantially harder than single-page retrieval, and that document-context scoring is especially helpful for within-document multi-page queries. VL-DocIR exposes these failure modes and provides a more demanding target for visual long-document retrieval. The new dataset and code will be made publicly available at this https URL.
[CV-172] DIME: Scaling Facial Representation Learning via Differential Masked Autoencoding
链接: https://arxiv.org/abs/2610.17015
作者: Hao Yu,Haoyu Chen,Hui Wei,Yan Jiang,Nicu Sebe,Guoying Zhao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Scaling up data and model capacity has driven substantial progress in visual representation learning. However, conventional pretext tasks exhibit diminishing returns on fine-grained facial analysis. We attribute this to a scaling bottleneck: these tasks model each image’s appearance and are dominated by identity, pose, and illumination, so subtle within-identity facial differences contribute little to the training signal. We introduce DIME, a differential masked autoencoder for large-scale facial representation learning. DIME builds on a simple observation: two images of the same person share identity-specific appearance, so their difference exposes facial changes. DIME reconstructs both masked images in an identity-matched pair and, through an Explicit Differential Direction Supervision (EDDS) objective, aligns the direction of their reconstructed difference with the true difference, while source isolation makes each reconstruction depend only on its own image. Shared appearance cancels in this difference, and because EDDS ignores magnitude, subtle facial changes count as much as large pose or illumination changes. DIME uses identity labels only to form pretraining pairs, requires no dense annotations, and learns dense facial representations with emergent part-level semantics. We pretrain DIME on 20 million high-resolution facial images consolidated from existing face datasets. DIME’s AU detection performance continues to improve as pretraining data scales from 6M to 20M images, with its BP4D F1 increasing by 5.9 points, while MAE gains only 0.5 points on the same subsets. DIME also outperforms specialized facial methods on AU detection, AU intensity estimation, cross-domain anti-spoofing, landmark localization, head pose estimation, face parsing, and face generation.
[CV-173] MAGNET: Manifold-Aware Graph Diffusion Network for Connectome Generation NEURIPS2026
链接: https://arxiv.org/abs/2610.17014
作者: Protyay Dey,Ayush Roy,Hyuna Cho,Won Hwa Kim,Vishnu Suresh Lokhande
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at NeurIPS 2026
Abstract:Functional brain connectome represents neural connectivity as a matrix of pairwise interactions between brain regions. Generation of functional connectomes is not only a question of validity; rather, having satisfied the constraints on the correlation matrix, the next step is to recover the class-conditional geometry buried under coarse labels. We propose MAGNET, a Manifold-Aware Graph Diffusion Network which uses a normalized-Cholesky representation of the manifold of correlation matrices that guarantees validity. MAGNET lifts noisy latent states into ROI-level region tokens and performs denoising with a relational inductive bias over brain atlas regions. To deal with structural problems induced by coarse labels, MAGNET employs class-anchored conditioning, amortized structural bridge, and relevance-preserving corruption. Across ABIDE, ADNI, and OASIS-3, MAGNET consistently achieves favorable results compared to previous manifold-aware approaches, demonstrating improvements of 7-21% in class-conditional fidelity ( \alpha,\beta -F1) across three cohorts and better sampling efficiency. Moreover, while training only with strict binary labels, MAGNET is capable of maintaining clinical heterogeneity through fine substructure of the connectomes in a zero-shot setting, improving subclass covariance alignment ( \lambda -MSE) by over 30%. These results suggest that geometric validity is a necessary but insufficient condition for clinical utility in connectome synthesis. Moreover, efforts in making the diffusion denoising class-conditional manifold aware finds utility beyond the highly curved brain connectome generation as this is a critical problem in various general settings. The code repository can be found at this https URL.
[CV-174] MaskSense: Confronting the Visual Exploration Trap in Masked Image Generation NEURIPS2026
链接: https://arxiv.org/abs/2610.17009
作者: Yawen Shao,Jie Xiao,Kai Zhu,Yu Liu,Hongchen Luo,Xueyang Fu,Yang Cao,Wei Zhai,Zheng-Jun Zha
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to NeurIPS 2026. Project Page: this https URL
Abstract:Reinforcement learning (RL) has shown strong potential for aligning generative models with human intent, but adapting RL to masked generative models (MGMs) remains largely underexplored. Existing approaches that directly adapt Group Relative Policy Optimization (GRPO) to MGMs leave a foundational conflict unaddressed: MGMs’ exploitative decoding nature is inherently at odds with the exploratory diversity that GRPO presupposes. In this work, we reveal a structural decoding bias termed the visual exploration trap: confidence-based sampling trades the exploration of diverse subject realizations for the greedy resolution of low-uncertainty background regions, causing premature collapse of the generative exploration space and starving GRPO of the rollout diversity it depends on. To this end, we propose MaskSense, a novel RL framework for MGMs that confronts the visual exploration trap at both the sampling and optimization stages. Specifically, we introduce a semantic anchored routing sampling mechanism that leverages semantic priors to preserve high-entropy subject token exploration while amplifying intra-group reward variance through exploratory-exploitative routing to yield more discriminative advantage estimates. Furthermore, we design a global entropy transition anchoring strategy to identify the most consequential decoding steps and concentrate gradient updates on them. Extensive experiments on multiple text-to-image benchmarks demonstrate that MaskSense substantially improves upon the base model, with GenEval accuracy improving from 54% to 86% and HPS from 28.89 to 36.39, achieving state-of-the-art performance.
[CV-175] Self-Supervised Learning of Semantically Consistent Visual Features via a Simple 3D Prior
链接: https://arxiv.org/abs/2610.16988
作者: Sebastian Day,Oisin Mac Aodha
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Visual features extracted from vision models like DINOv2 or Stable Diffusion enable high quality semantic correspondence between images. However, such features exhibit documented limitations that hinder accurate correspondence estimation such as the inability to disambiguate the left from right for categories exhibiting symmetries or repeated parts. Previous attempts to improve these features rely on using supervised data such as annotated keypoint pairs or camera pose information, or make restrictive topological assumptions, such as modelling the object of interest as a 3D sphere. We introduce a new self-supervised approach that allows us to learn features for topologically complex object categories using a simple prior. Instead of relying on an expensive instance-level analysis by synthesis pipeline, our approach makes use of contrastive learning and distribution matching at the global dataset-level to learn the coarse shape and appearance of a category. Combined with additional geometric losses, this enables us to replicate the global and local statistics of visual features which match those that are extracted from real images. Quantitative results are presented across several challenging object categories, demonstrating the effectiveness of our approach.
[CV-176] AUGUstus: Autonomous Driving VLA Model with Unified Generation and Understanding
链接: https://arxiv.org/abs/2610.16985
作者: Yuqi Ye,Caigao Jiang,Zhuonan Wang,Congyuan Pei,Chao Qu,Changhao Peng,Junhong Lin,Qize Yu,Zetian Song,Guangzhao Fu,Chongruo Wu,Shiyu Huang,Shuchang Zhou,Honggou Yang,Wei Gao
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Vision-language-action (VLA) models show strong potential for end-to-end autonomous driving by leveraging world knowledge and understanding to excel in long-tail scenarios. However, current VLAs have two limitations: 1) Degeneration into vision-action (VA) models. For low latency, regression/flow-based methods directly generate actions but underuse VLM understanding, causing unsafe, unexplainable actions. 2) Autoregressive high latency. Autoregressive methods sequentially generate chain-of-thought (CoT) and action tokens, fully leveraging VLM understanding, but high latency hinders deployment. To enable low-latency deployment while using CoT to assist action generation, we propose AUGUstus, an Autonomous-driving VLA model with Unified Generation and Understanding. We unify diffusion-based action generation and autoregressive CoT understanding in one model, where understanding only serves generation. Joint CoT supervision preserves VLM spatial understanding in shared representation for safer, more accurate trajectories. AUGUstus-mode orders action generation before CoT, so CoT supervises reliable actions during training while inference needs only action output, enabling low latency. AUGUstus-GRPO, a unified GRPO algorithm, jointly optimizes action and CoT, improving action performance. Experiments show AUGUstus achieves 94.6 PDMS on NAVSIMv1 and 90.7 EPDMS on NAVSIMv2, SOTA planning without scorer module. On nuScenes, it achieves average L2 error 0.27 m and average collision rate 0.05%, best among compared methods. To validate AUGUstus practicality, we train on in-house data with 512 NVIDIA H800 GPUs for ~2 days, evaluate on an internal 1,000 km dataset, and preliminarily deploy on a real vehicle, achieving 12 Hz operation. Project page: this https URL
[CV-177] he Value of Being Wrong: Self-Mined Visual In-Context Learning from Errors NEURIPS2026
链接: https://arxiv.org/abs/2610.16977
作者: Shu Zhao,Tan Yu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: NeurIPS 2026 Spotlight. 19 pages, 7 figures, 9 tables. Project page: this https URL
Abstract:Visual in-context learning (ICL) enables vision-language models (VLMs) to adapt at test time via demonstration examples, but its effectiveness depends critically on the examples selected. Recent counterfactual selection methods improve ICL by constructing demonstrations that expose how individual attribute changes affect the answer. However, they require multi-stage pipelines involving attribute extraction, caption engineering, and composed image matching. We propose a simpler alternative grounded in a key observation: a model’s own prediction errors are natural counterfactual demonstrations that directly encode its failure modes. Our method, SMILE (Self-Mined Visual In-Context Learning from Errors), builds a model-specific pool of hard negatives offline by recording the target VLM’s own mistakes. At test time, the prediction serves as a direction signal that surfaces, in the semantic embedding space, the past mistakes most likely to recur for the current query. This prediction-conditioned identification provides a unified mechanism for both closed-set classification and open-ended visual question answering without task-specific design. Across four benchmarks and four VLMs, SMILE delivers an average task-score gain of +3.09 points over the SOTA baseline at 5x lower latency. Project page: this https URL
[CV-178] UniLAP: A Unified Latent Action Policy for Cross-Embodiment Manipulation
链接: https://arxiv.org/abs/2610.16969
作者: Lei Yao,Qiao Feng,Mutian Tong,Qihao Yun,Lap-Pui Chau,Lingjie Liu
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Latent Action Models (LAMs) can capture manipulation priors from action-free videos, but discrete latents usually impose a granularity bottleneck, limiting their ability to represent fine-grained action. Recent continuous alternatives mitigate this, but they are typically used as proxies for world modeling and require a further planning step to produce executable actions. We thus introduce UniLAP, a unified continuous latent-action policy that models the latent space via flow matching and bypasses reliance on future frames during inference. By distilling a LAM teacher with a flow-adaptive objective, UniLAP inherits the teacher’s action representation while retaining the multimodal distribution of feasible behaviors. To decouple latent representations from task-irrelevant appearance, we extract a label-free motion target to isolate relevant kinematic cues from spurious visual dynamics. With the established embodiment-agnostic latent manifold, UniLAP can be easily adapted to novel platforms. Pretrained on multi-source videos at a compact 1.1B parameter scale, UniLAP outperforms state-of-the-art latent-action and Vision-Language-Action (VLA) baselines across hand pose forecasting, single-arm and humanoid manipulation, and real-world deployment.
[CV-179] Proactive Physical AI through Continuous Observation and Timely Intervention
链接: https://arxiv.org/abs/2610.16967
作者: Zhenhua Yang,Jiafeng Liang,Xingyu Zhou,Yi Jiang,Xiayu Cao,Zui Chen,Luqi Wang,Haiyang Sun,Yufei Xiang,Chang Li,Rongchuan Mu,Bowen Chi,Zihe Tong,Mingyu Li,Huijie Ma,Yiqiao Yang,Baoqi Ren,Zhiming Ma,Zhihao Zhu,Jiahao Ji,Zhuoyao Wang,Qianyu Wang,Xiangyu Liu,Wang Zixiang,Shunliang Zhang,Ming Liu,Xiaocheng Feng,Duyu Tang,Bing Qin
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: 33 pages, 5 figures, including appendices
Abstract:Physical AI is commonly formulated around an explicitly specified task: given a goal or instruction, a system perceives, reasons, and acts. In many real-world settings, however, important tasks emerge from continuously changing world states before a person explicitly formulates them. We study this complementary problem as Proactive Physical AI, where an AI system continuously observes the physical world, identifies situations that warrant intervention, and initiates appropriate actions at the right time. We develop ProAgent, an end-to-end system built around two components: a Jev-style proactive timing model (dubbed StreamingJev) that determines whether and when intervention is needed, and a Response Model that generates responses for people or invokes available device capabilities for physical execution. To support learning at scale, we develop an automated data pipeline that converts videos into supervision for proactive timing. Experiments demonstrate strong proactive performance under both explicit objectives and standing roles, combining more accurate intervention timing with fewer false triggers than the source-backbone baseline. Real-robot evaluations further demonstrate ProAgent’s ability to identify emerging needs and translate them into appropriate physical actions.
[CV-180] DDVR-Net: Degradation-Aware Decoupling and Visibility-Guided Restoration for Historical Document Recognition ACM-MM2026
链接: https://arxiv.org/abs/2610.16956
作者: Jiu Sha,Yu Weng,Mengxiao Zhu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 18 pages. Accepted to the 34th ACM International Conference on Multimedia (ACM MM 2026)
Abstract:Historical document analysis is crucial for cultural heritage preservation and low-resource language digitization. However, real historical documents often suffer from severe degradation, such as ink erosion and structural damage, causing mixed-visibility character sequences. Conventional OCR systems struggle in such evidence-limited scenarios, while contextual language models may over-correct visually clear characters. To address these challenges, we propose DDVR-Net, a degradation-aware decoupling and visibility-guided restoration framework for historical document recognition. DDVR-Net decouples intrinsic glyph structures from physical corruption at the feature level and predicts local visibility to estimate character reliability. A contextual decoder is selectively activated through a visibility-guided gate, restoring degraded characters while preserving clear visual recognition. Furthermore, we construct a low-resource multilingual benchmark covering Tibetan, Mongolian, and Uyghur scripts with 187000 document pairs, combining large-scale synthetic data and progressive training with real-world annotations to reduce the synthetic-to-real gap. Extensive experiments demonstrate that DDVR-Net consistently outperforms existing methods across three languages and diverse reading orders, providing an effective framework and benchmark for low-resource historical document recognition. Code and Benchmark: this https URL
[CV-181] UniMetric: Universal Metric Scale Recovery for Visual Geometry Foundation Models
链接: https://arxiv.org/abs/2610.16945
作者: Lei Yang,Yuner Zhang,Ziyang Liu,Ziying Song,Kai WU,Yiyang Gao,Jiangtao Su,Yiming Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 39 pages. Technical report. Code: this https URL ; Data: this https URL
Abstract:Recent advances in multimodal foundation models, exemplified by GPT-6 Astra, raise the prospect of general-purpose embodied reasoning and planning. Using Internet videos for embodied pretraining calls for reliable physical-scale annotations. Yet visual geometry foundation models often recover scene structure and camera motion without reliable metric scale. We show that uncertain size priors of familiar objects can jointly constrain a reconstruction’s shared scale. Building on this insight, we present UniMetric, a universal and training-free metric scale recovery framework for geometry foundation models, without additional sensors or task-specific metric supervision. Specifically, Metric Prior Acquisition elicits object height intervals from a vision-language model. Semantic-Geometric Alignment measures reconstructed object heights along the estimated support-surface normal in each frame and converts the priors into scale intervals. Global Scale Recovery reconciles these intervals through a convex objective, yielding a jointly feasible scale when they overlap and a unique compromise otherwise. The estimated sequence-level scale is applied uniformly to reconstructed points and camera translations. Controlled experiments with real VLM priors show that combining observations improves scale accuracy with exact relative object heights. To evaluate metric accuracy on GFM reconstructions without ground-truth scale alignment, we introduce MetricBench, covering twelve embodied, driving, and roadside subsets with sensor-derived ground truth. Across nine geometry foundation models, UniMetric reduces mean depth Abs Rel, reconstruction Chamfer distance, and trajectory error by 54.2%, 83.8%, and 65.6%, respectively, improving every model on all three tasks at the suite level. We release the UniMetric code and MetricBench data.
[CV-182] FOCUS: A Subspecialty Foundation Model for Intraoperative Central Nervous System Pathology
链接: https://arxiv.org/abs/2610.16914
作者: Xing Liu,Yifan Guo,Hongtai Lv,Lei Lou,Yuyi Tan,Jiang Du,Xijia Zhang,Ming Zhang,Chao Pang,Minglei Li,Wei Liu,Xiaobo Zhu,Song Zhang,Rong Yu,Hanlin Shu,Wenzhuo Xu,Tao Jiang,Cheng Lu,Jie Tian,Qing Chang,Zhenyu Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Intraoperative frozen-section diagnosis of central nervous system (CNS) lesions guides surgical decisions but is challenged by artefacts, diagnostic complexity and time constraints. Most pathology foundation models have limited exposure to CNS frozen tissue. Here we present FOCUS, a subspecialty foundation model pretrained on 27,930 frozen-sections enriched for CNS tissue. FOCUS led 11 internal diagnostic and 12 biomarker-prediction tasks, with mean AUCs of 0.969 and 0.877, respectively, retaining leading performance in temporal and external cohorts. To provide more comprehensive diagnostic information, we further organized ten diagnostic tasks into a WHO-aligned diagnostic hierarchy and separately evaluated FOCUS for lymphoma detection, a critical task for intraoperative surgical decision-making. In reader assessments by six neuropathologists, FOCUS assistance enabled junior readers to approach or exceed unassisted senior readers in hierarchical diagnosis and raised junior readers’ lymphoma sensitivity from 81.6% to 94.7%. FOCUS could improve diagnostic accuracy and granularity, inform lymphoma-related resection decisions and provide provisional molecular information. Such capabilities highlight the potential of FOCUS to provide robust, clinically actionable support for intraoperative CNS pathology.
[CV-183] ConDenseDrive: Knowledge-Dense Token-Sparse Visual Encoder for End-to-End Autonomous Driving
链接: https://arxiv.org/abs/2610.16908
作者: Shengzhen Li,Yongkang Li,Hao Gao,Chenxi Wu,Yajing Sun,Lu Hou,Wenyu Liu,Xinggang Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: 17 pages, 8 figures
Abstract:End-to-end driving requires rich spatiotemporal representations to support scene understanding and trajectory planning under strict latency constraints. However, most systems initialize their visual encoders from generic image pretraining and densely process all patch tokens for each camera view and input frame, limiting both representation quality and inference efficiency. To address these challenges, we present ConDenseDrive, a knowledge-dense and token-sparse visual encoder for end-to-end autonomous driving. Our three-stage framework first makes the encoder knowledge-dense by distilling complementary visual, semantic, and geometric knowledge from three vision foundation models. It then adapts the distilled encoder to multi-frame end-to-end driving and integrates spatiotemporal context into register tokens for planning. Finally, it makes the encoder token-sparse by applying register-guided token compression, reducing redundant patch computation and yielding a 2.0 \times speedup while improving planning performance. ConDenseDrive achieves 94.5 PDMS on NAVSIM v1 and 90.8 EPDMS on NAVSIM v2, and reaches 37.4 HD-Score in zero-shot evaluation on HUGSIM. Extensive experiments and qualitative analyses demonstrate that visual encoder design provides an effective route to reliable and efficient end-to-end driving.
[CV-184] Mining Cross-Modal Perception Gaps in Autonomous Driving: When Camera and LiDAR Disagree
链接: https://arxiv.org/abs/2610.16906
作者: Hongsen Cao,Mona Jaber,Shanxin Yuan,Ahmed Sayed
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multi-modal fusion for 3D object detection assumes that cameras and LiDAR are broadly complementary. In practice, how much each modality is trusted at a given location is decided by learned weights that are neither characterized nor auditable. We label every ground-truth object by which modalities detected it, then mine the resulting 0.4M-object database across nuScenes, KITTI, and Waymo. Because the patterns shift with distance from the ego vehicle, we read the mined model through Spatially Stratified Feature Attribution (SSFA), which conditions Shapley attribution on distance zones and exposes structure that global attribution averages away. Three assumptions built into the current fusion design do not hold. The modality-preference boundary tracks the object itself: a traffic cone is nearly six times as likely as a truck to be seen by the camera alone (16.8% vs. 2.9%), regardless of range. Camera-only detections peak at 10-30 m, inside the band usually ceded to LiDAR, and fall away on either side. The drivers of failure are not additive either: the distance - density interaction alone outweighs the entire main effect of viewing direction. We further find that modality dominance is directionally non-stationary, with objects behind the vehicle being markedly more LiDAR-dominant than those in front. We compile the mined structure into a Modality Reliability Map and inject it into four pretrained fusion detectors through a lightweight attention layer. This improves performance by up to 1.3 mean Average Precision (mAP) while fine-tuning under 4% of their parameters in 2-3 GPU-hours. Our work lifts reliability out of the weights and treats it as a quantity that can be examined before deployment. That is what a safety measure for multi-modal perception must rest on.
[CV-185] Barycentric Guidance: Turning Foundation Image Editors into Continuous Affective Controllers NEURIPS2026
链接: https://arxiv.org/abs/2610.16892
作者: Harvey Mannering,Zhiwu Huang,Adam Prügel-Bennett
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: NeurIPS 2026
Abstract:Emotions are expressed on a continuum and are naturally described in the valence-arousal (VA) space. Existing VA-based image editors are typically training-based and domain-specific, limiting them to VA regions seen during training and making them brittle across datasets and domains. Meanwhile, despite producing high-quality edits, state-of-the-art foundation image editors are not continuous affective controllers: non-linear emotion representations in text encoders makes prompt interpolation poorly calibrated, so accurate and continuous VA control remains unreliable. To fill this gap, we introduce Barycentric Guidance, a training-free framework that turns pretrained foundation image editors into continuous affective controllers over VA, requiring no finetuning, extra data, or model changes. Our method maps any target VA point to simplex-based guidance weights, enabling VA controllability in three aspects: (i) accurate targeting of VA states, (ii) broad coverage across the VA plane, and (iii) continuous control. Across three face datasets, we improve VA target accuracy by 25-29%, expand VA-plane coverage by 28-50%, and increase expression diversity by 29-49%, over the strongest baseline. Unlike prior domain-specific methods, our training-free approach exploits the foundation model’s inherent versatility, enabling generalization beyond human faces to animals, artworks, and complex scenes within a single framework. Code is available at this https URL.
[CV-186] EviStress: Evidence Stress Learning for Generalizable Face Forgery Detection with Textual Explanations
链接: https://arxiv.org/abs/2610.16879
作者: Jiuyao Jing,Xuanyu Yi,Yu Zheng,Chunlei Peng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The rapid spread of face forgeries on the Web challenges the credibility of visual content. In practical Web environments, manipulated media originate from evolving generation pipelines and undergo diverse processing during dissemination, making reliable authenticity assessment under unseen manipulations and visual changes a critical challenge. We observe that a standard LoRA-adapted CLIP detector exhibits uneven patch-level decision dependence, concen trating a disproportionate share of class-specific relevance on a small fraction of visual patches. To improve robustness to variations in such highly dependent evidence, we propose EviStress, a two-stage framework for generalizable face forgery detection with textual explanations. In the first stage, Evidence-Targeted Patch Perturbation (ETPP) estimates patch dependence from global, relational, and semantic cues and assigns positively correlated perturbation probabilities, explicitly stress-testing highly dependent visual evidence during training. We further introduce Discriminative Variation Preservation (DVP), which constrains the residual between clean and perturbed representations to retain authenticity discriminative information. In the second stage, Intrinsic Evidence Instruction Tuning (IEIT) transfers the adapted CLIP encoder into LLaVA as its vision tower and performs parameter-efficient multimodal instruction tuning, enabling authenticity prediction and textual explanation directly from the learned forgery-sensitive representation without an external detector. Experiments demonstrate improved cross-dataset and cross-method generalization and corruption robustness, with the multimodal model providing authenticity predictions and human-readable explanations.
[CV-187] Vision Models Capture Complementary Representations in Children That Converge in Adults DATE
链接: https://arxiv.org/abs/2610.16860
作者: Domenic Bersch,Timothy Schaumlöffel,Michela Proietti,Antonia Franaszek-Traczewska,Hannah Elisabeth Zwad,Siying Xie,Marlena Baldauf,Teresa Sylvester,Christina Maria Schätz,Bert Turtleton,Moritz Köster,Stefanie Höhl,Radoslaw Cichy,Gemma Roig
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 4 pages, 1 figure. Updated version of the CCN 2026 accepted abstract. Some analyses have been revised since presentation
Abstract:How the relationship between image-text aligned and vision-only models changes across development remains unknown. We applied time-resolved representational similarity analysis to EEG data from 8-year-olds, 12-year-olds, and adults. CLIP showed significant model-brain alignment already in children, whereas vision-only models aligned more strongly later in development. Residual analyses revealed complementary structure, with CLIP retaining alignment beyond an architecture-matched vision-only model from age 8. Much of the model-brain alignment reflected semantic structure, with EEG representations of category and animacy strengthening toward adulthood, while alignment with low-level image properties remained weak across age. Together, these findings suggest that image-text aligned and vision-only models capture complementary aspects of developing visual representations and may provide useful constraints for AI systems that aim to learn and adapt over time.
[CV-188] owards Cost-Efficient Video Dataset Construction with Large Language Models
链接: https://arxiv.org/abs/2610.16857
作者: Mohammed Shabot,Jonathan Adams,Jan Van den Stock,Joost Vennekens
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 18 pages, 9 tables, 1 figure
Abstract:Research on video classification models has witnessed significant growth recently, accompanied by a significant increase in the demand for datasets to train such models. Existing video dataset curation approaches share a common bottleneck: candidate videos must be downloaded before any relevance judgment is made, which results in storage, time and annotation costs. In this work, we present a filtering step, using Large Language Models (LLMs) to pre-screen videos based on their metadata, before actually downloading these videos, thereby saving on these costs. We evaluate the LLMs against a strict human majority vote derived from four independent annotators and validate the reliability of the annotations through inter-annotator agreement analysis (Gwet’s AC1). We compare the LLM pre-filter against two no-LLM baselines, namely two keyword heuristics. Via a Leave-One-Rater-Out (LOO) mechanism and an Add-One-Rater-In (AORI), we evaluate the performance of the LLMs as virtual annotators. Finally, we report cost comparisons relative to existing approaches. To ensure reproducability, we made our implementation available on github.
[CV-189] A Multi-Task Deep Learning Framework for Pose-Invariant Face Recognition and Facial Landmark Localization
链接: https://arxiv.org/abs/2610.16849
作者: Ali Ashkeyev,Zukhra Abdiakhmetova
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 7 figures, 1 table
Abstract:Face recognition systems often experience performance degradation under large pose variations due to changes in facial appearance and visibility of discriminative features. This study presents a multi-task deep learning framework that jointly addresses pose-invariant face recognition and facial landmark localization within a unified architecture. To enable simultaneous optimization of identity recognition and facial structure learning, the original Cross-Pose Labeled Faces in the Wild (CPLFW) dataset was adapted into a multi-task learning dataset by associating facial images with identity labels and five-point landmark annotations. The proposed framework consists of two task-specific branches operating on a common facial input: a ResNet50-based branch for identity recognition and an EfficientNetV2B0-based branch with spatial attention for facial landmark localization. The two objectives are jointly optimized using a weighted multi-task loss, allowing the model to learn complementary semantic and spatial facial representations. The proposed framework was evaluated on the prepared CPLFW-based multi-task dataset. Unlike the original CPLFW benchmark, which focuses primarily on identity recognition, the proposed dataset adaptation incorporates five-point facial landmark annotations to enable joint recognition and localization learning. Experimental results demonstrate that the model achieves a validation recognition accuracy of 81.58% and a facial landmark localization mean squared error of 96.38. The results indicate that combining identity classification and landmark regression within a single optimization framework provides an effective approach for learning robust facial representations under cross-pose conditions The proposed method demonstrates the potential of multi-task learning and attention-based feature refinement for unconstrained facial analysis. Comments: 10 pages, 7 figures, 1 table Subjects: Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2610.16849 [cs.CV] (or arXiv:2610.16849v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2610.16849 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Ali Ashkeyev [view email] [v1] Mon, 12 Oct 2026 14:23:49 UTC (4,077 KB)
[CV-190] Beyond User-Side Defenses: Image Protection in Open-Weight Diffusion Models
链接: https://arxiv.org/abs/2610.16835
作者: Jihye Kang,Jiwon Park,Namhyuk Ahn
类目: Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Model providers can support image protection, but open-weight deployment introduces practical constraints. We propose four requirements: protection effectiveness, clean fidelity, compatibility, and tamper resistance. We incorporate long-horizon attack simulations to strengthen resistance to protection removal. Experiments with a single protection tool show improved resistance to re-alignment at a cost to clean fidelity, motivating further research toward practical model-side protection.
[CV-191] Unsupervised Data Drift Monitoring for Deployed Deepfake Speech Detectors
链接: https://arxiv.org/abs/2610.16826
作者: Lukáš Prokeš,Vojtěch Staněk,Anton Firc,Kamil Malinka
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Accepted at IEEE International Workshop on Information Forensics and Security (WIFS) 2026
Abstract:Deepfake speech detectors are typically evaluated as binary classifiers on fixed benchmarks, but deployed systems face evolving attacks, codecs, speakers, and recording conditions, with labels unavailable during operation. This makes it difficult to know when a detector operates in familiar conditions and can be relied on. We propose an unsupervised drift monitor that operates on the internal embeddings of a deployed detector. The monitor compares each input against bona fide and spoof references using regularized Mahalanobis distances. These distances define trusted, ambiguous, and novel regions of the detector embedding space, allowing unfamiliar inputs to be routed for review. We evaluate the monitor on four modern detectors on three datasets. The monitor raises alerts for dataset mismatches and substantially reduces the security-critical case of accepting deepfakes at the cost of reduced automatic coverage. The results also show that drift affects both classes: bona fide recordings can move away from the training reference, increasing alert rates even in in-domain settings. Our findings suggest that deployed deepfake speech detectors should be treated not only as a binary classification, but also as a reliability problem.
[CV-192] Growing to See: Efficient Pretraining of Self-Supervised Vision Transformers
链接: https://arxiv.org/abs/2610.16817
作者: Amir Mohammad Karimi Mamaghan,Ferdinand Kapl,Andrea Dittadi,Karl Henrik Johansson,Kaitlin Maile,Johannes von Oswald,Stefan Bauer
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Progressive depth growth builds large networks by starting with shallow models and inserting layers during training. Recent advances in large language models have renewed interest in this longstanding approach, demonstrating benefits for training efficiency and reasoning alongside changes in internal computations. However, how growing affects representation quality and computational patterns in self-supervised vision transformers remains unclear. In this work, we systematically investigate progressive depth growth in two popular self-supervised families, namely masked autoencoders (MAE) and DINOv3, and analyze how growing affects downstream performance, intermediate representations, and internal computations. Across global and dense tasks, we show that growing improves downstream performance while reducing training FLOPs by 13.9–24.8%, with ImageNet linear-probe accuracy gains of up to 3.36 percentage points. Growing also improves downstream performance at intermediate layers and reshapes internal computational patterns, introducing recurring alternations between global and localized attention in MAE and earlier local-to-global attention progression in self-distilled DINOv3. Together, these findings establish progressive depth growth as an effective and efficient pretraining strategy for improving downstream performance across global and dense visual tasks with less computation, and advance our understanding of how growing shapes visual computation.
[CV-193] Carve VFM: Carving Specialists out of Generalists Automatic Task-Specific Compression of Vision Foundation Models
链接: https://arxiv.org/abs/2610.16813
作者: Lotfi Abdelkrim Mecharbat,Abdelrahman Elsayed,Ibrahim Almakky,Martin Takáč,Mohammad Yaqub
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 24 pages, 6 figures, 11 tables
Abstract:Vision foundation models (VFMs) provide transferable visual representations, but their inference cost limits deployment in many settings. Fine-tuning specializes a VFM without shrinking it; distillation shrinks it, but through expensive teacher-guided training. Structured pruning offers a more direct route by extracting a compact subnetwork from the VFM. However, as compression gets aggressive, existing methods fail because they treat subnetwork selection and recovery as separate axes. Across layers, capacity must follow task sensitivity rather than spread uniformly. Within a layer, a removal cannot be judged by importance alone; its impact also depends on how well the lost contribution can be recovered from what remains. Selection and recovery must therefore be decided jointly. We introduce Carve VFM, which resolves this interdependence. Given a VFM, a calibration set, and a parameter budget, it (i) searches for task-specific capacity allocations across layers; (ii) guides the search via an adaptive recovery-aware fitness criterion; and (iii) reconstructs removed behavior through iterative closed-form low-rank corrections merged into the surviving weights. The result is a dense encoder produced in minutes on a single GPU without backbone retraining. Across twelve benchmarks and four models from DINOv3 and MetaCLIP-2, Carve VFM achieves the highest reported mean accuracy in seven of eight backbone-budget configurations among the evaluated dense structured-pruning methods. Further fine-tuning surpasses the strongest evaluated edge model (88.1% vs. 85.8%), and carved models cut compute and memory while delivering 1.3-2.0x higher GPU throughput than the uncompressed one. These results show that the compact model you need is inside the VFM; it only has to be carved out.
[CV-194] X-Rag ent: Tool-Augmented Reinforcement Learning for Evidence-Grounded Chest X-Ray Reasoning
链接: https://arxiv.org/abs/2610.16811
作者: Yi Wang,Zhenlong Yuan,Yuchan Guo,Maolei Huang,Zijian Song,Dapeng Zhang,Rongbin Tan,Jiyuan Wang,Zhiyuan Wang,Fangfang Lin,Shuhan Song,Huawei Cao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Chest X-ray (CXR) interpretation often requires integrating visual findings with task-specific evidence rather than relying on direct visual recognition alone. While tool-augmented medical agents can supply such evidence via external modules, they often lack explicit mechanisms for learning when auxiliary evidence is needed and for preventing inapplicable quantitative evidence from entering diagnostic reasoning. To address this, we propose X-Ragent, a post-training framework for tool-augmented reinforcement learning that decouples learned auxiliary-evidence seeking from validity-gated quantitative measurement acquisition for evidence-grounded CXR reasoning. We construct Rad-CoT, a 7.7K-sample corpus combining causally ordered tool-augmented trajectories with report-style diagnostic reasoning, and use it to initialize Qwen3-VL-8B via agentic supervised fine-tuning (SFT). We then optimize the policy with Group Relative Policy Optimization (GRPO) using an accuracy-gated reward that applies process-level terms for measurement consistency and auxiliary-tool cost only to correct trajectories, preventing plausible process behavior from compensating for an incorrect diagnosis. Across three CXR benchmarks, X-Ragent achieves the strongest results, including an 8.6-point improvement on the fully held-out ChestAgentBench. By seeking evidence selectively and admitting only valid measurements, X-Ragent supports more faithful and evidence-grounded CXR reasoning.
[CV-195] ViewStruct: View-Grounded Articulation and Connectivity Inference
链接: https://arxiv.org/abs/2610.16793
作者: Wenjing Bian,Ranran Huang,Andreas Geiger
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Interacting with unfamiliar objects requires understanding how their parts connect and move. We introduce ViewStruct, a framework for jointly predicting connectivity and articulation in objects and scenes. Given an RGB image with aligned part segmentation and depth, optionally supplemented by images or video, ViewStruct predicts rigid-body groupings, attachments, and joint parameters. Its view-grounded representation links predictions to observed regions and expresses joint geometry in a gravity-aligned, view-relative frame, without requiring category-specific canonical orientations. A pretrained vision–language model is adapted to predict structure in seconds, without requiring observed articulation motion or per-instance optimisation. Experiments demonstrate improved structural prediction over evaluated baselines and transfer to unseen categories, asset sources, and real-world human-interaction videos, despite task-specific training exclusively on synthetic data. Simulated manipulation experiments further demonstrate the utility of predicted joints for generating robot motion trajectories.
[CV-196] InfoVis: Unified Slide Image Generation and Editing
链接: https://arxiv.org/abs/2610.16784
作者: Jiahao Wang,Xiyan Jiang,Chenyuan Wu,Haoqiang Zhang,Xin Luo,Yueze Wang,Defu Lian,Bo Zhang,Jiajun Zhang,Zheng Liu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Code: this https URL
Abstract:Generating and editing presentation slides requires models to combine accurate text rendering, instruction following, visual grounding, aesthetic design, and precise editing within one canvas. Existing work covers only subsets of these capabilities and lacks large-scale slide-native supervision and diagnostic evaluation. We formulate four slide-centric tasks and build a scalable data engine with over 13M continual-pretraining records and 1.9M supervised fine-tuning records. We also introduce VisSlideBench, a benchmark of 1,527 manually screened instances that covers all four tasks with task-specific, multidimensional evaluation. Its automatic rankings agree with blinded, dimension-wise human rankings. Using these resources, we train InfoVis, a unified 9B slide generation and editing model. On VisSlideBench, InfoVis leads every evaluated open-source model on all eleven metrics and scores 7.94 overall against 7.99 for Nano-Banana-2. Starting from FLUX.2-klein-base-9B, slide-specific post-training adds 1.6 points overall, with the largest generation gains in content faithfulness; the remaining gap to proprietary models is smallest in content faithfulness and largest in aesthetic quality. Built on InfoVis, our document-to-slide workflow achieves the second-highest Aesthetics score on the external SlidesGen-Bench. We will release the model, code, data, and benchmark.
[CV-197] owards Complementary Keypoint Detection via Mixture of Detectors NEURIPS2026
链接: https://arxiv.org/abs/2610.16762
作者: Xiaoyong Lu,Guobao Xiao,Dong Liang,Songlin Du
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: NeurIPS 2026
Abstract:Keypoint detection plays a central role in geometric vision pipelines. Existing detectors exhibit complementary strengths across different scene structures and keypoint budgets, and reliance on a single detector often results in suboptimal keypoint distributions. Motivated by this observation, we propose \textbfMixture of Detectors (MoD) for dense matching pipelines, a unified framework for complementary keypoint detection through the adaptive composition of heterogeneous detectors. MoD first distills heterogeneous detectors into a unified model, allowing diverse detection maps to be predicted efficiently within a single forward pass. Detector selection is subsequently formulated as a patch-wise, budget-conditioned decision process, in which a lightweight router assigns the most suitable detector to each local region. To avoid heuristic supervision, the router is trained with task-driven rewards derived from downstream matching quality, jointly capturing both keypoint quantity and precision. Extensive experiments on MegaDepth, ScanNet, and HPatches demonstrate that MoD consistently outperforms strong baselines across a wide range of keypoint budgets, demonstrating its effectiveness in improving keypoint distribution and downstream geometric performance.
[CV-198] Image Tokenizers Should Be Trained Between Images
链接: https://arxiv.org/abs/2610.16759
作者: Ole Beißwenger,Jan-Niklas Dihlmann,Hendrik Lensch
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Latent diffusion models generate images as codes of an image tokenizer, whose decoder turns them into the pixel image. Recent tokenizers improve how well these codes reconstruct an image and how easily a generator learns them. Yet a generator does not reproduce the codes of training images, and when a generated code lands outside what the decoder expects, the decoded image contains imperfections. Current methods make the decoder more robust to generated codes by adding noise during training. However, generated codes lie between the training codes and resemble their mixture more than one of them plus noise. To improve generated images, decoders should therefore be trained on codes in the space between training samples. We propose InterpTok, which fine-tunes MAETok to decode mixtures of image codes into natural images. Since no target image exists between two training samples, a discriminator judges whether each decoded mixture looks as natural as its two sources. A mixing ratio loss requires the decoded mixture to keep the mixing ratio of the two codes. Together, they prevent collapse to a trivial solution and yield a sharp and natural composition of both images. InterpTok keeps MAETok’s reconstruction quality and improves interpolation quality more than threefold, the best of all tokenizers in a recent benchmark. Beyond interpolation, InterpTok improves the generation FID of a matched diffusion transformer from 11.61 to 8.48 without guidance, and even reaches 8.11 at an eighth of the sampling steps, where the generator on MAETok’s codes worsens. The generator trained on our tokenizer also reaches the final result of the one trained on MAETok at half the training steps. Our findings show that the space between image codes can and should be explicitly supervised rather than ignored.
[CV-199] he Midas Touch for Metric Depth: Interpretable Metricization for Generalizable Depth Perception
链接: https://arxiv.org/abs/2610.16749
作者: Mengtan Zhang,Yu Ma,Shiqi Jiang,Yi Feng,Wei Ye,Xieyuanli Chen,Hanli Wang,Qijun Chen,Rui Fan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 18 pages, 7 figures
Abstract:Recent foundation models have drastically improved the generalization of monocular relative depth estimation, yet recovering accurate metric depth from their estimates for depth perception tasks remains challenging due to scale ambiguity and local metric-scale inconsistency. Existing approaches recover metric depth from sparse observations using either explicit global affine transformation algorithms or implicit data-driven depth completion networks. The former cannot accommodate spatially varying relative-to-metric relationships, while the latter are sensitive to the distribution of sparse observations and may not fully exploit generalizable relative depth priors. To simultaneously address these issues, this study extends our previous work MTD and proposes MTDv2, a hybrid metric depth perception framework that combines explicit depth metricization with data-driven metric depth refinement. The explicit depth metricization process performs segment-wise affine transformations to recover spatially varying metric scales while preserving the geometric structure of relative depth, followed by pixel-wise depth refinement that propagates metric observations along paths with minimal accumulated geometric variation. The data-driven depth refinement process subsequently extends the explicit single-path depth propagation to a learnable multi-path formulation while preserving its interpretability. Extensive experiments conducted on 16 public datasets demonstrate the robust and generalizable depth scale recovery capability of MTDv2 across diverse sources of sparse observations and different depth estimation paradigms, as well as its broad applicability to downstream 3D perception tasks. Code will be publicly available at this https URL upon publication.
[CV-200] FoveaTree: Information-Density Trees for Query-Guided Visual Token Pruning in Video Large Language Models
链接: https://arxiv.org/abs/2610.16744
作者: Zihan Meng,Weidong Chen,Dexiang Hong,Xinyan Liu,Zhendong Mao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Video large language models (Video LLMs) encode each frame into hundreds of visual tokens, making video inference expensive. Existing methods reduce this cost through similarity-based merging or attention-based selection but face two challenges: Hierarchy-blind Pruning, which overlooks regional compressibility and can discard critical evidence, and Attention Drift, where positional effects and attention sinks shift text-to-vision attention away from task relevance. We propose FoveaTree, a two-stage, training-free visual-token pruning framework. First, Merge-Gain Density Estimation measures the feature distortion caused by regional merging and constructs a quadtree-derived density map to guide token dropping, merging, and retention. Second, Attention Tree Pruning uses the surviving units and inherited tree topology to refine text-to-vision attention in intermediate LLM layers, preserving relevant regional evidence while removing low-relevance content. We evaluate FoveaTree on InternVL3.5-8B and Qwen3-VL-8B across five video question-answering benchmarks and one captioning benchmark at 50%, 30%, and 10% visual-token retention. FoveaTree consistently achieves overall state-of-the-art performance across the evaluated settings, offering a strong accuracy-efficiency trade-off.
[CV-201] VAttnPO: Visual Attention-Guided Policy Optimization
链接: https://arxiv.org/abs/2610.16735
作者: Lihao Zheng,Zhenwei Shao,Zhou Yu,Yan Yang,Haoyi Sun,Qian Wang,Ning Mao,Pan Zhou
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Multimodal reinforcement learning with verifiable rewards commonly assigns one sequence advantage to an entire response. This outcome-level signal distinguishes overall response quality but leaves changes in visual-evidence use across generated tokens implicit. We introduce VAttnPO (Visual Attention-Guided Policy Optimization), which uses answer-conditioned visual attention to refine token-level credit while preserving the verifier-derived trajectory-mean advantage. A frozen old policy constructs a visual reference by teacher forcing the known correct answer and produces a visual distribution for each rollout token. VAttnPO measures their agreement in visual-attention space using Jensen–Shannon divergence, rather than reading guidance from teacher output-token distributions or introducing an attention-matching objective. Trajectory centering, relative-position weighting, and credit balancing turn this agreement into a zero-mean, position-aware correction to the GRPO sequence advantage. Token intervention, visual-index permutation, and trajectory analysis provide diagnostic evidence linking this signal to model behavior, visual correspondence, and generation position. On Qwen3-VL-8B-Instruct, VAttnPO raises six-task image AVE from 70.97 with GRPO to 73.50 and scores higher on every reported structured, video, and MME metric. Gains over GRPO also hold on Qwen2.5-VL-7B-Instruct and InternVL3-8B-Instruct.
[CV-202] PathPilot: An Adaptive Question-Conditioned Multi-Magnification Navigation Agent for Whole-Slide Image Analysis
链接: https://arxiv.org/abs/2610.16734
作者: Zhengyang Xu,Han Li,Jingsong Liu,Ayako Ura,Shihui Zu,Shaohua Kevin Zhou,Nassir Navab,Peter Schüffler
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The diagnostic capability of WSI agents relies on effective slide navigation to identify diagnostically relevant regions and efficiently integrate evidence across the slide. While in clinical practice, pathologists dynamically examine slides across multiple scales, integrating global and cellular evidence in a sequential manner, existing AI navigation approaches mostly operate at a single fixed magnification or rely on predefined magnification traversal. This mismatch prevents existing methods from modeling cross-magnification interactions and adaptive magnification selection inherent to real diagnostic workflows. We propose PathPilot, a clinically consistent multi-magnification WSI navigation agent that explicitly models cross-magnification localization and evidence-adaptive exploration. Specifically, we introduce a CrossMagLoc that learns magnification-specific diagnostic localization from pathologist viewing behavior and integrates information from adjacent magnifications to identify the diagnostically most relevant region within the current ROI, thereby providing a diagnostically informed starting point for subsequent question-conditioned exploration. We further introduce an EviRegionSearch that starts question-conditioned exploration from this region, progressively accumulating question-relevant evidence and adaptively determining the next navigation action according to the remaining evidence requirement. Their iterative interaction enables PathPilot to sequentially explore across ROIs and magnifications until sufficient evidence is acquired. Extensive QA evaluations across five cohorts demonstrate consistent superiority over competing SOTA methods on inter/external cohorts, respectively, validating question-conditioned evidence acquisition across regions and magnifications that mirrors the sequential, knowledge-driven navigation of pathologists.
[CV-203] 3D-VITAL: Visibility-aware Identity Training from Artificial Liftings of 3D Reconstruction model for Cross-View Object Re-Identification NEURIPS2026
链接: https://arxiv.org/abs/2610.16692
作者: William Grolleau,Astrid Sabourin,Guillaume Lapouge,Catherine Achard
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to NeurIPS 2026 (poster)
Abstract:We address the Cross-View Object Re-Identification (ReID) task, aiming to match identities across drastically different viewpoints. We focus on Single-View Aerial-Ground ReID (SV AG-ReID), where models train on one viewpoint and must generalize to another at inference. To bridge this viewpoint gap, related work employs generative models to synthesize novel views from source images as additional training data. We leverage recent advances in large-scale single-view 3D mesh reconstruction to lift source images into textured meshes and to render them from new viewpoints. However, this pipeline produces unreliable data in two distinct ways. First, the reconstruction may fail and produce geometrically incorrect meshes, resulting in severe corruption in rendered novel views. Second, even when the geometry is correct, the rendered novel views contain hallucinated textures in regions not visible in the source image. The reconstructed mesh carries a per-pixel partition between regions visible from the source view and regions that the reconstruction model had to hallucinate, available as a deterministic byproduct of rendering. We introduce this per-pixel visibility map and exploit it in a two-stage design. Cross-View Consistency Filtering (CVCF) discards geometrically incorrect meshes by ensuring multi-view consistency between source and novel views over visible patches. Visibility-Aware Attention Supervision (VAAS) leverages the visibility map to guide the model’s attention toward reliable regions of the rendered view during training. Combined, these two components enable our method, 3D-VITAL, to better leverage the synthetic signal from single-view 3D mesh reconstruction. 3D-VITAL sets a new state-of-the-art on three SV AG-ReID benchmarks. Code and data will be released at this https URL.
[CV-204] EvoSpatial: Experience-Driven Skill and Tool Evolution for Spatial Reasoning Agents
链接: https://arxiv.org/abs/2610.16679
作者: Bin Yu,Shijie Lian,Zhaolong Shen,Xiaopeng Lin,Yao Zhang,Haishan Liu,Haotian Li,Yuliang Wei,Kai Hu,Yichao Du,Zhirui Zhang,Bailing Wang,Kai Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Tool-augmented vision-language agents compose perception and measurement to reason about spatial relationships, but effective execution also requires tracking geometric assumptions and checking intermediate evidence. Preserving this experience across tasks requires retaining procedural knowledge together with the computations it governs. We introduce EvoSpatial, an experience-driven framework that evolves a library of linked skills and tools. Our key insight is that successful and failed spatial trajectories reveal both how to compute and when to trust the result. EvoSpatial retrieves skills, reuses tools, and synthesizes reusable computational functions (generated helpers) during reasoning. After answer submission, supervised outcome feedback guides skill refinement and tool consolidation for subsequent tasks. Across 20 benchmarks, EvoSpatial attains the highest average score among the compared spatial agents, exceeding SpatialClaw by 4.0 percentage points. Further experiments show cross-benchmark transfer and improved reasoning through skill-guided distillation without external skills at evaluation. These results support accumulating spatial experience for both inference-time reuse and trajectory-based training.
[CV-205] Revisiting Natural Language Prompts from Large Language Models for Open-Vocabulary Video Recognition
链接: https://arxiv.org/abs/2610.16666
作者: Haichen He,Weibin Liu,Weiwei Xing
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-language models (VLMs) like CLIP have shown great success in image tasks, however, applying CLIP to open-vocabulary video recognition poses challenges. Due to the limited text information, directly fine-tuning CLIP on video recognition datasets may lead to overfitting and hinder its generalizability. A prevalent method is to enrich text information by generating prompts using Large Language Models (LLMs). However, limited efforts have been made to explore how to generate higher-quality prompts that can contribute to classify unseen categories, and how to further enhance vision-text alignment based on these prompts. In this paper, we propose P2CLIP (Prompt-guided CLIP), which aims to bridge this gap by exploring the impact of prompts across three aspects: category, video, and frame. At the category level, we argue that LLM-generated prompts under simple instructions may suffer from semantic overgeneralization, which can blur the category boundaries and hinder the performance of the model. To address this issue, we design Generic-Discriminative Prompt using LLMs to obtain discriminative category embeddings with rich semantic information. At the video and frame levels, we first use VLM and LLM to generate frame and video descriptions for enhancing vision-text alignment. Then we propose a Knowledge-Guided Temporal Aggregation mechanism, which leverages the expert knowledge derived from LLMs to guide the temporal modeling of videos and encourage a better construction of the joint visual-textual feature space. Extensive experiments across various benchmarks, including zero-shot and few-shot settings, demonstrate the effectiveness of our method.
[CV-206] Language2Drive: Do Vision-Language-Action Models Really Follow the Instructions in Autonomous Driving?
链接: https://arxiv.org/abs/2610.16634
作者: Zihan You,Zehua Fan,Wenqi Lyu,Aoqi Wang,Ni Wang,Yifei Yang,Linge Zhao,Huan Li,Sining Ang,Yuguang Yang,Jueyu Chen,Xiangyu Li,Qianshan Wei,Jieyuan Pei,Yuxin Wang,Zhiwen Chen,Jiaqi Li,Wenxuan Song,Yanyuan Qiao,Benjin Zhu,Haoang Li,Yan Wang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Recent Vision-Language-Action (VLA) models increasingly use natural-language instructions to condition autonomous driving behavior. Yet existing closed-loop benchmarks primarily measure route completion, safety, and rule compliance, leaving a fundamental question unresolved: does language actually determine which feasible behavior the vehicle executes? A trajectory may be safe and successful while still violating or even ignoring the passenger’s instruction. To address this, we introduce \ltd, a benchmark designed to directly test this capability. Its key evaluation unit, the paired counterfactual unit (PCU), holds the initial driving scene fixed while changing only the instruction. This design enables us to distinguish isolated instruction success from counterfactual paired success across four instruction-following scopes. To test whether this distinction persists across evaluation regimes, we instantiate Language2Drive in complementary closed-loop CARLA simulation, reconstructed real-world scenes, and open-loop driving benchmarks, including our released CARLA-Language2Drive simulator with purpose-built maps. Evaluating representative driving VLAs across these settings reveals a striking mismatch between strong driving competence and weak language control: 93.2% of successful rollouts do not have a successful counterpart under the alternative feasible instruction. This failure is dominated by one-sided completion, where a model repeatedly executes one preferred behavior but fails when the same scene requires the counterfactual alternative. These results expose two fundamental limitations of current evaluation: isolated success and driving scores, which are widely used by conventional driving benchmarks, can substantially overstate language-following capability, while current driving VLAs themselves lack reliable instruction following under matched counterfactual tests.
[CV-207] REEL: Learning to Retrieve Evaluate and Evolve Visual Memory for Any Long-Horizon Multi-Shot Video Generation
链接: https://arxiv.org/abs/2610.16632
作者: Hang Xu,Guohui Zhang,Xiaoxiao Ma,Hu Yu,Jie Huang,Jiayin Xiao,Lin Song,Haoyang Huang,Nan Duan,Feng Zhao
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: this https URL
Abstract:Memory-based long-video generation aims to combine user control with long-range visual consistency. We organize this process into three functional components: a planner, an adapter, and a generator. Representative multi-shot systems couple memory to generator-specific interfaces or use predefined agent workflows, rather than independently training a memory policy for cross-shot objectives. We introduce REEL, a general-purpose visual-memory planner. Through a unified tool interface, REEL learns to retrieve references, create missing visual evidence, and update persistent memory. Backend-specific adapters translate its evidence plans into supported conditioning inputs, allowing the same policy to guide different black-box video generators. To train REEL, we construct two high-quality datasets, REEL-SFT-4k and REEL-RL-1k, covering diverse interaction patterns and prompt types. Training proceeds in three stages: supervised fine-tuning, single-shot Group Relative Policy Optimization (GRPO), and multi-shot GRPO. Multi-shot GRPO carries policy-generated memory across consecutive shots, so later outcomes guide earlier reference selection and memory updates. REEL-Bench tests identity preservation and intended state changes under text-only requests and timed reference-image uploads. On JoyAI-Echo~1.5 and MiniMax-H3, REEL improves cross-shot consistency on ST-Bench by 15.1% and 20.7%, and Core-100 on our proposed REEL-Bench by 2.41 and 8.32 points, respectively, over the no-memory baseline.
[CV-208] Adaptive Interaction Network for Human Motion Prediction During Human-Robot Collaboration
链接: https://arxiv.org/abs/2610.16624
作者: Mengyuan Liu,Yangting Lin,Qiongjie Cui
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 12 pages, 10 figures, 10 tables. Author accepted manuscript. Published in IEEE Transactions on Image Processing. Code: this https URL
Abstract:Human motion prediction during human-robot collaboration is critical for achieving safe and efficient interactions in shared environments. Unlike previous studies that have primarily focused on humans and objects, we focus on the robot-aware human motion prediction task, which explicitly models the influence of robots on human motion. This task presents new challenges, such as capturing human-robot heterogeneity and modeling complex spatial interactions. To address these issues, we develop an adaptive interaction network (AINet) model that consists of two branches: a primary branch that predicts future human motion and an auxiliary branch that estimates robot trajectories. The two branches are jointly optimized and coupled via a Local-Global Spatial Interaction (LGSI) Module, which effectively captures fine-grained and global contextual dependencies between human and robot motion sequences. Then, we introduce an Adaptive Weighted Aggregation (AWA) Module to dynamically fuse motion features using context-dependent weights, thereby increasing adaptability across diverse scenarios. Furthermore, we adopt a coarse-to-fine prediction strategy, in which a coarse pseudo-trajectory is first predicted, followed by the refinement of detailed human poses based on that trajectory. Extensive experiments on three challenging datasets reveal that our proposed method achieves superior performance, validating its effectiveness in human-robot collaboration scenarios. Our code is available at: this https URL.
[CV-209] Answer It or Not? Towards Prompt-Adapted Adversarial Images on Vision Language Models
链接: https://arxiv.org/abs/2610.16622
作者: Shunyuan Bai,Yuchen Ren,Le Yang,Chenhao Lin,Chao Shen,Zhengyu Zhao
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 20 pages, 7 figures
Abstract:With the advancement in language modeling, research on adversarial images has recently shifted from image classifiers to vision-language models (VLMs). Despite the new, text-prompt-driven nature of VLMs, we find that existing VLM adversaries have largely overlooked the dynamics of text prompts. In a typical use case, adversarial images can protect visual privacy by rejecting VLMs from answering questions about the image. However, existing protections are prompt-agnostic, i.e., indiscriminately rejecting any question, including those for potential legitimate use of the image. To address this limitation, we propose Prompt-Adapted (ProAda) adversarial images, which are optimized to balance the rejection of harmful prompts and the correct response of legitimate prompts. In particular, to achieve a more accurate adversarial direction, the text prompt is also iteratively updated with a simple classifier judging whether the harmfulness of the prompt remains. Extensive experiments on visual-privacy-related datasets validate that ProAda can reject over 90% of harmful prompts while allowing over 90% of legitimate ones, demonstrating superior overall performance compared to existing prompt-agnostic VLM adversaries. In addition, our ProAda achieves competitive overall cross-model transferability and improved robustness against countermeasures.
[CV-210] Detecting Chain-of-Thought Hijacking in Backdoored Multimodal Large Reasoning Models
链接: https://arxiv.org/abs/2610.16598
作者: Zhifang Zhang,Qi Wei,Weitong Chen,Olaf Maennel,Wei Emma Zhang,zuomin ren,Lei Feng,Miao Xu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Chain-of-thought (CoT) reasoning enables multimodal large reasoning models (MLRMs) to solve complex visual problems, but also exposes them to CoT hijacking: an adversary implants a backdoor through poisoned fine-tuning so that a visual trigger induces a hijacking turn, redirecting an otherwise faithful reasoning chain to an attacker-chosen answer. However, since the hijack acts within the reasoning chain, existing multimodal backdoor defenses designed for MLLMs that answer directly struggle to detect it. Moreover, inspecting the chain itself does not reliably expose the hijack, since the hijacking turn can resemble honest self-correction in form. In this paper, we find that honest reasoning tends to revise its answer at high-entropy “aha moments”, whereas a hijack can overturn a confident answer through a near-deterministic turn. To quantify this contrast, we formalize this phenomenon as belief inertia, the cumulative next-token entropy of a reasoning span minus the answer-belief revision it induces, with markedly lower values for hijacked spans than for honest reasoning spans. Since hijacking tokens are also image-dependent, we further improve detection by restricting the measure to image-dependent spans, yielding visual belief inertia (VBI). Building on this, we propose InertiaGuard, a training-free, test-time detector that flags a trace when its VBI falls below a threshold calibrated on a small set of clean traces. Across various attacks and benchmarks, InertiaGuard achieves higher detection performance than existing baselines, while generalizing to different model architectures and trigger types.
[CV-211] JtT: Rethinking Point Trajectory Diffusion for Diverse and Rigid Motion from an Image
链接: https://arxiv.org/abs/2610.16591
作者: Gabrijel Boduljak,Yushi Lan,Christian Rupprecht,Andrea Vedaldi
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:We study forecasting rigid-body motion from a single image: generating diverse and physically plausible future trajectories for arbitrary points in a scene, without building rigidity into the architecture or relying on geometric inputs such as depth. Like video generators, recent trajectory generators fail at this task, even in simple synthetic Kubric scenes, despite scaling up model size and training on real data. We show how to solve this with the right architecture and parameterization. Our model, JtT, is a simple diffusion transformer that operates directly in coordinate space and decodes arbitrary and dense query patterns, without the trajectory VAEs or fixed grids assumed in prior work. Through a systematic study of the design space of point trajectory diffusion, we find that diffusing per-frame velocities is the only parameterization that captures rigidity, that it must be paired with the right timestep schedule, and that adding full-resolution RGB at the query points resolves the boundary failures of prior methods. However, evaluating such 2D models rigorously is itself non-trivial. To our knowledge, we are the first to observe that physical plausibility and diversity do not improve together during training: larger models, as used in prior work, memorize the single ground-truth future per training image, becoming more rigid while covering fewer possible futures. Yet prior work has paid little attention to this, so we show how model rankings depend on model size, metric, and stopping time, and adopt a stricter, checkpoint-averaged evaluation protocol. Under this protocol, JtT improves over prior methods on both distributional and rigidity metrics and approaches ground-truth rigidity, with only about 50M parameters. Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) Cite as: arXiv:2610.16591 [cs.CV] (or arXiv:2610.16591v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2610.16591 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Gabrijel Boduljak [view email] [v1] Mon, 12 Oct 2026 11:13:37 UTC (11,925 KB)
[CV-212] Code-Interactive Visual Reasoning for Financial Candlestick Understanding
链接: https://arxiv.org/abs/2610.16587
作者: Chenxi Zhang,Youwei Pang,Xiaoqi Zhao,Ziliang Gan,Rongjunchen Zhang,Shanshan Feng,Hao Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Financial candlestick understanding requires locating visual elements, aligning them across trading dates and panels, and combining observations through numerical and financial reasoning. Answer accuracy alone does not reveal whether predictions are supported by chart evidence. We introduce Code-Interactive Financial Candlestick Reasoning and present Fin-Thinker, a model based on Qwen3.5-9B that generates and executes code to acquire evidence and perform computations. The model uses the returned observations to guide subsequent reasoning. To train these capabilities, we construct FinKLine-32K with 32,184 replay-verified code-observation trajectories collected through teacher-guided student exploration. These include successful solutions with and without error recovery. Our three-stage training framework combines foundational visual-code fine-tuning, financial process supervision of successful reasoning and corrective continuations, and execution-feedback reinforcement learning. For evaluation, we introduce FinKLineBench with 1,200 questions across 138 financial task types to assess answer correctness, code execution, and evidence grounding separately. On FinKLineBench, Fin-Thinker improves strict accuracy by 13.1 and 33.2 percentage points over the best-performing direct-answer baseline among those evaluated and its base model with sandbox access, respectively. Training-stage ablations further show that financial process supervision and execution-feedback reinforcement learning yield successive improvements in both answer accuracy and GPT-5.6-judged grounded accuracy.
[CV-213] CineRecon: Benchmarking Frontier Models for Agent ic Cinematic Reconstruction
链接: https://arxiv.org/abs/2610.16577
作者: Jialei Wang,Zehan Wang,XiaoYang Wu,Ziang Zhang,Siyu Chen,Jiaer Huang,Yujia Zhang,Tianyu Pang,Hengshuang Zhao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 24 pages, 19 figures. Project page: this https URL
Abstract:Reconstructing reference videos as editable dynamic 3D scenes is an important path toward controllable video generation for filmmaking: these scenes provide modifiable, renderable visual proxies that enable explicit control over scene layout, subject actions, and camera behavior. To evaluate agents’ ability to reconstruct editable dynamic 3D scenes from dynamic reference videos, we introduce CineRecon, a benchmark comprising 50 reference videos and 360 annotated evaluation clips. The reference videos span film, real-world footage, and games. Given a complete video, an agent uses tools to construct, animate, render, and refine its reconstruction. We evaluate this task under three representation settings: Proxy, Simplified, and Detailed. These settings increase geometric and appearance requirements while retaining the same requirements for spatial relations, camera behavior, and subject dynamics. This design tests whether agents can reproduce the events in a video when detailed appearance is unnecessary, and how performance changes as representation requirements increase. For evaluation, we construct checklists that associate observable properties and events with specific entities and times. A vision-language judge compares timestamp-aligned reference and rendered frames across 11 categories, grouped into camera, static, and dynamic consistency dimensions. We evaluate 12 models using their corresponding agent harnesses and find failures across all three dimensions. Subject motions and interactions remain difficult to reproduce even under the Proxy setting.
[CV-214] Dense Endoluminal Mapping and Localisation from Monocular Bronchoscopy
链接: https://arxiv.org/abs/2610.16570
作者: Salvatore Esposito,Matias Mattamala,Daniel Rebain,Liang Zhao,Kev Dhaliwal,Mohsen Khadem,Subramanian Ramamoorthy
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Bronchoscopy inserts a camera through the airways to diagnose endoluminal disease. Locating the camera along the airway usually requires additional equipment, such as an electromagnetic tracker and a preoperative CT scan. In this paper, we investigate the use of the monocular camera stream to localise the bronchoscope without requiring additional sensing. We introduce a monocular SLAM system that localises the tip of a bronchoscope and reconstructs the airway surface around it from RGB video alone. A Depth Anything 3 backbone, fine-tuned on over a million photorealistic airway renderings, returns metric per-pixel depth and is the system’s only source of scale. DINOv3 features routed through a custom cross-attention matcher supply sub-pixel correspondences on mucosa where classical descriptors fail. Both enter a factor graph alongside a dense constraint from projective iterative closest point (ICP) registration against the system’s own running reconstruction, so surface evidence stabilises the directions the visual factors leave weakly observable. On six airways rendered from patient CT scans, the system reaches 26.1 mm mean absolute trajectory error and 77.5% mean surface F-score. Its mean surface accuracy exceeds that of five recent monocular SLAM and feed-forward baselines. MASt3R-SLAM, which fits a free scale to the ground truth, has a lower mean trajectory error (22.0 mm). ORB-SLAM3 completes only two of the six airways, whereas our system completes all six and also reconstructs two ex vivo specimens.
[CV-215] VLA-GAUGE: Systematically Evaluating Generalization and Robustness in Vision-Language-Action Models
链接: https://arxiv.org/abs/2610.16567
作者: Jun Xiao,Rui Ye,Dong Cao,Jinyu Gu,Ying Wang,Zhiliang Zhu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: 18 pages, 8 figures, 4 tables
Abstract:Vision-Language-Action models (VLAs) have achieved high success rates on familiar manipulation tasks, yet it remains unclear whether such scores reflect genuine zero-shot generalization and perturbation robustness. Existing evaluation frameworks provide limited capability-level diagnosis and combinatorial perturbation coverage, often without connecting diagnosis to targeted model refinement. We introduce VLA-GAUGE, a diagnostic framework that unifies both evaluations under a shared executable task representation. It combines (1) paired in-distribution and out-of-distribution tasks across six capability suites, (2) single-factor sensitivity analysis with combinatorial perturbation screening and evaluation, and (3) diagnosis-driven refinement through targeted data augmentation and model fine-tuning. We evaluate 12 VLAs spanning different model sizes and action-modeling paradigms under a unified protocol. We find that (1) The mean Zero-Shot Score on VLA-GAUGE is only 22.3%, with SimpleVLA-RL scoring 8.0% and ranking second to last among the evaluated models. (2) Under three representative combinatorial perturbation configurations, the mean R-Score is 9.3. (3) Diagnosis-driven fine-tuning improves \pi_0.5 's zero-shot generalization and robustness under a selected perturbation combination on held-out scenarios by 72.4 and 31.8 scoring points, respectively; the latter exceeds the 10.9-point gain from LIBERO-Plus single-factor fine-tuning despite fewer successful training trajectories. These results show that high reported success rates on familiar manipulation tasks do not guarantee strong zero-shot generalization to unseen tasks or robustness to combinatorial perturbations, while diagnostic evaluation provides actionable signals for targeted model improvement.
[CV-216] NoduleFlow: Automatic 3D Lung Nodule Detection and Segmentation
链接: https://arxiv.org/abs/2610.16548
作者: Giulia Romoli,Filippo Ruffini,Francesco Di Feola,Valerio Guarrasi,Matteo Tortora,Paolo Soda
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Low dose Computed Tomography (CT) is the standard imaging modality for lung cancer screening. On CT scans, lung cancer typically manifests as pulmonary nodules, whose manual identification is time-consuming and prone to observer variability, image quality limitations, and suboptimal reading conditions. Moreover, subtle temporal changes are often difficult to quantify reliably in CT scans from the same patient across follow-up examinations. Deep learning-based approaches for the automatic detection and segmentation of lung nodules have demonstrated improved accuracy and efficiency. However, most existing approaches provide only coarse region-of-interest localisation, rely on prior lesion annotations or human supervision, or treat detection and segmentation as separate tasks. In this work, we present NoduleFlow, a framework for lung nodule detection, classification, and segmentation directly from series of CT volumes. Our proposal consists of four interconnected modules and operates in an automated manner at inference time, without requiring prior lesion annotations, manual initialisation, or region-of-interest pre-selection. We further investigate whether nodule localization benefits from the propagation of candidate regions across follow-up scans through inter-scan registration fields, as opposed to conventional independent per-scan analysis. The generalisability of the framework is systematically evaluated across four independent datasets. The results underscore the potential of automated pipelines for lung nodule analysis in screening and diagnostic settings.
[CV-217] Heuristics are Not Enough: Calibrated and Corrected Caching for Video Diffusion Acceleration
链接: https://arxiv.org/abs/2610.16545
作者: Jiyue Jiang,Zhiyu Huo,Yuhan Chen,Yanruisheng Shao,Gong Zhang,Lei Deng
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Feature caching has become a mainstream training-free strategy for accelerating video diffusion models by reusing intermediate outputs across denoising steps. However, existing adaptive methods rely on heuristic difference measures misaligned with actual generation quality, employ a single global threshold that ignores heterogeneous phase dynamics, and never validate their budget estimates against true approximation error, allowing systematic bias to propagate uncorrected. We propose \textbfAlignCache, a closed-loop caching framework that directly aligns scheduling decisions with their true influence on the final generated video through three tightly coupled components: (1) phase-adaptive budget routing that partitions the denoising trajectory into structurally distinct phases, each governed by an independent threshold and a staleness-aware guard; (2) closed-loop threshold calibration that exploits counterfactual error measurement at the classifier-free guided output level to adjust decision boundaries online via an EMA feedback loop; and (3) momentum-augmented draft synthesis combining direction-magnitude gated residual extrapolation with anchor-guided unconditional correction. By transforming scheduling from a static risk accumulator into a feedback-driven, self-correcting controller, AlignCache consistently achieves superior acceleration-quality trade-offs across state-of-the-art video generation models. Our code is available at this https URL.
[CV-218] Deep Learning-based Mechanical Characterization of Synthetically Generated Plant Fibers under Transverse Compression
链接: https://arxiv.org/abs/2610.16529
作者: Nikolaos Kegkeroglou,Bilal Sabillah,Myrto Inglezou,Vincent Placet,Céndric Clevy,Antonios Porichis
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Plant fibers are increasingly used as sustainable reinforcements in composites, but the intrinsic variability of their microstructure causes substantial scatter in transverse mechanical behavior, complicating composite design. Characterizing this behavior relies on finite element (FE) simulations requiring several minutes of high-performance computing per fiber, making high-throughput characterization across realistic fiber populations impractical. In this study, we propose a deep learning framework replacing this FE bottleneck with a fast pipeline. It includes two coupled recursive U-Nets predicting the step-by-step evolution of fiber deformation and internal stress fields under transverse compression, followed by a regression CNN estimating the transverse Young’s modulus from the final stress field. Unlike prior approaches predicting stress fields for fixed configurations or idealized geometries, our framework recursively resolves the full temporal stress evolution for synthetic fibers derived from real hemp cross-sections, with or without a lumen. Trained on 1,000 synthetic fiber geometries with FE ground truth, it reproduces the stress field sequence with a Structural Similarity Index Measure (SSIM) of 0.9574 \pm 0.0074 and a Root Mean Squared Error (RMSE) of 0.0345 \pm 0.0079 . It estimates transverse Young’s modulus with an R^2 of 0.938 \pm 0.024 , a Mean Absolute Error (MAE) of 24.31 \pm 1.23 MPa, and a Mean Absolute Percentage Error (MAPE) of 5.25 \pm 0.91% . Inference requires under one second per fiber versus five minutes per FE simulation, drastically reducing computational cost. This offers a fast, reliable tool for mechanical characterization of synthetic plant fibers, with promising applications in designing natural fiber-reinforced composites.
[CV-219] KARAMEL: Knowledge-Aware Ranking for Multimodal Entity Linking in the Arts via Heterogeneous Graphs
链接: https://arxiv.org/abs/2610.16502
作者: Raffaele Scaringi,Alejandro Sierra-Múnera,Gennaro Vessio,Giovanna Castellano,Ralf Krestel
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 19 pages, 7 figures, 5 tables, journal: Machine Learning
Abstract:Discovering and disambiguating entities from multimodal and heterogeneous data is a key challenge in discovery science, where relevant knowledge is often implicit and distributed across modalities. We present KARAMEL, a framework for multimodal entity linking (MEL) - the task of aligning visual and textual mentions to entities in a knowledge base - in the cultural heritage domain. Unlike prior vision-language models that primarily rely on surface-level multimodal alignment, our approach incorporates sociohistorical and contextual knowledge through a knowledge-aware architecture that fuses a vision-language encoder with a heterogeneous graph neural network. This design enables contextual reasoning over complex relational structures and supports robust entity disambiguation under ambiguous or underspecified evidence. To benchmark multimodal discovery in this setting, we introduce WikiMuSA (Wikidata-based Multimodal Semantic data for Art), a new dataset that links artworks to Wikidata entities via images, textual descriptions, and structured contextual knowledge. Experimental results show that KARAMEL surpasses strong baselines, particularly on implicit or abstract mentions, while ablation studies highlight the critical role of domain-specific, structured context in learning expressive representations. While instantiated in the cultural heritage domain, KARAMEL addresses a general knowledge discovery problem and applies to other scientific settings involving multimodal observations and heterogeneous background knowledge. Both the dataset and source code are made publicly available.
[CV-220] Self-Verifier Capture in Unified Visual Generators: Diagnosing and Repairing Closed-Loop Self-Reward
链接: https://arxiv.org/abs/2610.16500
作者: Binyan Chen,Tong Pan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 36 pages, 5 figures; supplementary material included
Abstract:Image-space reward requires decoding and a separate judge; unified visual generators can instead reuse their understanding pathway for decode-free, self-supplied reward. Optimizing against this verifier closes the self-reward loop; shared trainable weights let one update move generation and scoring even with the reward gradient stopped. We call the resulting failure self-verifier capture: self-reward rises while independently measured quality falls. Matched semantic/denoising by shared/frozen cells on Scale-RAE isolate two routes: the policy exploits on-manifold blind spots of a frozen semantic judge, while shared updates additionally rewrite the judge. A frozen judge also becomes stale, and repeated self-refresh yields progressively weaker recoveries. The diagnosis yields a two-part repair: locking the warmed reward read blocks rewriting, while sparse, parameter-disjoint external anchors counter residual exploitation. Unanchored rollouts retain semantic reward across multiple latent noise levels, providing dense latent reward that terminal image-space judging cannot supply, while a drift-triggered controller allocates anchors to high-drift rollouts and propagates their calibration within each group. At matched budgets, the controller improves on fixed-dose anchoring, increasingly so as coverage tightens. Both the unanchored failure and the repair transfer to the higher-performing, architecturally distinct Show-o2. Across both bases, the complete method significantly outperforms a construct-matched every-rollout judge on GenEval and DPG-Bench while externally scoring only 1/16 of rollouts. At 1/32 coverage, it scores higher in all four comparisons and remains jointly non-inferior.
[CV-221] MiMaAlign: Hierarchical Micro-Macro Alignment for Physically Plausible Video Generation
链接: https://arxiv.org/abs/2610.16482
作者: Yansong Liu,Litong Gong,Weijie Li,Tiezheng Ge,Ming Zeng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 20 pages, 14 figures
Abstract:Text-to-video (T2V) diffusion models have made remarkable progress in generating high-fidelity visuals, yet their outputs still often exhibit physically implausible phenomena, including discontinuous dynamics and implausible event interactions. Existing representation alignment methods mainly transfer generic video features or short-range motion cues, treating physical plausibility as a single-level representation problem. To address this issue, we propose MiMaAlign, a hierarchical micro-macro alignment framework that transfers complementary dynamics-oriented and event-oriented knowledge into different abstraction depths of a pretrained video diffusion model. Specifically, micro-level continuous dynamics alignment imposes constraints by aligning diffusion features with a continuous-time motion representation learned by a Neural Ordinary Differential Equation (Neural ODE), while macro-level event-commonsense alignment provides representations of object interactions and event semantics through an instruction-prompted vision-language model (VLM). With training-only auxiliary objectives, we align the complementary supervision from the two teachers with shallower and deeper generator features, respectively, enabling the generator to internalize the corresponding physical priors. Empirical evaluations show that, on CogVideoX-2B, MiMaAlign achieves a relative improvement of up to 21.0% in the Physical Commonsense (PC) score on the VideoPhy benchmark over the baseline, and also delivers substantial gains on the highly challenging VideoPhy2 benchmark, while maintaining general video quality on VBench and being supported by blind user comparisons.
[CV-222] XAI-Guided Fourier Domain Generalisation for Pure-Simulation Sim-to-Real Deep Reinforcement Learning
链接: https://arxiv.org/abs/2610.16478
作者: Wei Yao,Chuyao Wang,Ziwei Wang,Nabil Aouf
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Deep reinforcement learning has shown strong potential for vision-based autonomous driving, but policies trained in simulation often suffer severe performance degradation when deployed in the real world. Existing sim-to-real methods commonly address this gap through image-transfer-based domain adaptation or policy-agnostic domain randomisation. However, the former requires laborious acquisition of target-domain data and is often limited to the deployment domain, whereas the latter introduces unguided perturbations that may not reflect the underlying sim-to-real discrepancy, leading to suboptimal transfer performance. To overcome these limitations, we propose XFD, an explainable artificial intelligence (XAI)-guided Fourier domain generalisation framework for pure-simulation sim-to-real reinforcement learning. The proposed framework extracts explainability attribution maps from simulated observations, and converts these attributions into policy-conditioned Fourier-domain interventions. By perturbing appearance-sensitive spectral components while preserving task-defining spatial structures, XFD constructs robustness-enhancing visual domains entirely within simulation. Subsequently, the proposed robust policy optimisation guides the policy towards attending to task-critical features while suppressing sensitivity to irrelevant visual information. Extensive experiments in simulation and zero-shot real-world deployment demonstrate that XFD consistently improves traveled distance and reward over standard policy and randomly domain-randomised baseline, especially on unseen maps with cluttered backgrounds. Attribution-based analysis further proves that the proposed policy preserves more consistent task-relevant visual evidence across simulation and real-world observations.
[CV-223] Auditing Observation-Quality Shortcuts in the Evaluation of Person-Centered Motion for Video Anomaly Detection
链接: https://arxiv.org/abs/2610.16471
作者: Chaehyun Kim,Changwon Lim
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 9 pages, 2 figures
Abstract:Video anomaly detection (VAD) identifies deviations from normal patterns, but such deviations do not necessarily correspond to safety-critical events. We focus on human-centered emergencies and violence and investigate whether motion-based anomaly discrimination reflects task-relevant motion information or observation-quality differences arising from person detection, pose estimation, and tracking. We introduce an observation-quality-aware evaluation framework that combines quality diagnostics, quality balancing, motion orthogonalization, and robustness analyses. On human-centered categories of UCF-Crime, quality-only features match or exceed the discrimination of multi-cue motion in the original cohort, revealing a strong observation-quality shortcut. After cardinality matching brings all 11 measured quality variables below a prespecified marginal balance threshold, multi-cue motion consistently outperforms quality-only features across Logistic Regression, Random Forest, and RBF-SVM, yielding ROC-AUC gains of 0.046-0.141. Because nonlinear quality-only prediction remains above chance after matching, we interpret this result as additional discrimination beyond measured marginal quality rather than as evidence of quality-free or causal motion understanding. These findings show that high VAD performance should not be interpreted as evidence of abnormal-motion modeling without auditing the reliability of the person-observation pipeline. Our study provides a systematic protocol for evaluating observation-quality shortcuts and for assessing whether motion-based discrimination survives explicit control of detection, pose-estimation, and tracking reliability.
[CV-224] aching Vision-Language Models to Decline: Support Completion Against Mirag e
链接: https://arxiv.org/abs/2610.16465
作者: Alfaj Uddin Ahmed,Md. Shaown Miah,Md Arid Hasan,Syed Ishtiaque Ahmed,Sayeed Shafayet Chowdhury
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 23 pages, 10 figures. Includes supplementary material
Abstract:Vision-language models answer questions about images they were never shown, a failure called mirage. Such an answer can be correct without being justified by the evidence, and the difference matters wherever a model’s output is checked by someone who did not see its input. We locate the failure in the training objective: cross-entropy separates a model that reads the image from one that ignores it by only the conditional mutual information between answer and image, and no training example rewards declining. We formalise support completion, TIDE, which also shows each training question with a blank image, a same-domain wrong image and an out-of-domain image, with targets that decline, and prove that under TIDE every image-blind predictor is suboptimal by at least 1.04 nats per example, that the optimum declines exactly on the counterfactual kinds while keeping its behaviour on valid images, and that for any finite model the error of its answer-or-decline decision, declining on valid input plus answering on invalid input, is bounded below by how well its representation separates valid from invalid evidence. On Qwen2.5-VL-7B under a frozen held-out protocol with two judges, the three-condition predecessor of TIDE reduced the invalid-evidence mirage rate from 0.759 to 0.097 and the blank-image rate from 0.700 to 0.003 at a false-abstention cost of 0.126; label-free grounding raised the rate to 0.859, a likelihood-advantage contrast moved it only to 0.718 because its empty-image term carries no gradient, and grounding applied after support completion re-opened the mirage. The four-condition protocol on three base models, with seeds and component ablations, is in progress and not reported here.
[CV-225] GHOST: Geometry-Constrained Hybrid Diffusion for Open-Vocabulary 3D Scene Graph Generation ACM-MM2026
链接: https://arxiv.org/abs/2610.16455
作者: Changqun Feng,Wangxiandi Yin,Xin Hu,Ruiting Dai,Feiyang Yang,Jielei Wang,Tao He
类目: Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM)
备注: 10 pages. Accepted at ACM MM 2026 (34th ACM International Conference on Multimedia)
Abstract:3D scene graph generation requires understanding both object semantics and metric spatial relations. However, existing open-vocabulary methods often struggle with spatial predicates because language similarity reflects semantic co-occurrence rather than explicit geometry. We propose GHOST, a Geometry-constrained Hybrid diffusion framework for Open-vocabulary 3D Scene graph generaTion. GHOST jointly models continuous node geometry and discrete edge predicates, following a geometry-to-constraint-to-predicate denoising process to ensure geometrically consistent relation prediction. It introduces language-grounded constraint induction, which maps predicate embeddings to differentiable geometric constraints that guide both layout refinement and predicate reweighting. A Geometry-Topology Collaborative Denoiser further couples node and edge reasoning through noise-adaptive interaction and orientation-aware modeling. Experiments on 3DSSG demonstrate state-of-the-art performance in both closed-set and open-vocabulary settings, including over 17-point gains over RelationField on spatial predicates.
[CV-226] Not Every Cue Is Evidence: Cue Supportability for Forensic Distillation in Face Forgery Detection
链接: https://arxiv.org/abs/2610.16452
作者: Binjia Zhou,Dawei Luo,Shuai Chen,Feng Xu,Haoyuan Li,Jia wen Seow,Jiachi Wang,Jie Lei,Zunlei Feng,Yijun Bei
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Explainable face forgery detection requires not only correct judgments but also forensic evidence supported by the input image. However, training cues are typically derived from the forgery generation pipeline. Since images available to the model are of limited quality and resolution, it is difficult to observe the fine-grained traces these cues refer to. Training methods that further supply such cues as privileged supervision exacerbate the problem, as the student is taught to articulate evidence it cannot perceive. The student consequently resorts to templated or over-generalized expressions, giving rise to the correct-answer, wrong-evidence (CAWE) failure. We propose a dual supportability audit that filters privileged cues before distillation. Each candidate cue must be predictable from the remaining privileged context and causally tied to high-attribution visual regions in the current image. Only cues satisfying both conditions receive substantial distillation weight, preventing unsupported details from leaking into the student. Experiments across eight benchmarks show that our method maintains or improves detection accuracy while significantly raising explanation faithfulness compared to existing paradigms.
[CV-227] khonov-Stabilized Bezier Representation Forecasting for Training-free Diffusion Acceleration
链接: https://arxiv.org/abs/2610.16448
作者: Lei Zhu,Mujie Lin,Ruochong Zheng,Guangyi Wang,Hao Li,Peng Jin,Chang Liu,Jie Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 6 figures, 5 tables
Abstract:Diffusion models, particularly Diffusion Transformers, achieve strong image and video generation quality but remain expensive at inference time due to repeated denoiser evaluations. Feature caching offers a deployment-friendly acceleration strategy by reusing or predicting intermediate representations without retraining the generator or modifying the sampler. However, existing cache-then-forecast methods often rely on Taylor-style finite-difference extrapolation, which can become unstable over longer cache intervals, and typically predict local module outputs whose errors may accumulate through subsequent denoiser blocks. We propose \textbfBeziCast, a training-free diffusion acceleration framework that forecasts output-proximal denoising representations using low-order B’ezier trajectories. BeziCast estimates B’ezier control points via Tikhonov-stabilized fitting, providing a smooth, capacity-controlled temporal parameterization that avoids explicit high-order derivative estimation and decouples trajectory capacity from cache-query density. To moderate aggressive extrapolation, we further derive equivalent forecasting weights and introduce a convex-hull-inspired guardrail that detects high-risk predictions and softly pulls them toward a conservative simplex-projected estimate. Extensive evaluations across advanced image and video diffusion models demonstrate the effectiveness of BeziCast. In particular, BeziCast accelerates FLUX.1 by up to (4.79\times) and HunyuanVideo by up to (4.11\times), while preserving substantially better generation quality than competing acceleration baselines.
[CV-228] Pixels or Code? Benchmarking Instruction-Guided Visual Artifact Operation
链接: https://arxiv.org/abs/2610.16444
作者: Shun Guo,Shuizhou Shi,Boao Wang,Wei Xiong,Siyu Sun,Jiangtong Li,Jie Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Code: this https URL
Abstract:Visual artifacts such as charts, technical diagrams, webpages, and document pages underpin scientific, commercial, and engineering workflows. Editing these artifacts requires direct modification, dependency-aware updates, preservation of unaffected content, and compliance with domain constraints. We define this capability as Visual Artifact Operation (VAO). Prior benchmarks test parts of this problem, including cascading updates in charts. We study direct edits, dependent updates, and preservation across visual domains and pixel, code, and agent paradigms under one image-only protocol. We introduce Visual Artifact Operation Bench (VAO-Bench), comprising 1,573 manually screened tasks across nine artifact categories and three difficulty levels. VAO-Bench uses executable artifacts and deterministic rendering to trace reference transformations and annotate direct edits, dependent updates, and preservation constraints. All systems receive the same source image and instruction; we evaluate pixel-space editors, code-generating multimodal systems, and tool-using agents on direct-edit accuracy, dependency-aware update, content preservation, domain validity, and visual quality. Experiments show that executable-code systems better support dependent updates, while pixel-space editors offer faster execution and strong visual quality. Explicitly stating dependencies improves both paradigms, identifying implicit-update inference as a central bottleneck. VAO-Bench provides a reproducible foundation for dependable visual artifact editing across generation paradigms.
[CV-229] Federated Coreset Subsampling for Distributed Memory Banks
链接: https://arxiv.org/abs/2610.16425
作者: Jason Witzemann,Daniel Pröll,Sebastian Hegenbart
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:In real-world scenarios, industrial processes can span multiple factories and goods are often produced at multiple sites, leading to a shared yet non-identical distribution of data across locations. To address this, we propose two federated approaches for i) combining existing coresets (Double Subsampling) and ii) creating new coresets (Federated Subsampling) in a distributed anomaly detection setting. In particular, we adapt the memory bank based PatchCore algorithm to a federated setting, resulting in increased cold-start performance, resilience to domain- and sensor-shift, and resilience to unreliable actor participation, a common problem in classical, network based federated learning strategies.
[CV-230] Learning ω-Powered Conditional Distributions with Importance-Weighted Flow Matching
链接: https://arxiv.org/abs/2610.16419
作者: Maciej Falkiewicz,Frantzeska Lavda,Jason Ramapuram,Alexandros Kalousis
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: Code available at this https URL
Abstract:Classifier-Free Guidance (CFG) with guidance scale \gamma is commonly motivated by the \omega -powered conditional distribution obtained by setting \omega=\gamma , but CFG does not, in general, sample from this distribution exactly. We study whether the powered target can instead be learned explicitly. We introduce Importance-Weighted Conditional Flow Matching (IW-CFM), which reweights conditional flow-matching examples using density ratios obtained from frozen conditional and unconditional flow models. With exact normalized weights, the resulting objective has the same gradient as flow matching on the powered target; in practice, we use self-normalized estimation and a regularized empirical approximation for high-dimensional data. On synthetic distributions, IW-CFM recovers analytical powered targets for which numerically matched CFG, \gamma=\omega , exhibits a visible mismatch. On ImageNet, an amortized lightweight corrector achieves better alignment than fixed- and matched-guidance CFG with the corresponding weighted and filtered reference distributions. We additionally present an exploratory CLIP analysis of powered-density rankings on real ImageNet data. Aggregate shifts across selected prompt-specific contrasts coexist with heterogeneous classwise responses, including nonmonotonic trajectories. These results provide a practical route for studying and approximating model-defined powered conditional distributions.
[CV-231] Not All Challenges Are Equal: Evaluating Interactive Liveness Detection Against Real-Time Face Swaps
链接: https://arxiv.org/abs/2610.16417
作者: Anton Firc,David Drtil,Martin Perešíni,Milan Šalko,Vojtěch Staněk,Kamil Malinka
类目: Computer Vision and Pattern Recognition (cs.CV); Cryptography and Security (cs.CR)
备注: Accepted at BIOSIG 2026
Abstract:Remote identity verification assumes that a camera stream shows a live person, yet public real-time face-swap tools can replace that person before the stream reaches the verifier. Passive deepfake detectors generalize poorly to new generators, while interactive challenges are often selected without evidence of which gestures expose current attacks. We evaluate randomized gesture challenges in 108 live sessions with six subjects, 24 impersonated identities, and two face-swap tools, DeepFaceLive and FaceFusion, injected through a virtual camera. The static calibration segment reaches a session-level AUC of 0.878 with a frozen UCF spatial detector, while the mean across the active-gesture windows reaches 0.988. Pose and combined gestures have the highest point estimates, whereas expressions show no clear improvement. Hand occlusion against FaceFusion, configured with an occlusion mask, scores below the static baseline. The end-to-end time required to satisfy the validator also distinguishes attacks from genuine sessions, achieving a session-level AUC of 0.732, although this signal includes synthesis and transport delays. These results show that challenge sets should be tested against current attack configurations and that response time can provide an additional, deployment-specific signal.
[CV-232] Recovering from Pose-Selection Failures in Repeated Visual Localization
链接: https://arxiv.org/abs/2610.16409
作者: Jaehyun Kim,Wonseok Kang,Tae-Wan Kim
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: 9 pages, 3 figures
Abstract:Repetitive structures can give an incorrect pose candidate more inlier support than an accurate one, causing visual localization to select it. History-Guided Pose Reselection (HPR) uses coarse position forecasts from up to five past localization outputs to distinguish well-separated current-query candidates. The best-fitting forecast proposes an alternative; HPR accepts it only if every well-fitting forecast is much closer to it than to the localizer’s pose. HPR needs no odometry and returns the selected candidate unchanged, using prediction for selection rather than pose refinement. With one development-set threshold, HPR improves recall within 5 meters and 10 degrees in all 12 main configurations of Hierarchical Localization (HLoc), MASt3R-based pose generation, and JIST-based retrieval, losing successes on fewer than 0.4 percent of queries. On NAVER Gangnam B1, HPR corrects 97 HLoc failures and loses six successes; gains persist within 0.25 meters and 2 degrees on all three HLoc test sets. Median HPR runtime on HLoc candidates is below one millisecond, excluding pose generation. Recovery weakens under less frequent localization.
[CV-233] hrowing4: Phase-Aware Pose Modeling for Throw-Distance Prediction in Elite Athletics
链接: https://arxiv.org/abs/2610.16383
作者: Bikash Kumar Badatya,Kartike Tiwari,Jyotirmoy Amin,Ravi Hegde
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted for publication in Sports Engineering
Abstract:Elite athletics throwing provides a structured biomechanical action sequence together with an objective continuous outcome, namely throw distance, making it a suitable setting for performance-oriented sports analytics. In this work, we investigate whether explicit phase alignment of pose dynamics improves throw-distance prediction under realistic broadcast conditions. To support this study, we curate Throwing4, a dataset of 500 elite throwing clips spanning javelin, discus, hammer, and shot put, with corresponding two-dimensional pose sequences, frame-level phase annotations, and official distance labels. We formulate throw-distance prediction from broadcast-derived pose as a regression problem and compare full-sequence modeling against a phase-conditioned approach that partitions each attempt into four technique stages and fuses phase-specific motion representations. Experiments across multiple graph-based pose encoders show a consistent reduction in prediction error when phase structure is incorporated, indicating that phase-aligned modeling provides a more stable representation than treating the entire motion as an unstructured sequence. These findings suggest that explicit biomechanical phase information is a useful inductive prior for pose-based performance modeling in elite athletics throwing. Beyond score prediction, the dataset and protocol provide a basis for future work on fine-grained technique analysis, interpretable sports analytics, and robust modeling from broadcast video.
[CV-234] From Sound to Sense: An Anatomy of Semantic Supervision for Speech-Driven Visual Grounding in Remote Sensing Imagery
链接: https://arxiv.org/abs/2610.16364
作者: Zhongyu Pang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Speech-driven visual grounding in remote sensing (RSVG) removes the typing bottleneck of text-based interaction and is intended to preserve prosodic intent, enabling hands-free target acquisition in time-critical missions. The direct-audio paradigm, mapping raw speech straight to a bounding box and bypassing automatic speech recognition (ASR), was recently adopted by RSTalker, but what such grounders actually learn about language remains unmeasured. SemEcho grafts a frozen Whisper speech pathway onto a VLTVG grounding host so that speech and text share one decoder, making the paired text, free by construction since speech corpora in this task are synthesized from DIOR-RSVG expressions, a training-time teacher and a fair, same-host text upper bound. On this instrument we dissect three semantic injection channels (feature-mimicry distillation, word-level CTC supervision, and two-stage ASR pretraining) with linear semantic probes, and release Audio-DIOR-R (80,973 training clips; 10 timbres x K=3; Qwen3-TTS synthesis with online sox noise lambda in [0, 0.4]) and Audio-DIOR-Comp, a re-split holding out 17 attribute-category combinations under a 2x2 acoustic-semantic protocol. Our anatomy yields four findings. (i) Whether direct alignment beats a domain-trained ASR cascade is decided by recognizer quality: against a weak in-domain cascade (CTC 0.53) direct wins by 9.0 Acc@0.5 points with transcription errors concentrated on spatial-relation words, but a strong in-domain cascade (CTC 0.029) reaches the text upper bound (80.8 vs. 81.0) and beats even the strongest direct model by 8.8 points; the anti-ASR premise holds only against weak recognizers.
[CV-235] VideoSlate: Stateful Commit Control for Temporally Grounded Agent ic Video Reasoning
链接: https://arxiv.org/abs/2610.16357
作者: Shuo Zheng,Yuanjie Dang,Peng Chen,Jianjun Xiang,Ronghua Liang,Mohammed Bennamoun,Qiuhong Ke
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 23 pages, including appendix
Abstract:Agentic video reasoning systems gather and assess video evidence across stages, selecting relevant spans as visual evidence for answering questions. New candidates and observations prompt reassessment of the current span, but revisions can correct localization errors or discard useful evidence already present in the adopted span. Systems therefore need to connect evidence organization across stages with temporal state control, allowing accumulated information to jointly inform revision assessment while explicit adoption decisions govern how assessment outcomes affect downstream visual evidence. We introduce VideoSlate, a stateful commit-control framework built around shared evidence. A typed evidence blackboard links candidate spans, provenance, construction, and verifier observations, enabling an incumbent-relative certifier to assess revision support from the candidate, current span, and related proposals. Based on certification, the controller authorizes a revision or retains the current span, separating evidence accumulation from state updates and recording each decision and its evidence on the blackboard. Extensive evaluations across video grounding and grounded question-answering benchmarks show that VideoSlate can improve temporal localization through selective revision.
[CV-236] SyncReward: Human-Aligned Synchronization Reward for Audio-Visual Generation
链接: https://arxiv.org/abs/2610.16338
作者: Kaisi Guan,Ying Ba,Wenjing Wang,Xihua Wang,Chen Li,Jing Lyu,Ruihau Song
类目: Computer Vision and Pattern Recognition (cs.CV); Sound (cs.SD)
备注:
Abstract:Recent years have witnessed rapid progress in joint audio–visual generation. However, audio–visual synchronization remains a challenge, as generated speech and sound often temporally misalign with visual events. A key bottleneck is the lack of human-annotated data, reliable evaluation benchmarks, and synchronization-aware evaluation models for generated audio–visual outputs. To address these limitations, we conduct a comprehensive study spanning data, benchmarking, and reward modeling. We first introduce SyncReward-Data-80K, a dataset containing 80K professional synchronization ratings for clips from eight audio–visual generators. A taxonomy-guided construction pipeline covers four sound-event families: Speech, Onset, Instrument, and Ambient. We further construct \textbfSyncReward-Bench, a held-out benchmark of 1K clips, each independently rated by multiple professionals. Using this benchmark, we systematically evaluate specialized synchronization metrics and omni-modal models, revealing their substantial misalignment with human judgments. Finally, we develop SyncReward, a reward model that combines semantic–temporal contrastive pretraining on real clips with human supervision on generated clips. As an evaluation metric, SyncReward achieves the strongest agreement with human ratings among the evaluated methods. As a reward, it improves synchronization through inference-time best-of- N selection and reinforcement learning based post-training. We will release the data, benchmark, model, and code at this https URL.
[CV-237] WorldRoPE: Probabilistic World-Centric Rotary Embedding for Multi-View Attention KR
链接: https://arxiv.org/abs/2610.16337
作者: Mijeong Kim,Kyujin Lim,Bohyung Han
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Project page: this https URL
Abstract:Rotary Positional Embedding (RoPE) is widely used to encode relative positions in Transformers. In multi-view settings, however, patch coordinates are defined independently in each image: the same 3D point can therefore have different coordinates across views, making them insufficient for establishing cross-view correspondences. We introduce WorldRoPE, a probabilistic world-centric RoPE for multi-view attention. At each layer, WorldRoPE back-projects each patch token into a set of volumetric Gaussian primitives in a shared world frame. These primitives represent multiple plausible 3D supports for a token, jointly capturing alternative depth hypotheses, depth uncertainty, and the finite spatial extent of its patch frustum. When a key token’s support is projected into a query view, it induces a distribution over where that key may appear in the query image. WorldRoPE integrates the rotary operator over this distribution in closed form, combining rotations at the projected primitive centers according to their mixture weights with covariance-dependent attenuation. This makes attention less sensitive to uncertain key locations in the query view, without rasterizing or warping token features. The formulation reduces to deterministic RoPE when the support collapses to a point. Across layers, the predicted geometry guides cross-view attention, while the updated features refine the geometry used at the next layer. Experiments on novel view synthesis, depth estimation, and 3D geometry reconstruction demonstrate the effectiveness of WorldRoPE for multi-view Transformers.
[CV-238] BAT3R: Robust Online 3D Reconstruction with Bayesian Adaptive State Updates
链接: https://arxiv.org/abs/2610.16330
作者: Lijian Li,Yuanpeng He,Yudian Zheng,Chung-ju Huang,Linyu Li,Dongming Jin,Wenpin Jiao,Zhi Jin,Chi-Man Pun
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Feed-forward transformer models have driven remarkable progress in 3D vision,yet their quadratic complexity renders them impractical for long video sequences. Streaming 3D reconstruction addresses this bottleneck by processing frames sequen?tially with constant memory. However, existing recurrent architectures inevitably suffer from progressive degradation over long sequences, as they fail to account for the intrinsic reliability of observations and the accumulated confidence of the historical state, leading to severe error accumulation. To address these issues, we introduce BAT3R, a training-free Bayesian adaptation framework comprising two synergistic modules: State Innovation Gating (SIG) and Bayesian-guided Adaptive State Updating (BASU). SIG acts as a distributional regularizer that anchors state evolution to its geometric initialization, effectively arresting representational drift during long-term inference. Building on this, BASU reformulates state update as a recursive Bayesian inference process within a Kalman filtering framework. By jointly modeling the temporal reliability of observations via information-theoretic gain and spatial relevance via cross-attention maps, BASU achieves an asymmetric spatio-temporal state fusion: well-consolidated historical regions are protected from redundant updates, while genuinely novel scene content is aggressively incor?porated. Extensive experiments on standard benchmarks (7-Scenes, TUM, ScanNet, KITTI, etc.) demonstrate that BAT3R consistently outperforms state-of-the-art online 3D reconstruction methods, with the most significant gains observed on long sequences and complex scenes.
[CV-239] Explicitly Aligned Multi-View Breast Cancer Risk Prediction from Variable-Length Screening Histories
链接: https://arxiv.org/abs/2610.16329
作者: Solveig Thrun,Zijun Sun,Suaiba A. Salahuddin,Kristoffer Wickstrøm,Elisabeth Wetzer,Stine Hansen,Robert Jenssen,Michael Kampffmeyer
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Accurate breast cancer risk prediction from screening mammography is essential for improved risk stratification and personalized screening. It can benefit from both temporal breast tissue changes across longitudinal examinations and complementary anatomical information from craniocaudal (CC) and mediolateral oblique (MLO) views. Leveraging longitudinal information requires identifying corresponding tissue regions across examinations, yet most approaches rely on implicit feature alignment without explicitly enforcing spatial consistency. Recent work has demonstrated the benefits of explicit temporal alignment; however, existing methods model each view independently, preventing them from exploiting complementary CC and MLO information. Furthermore, explicit approaches have been limited to a single prior examination. To fill these gaps, we propose (this http URL), a breast-wise longitudinal multi-view framework that jointly models CC and MLO views across screening examinations while accommodating a variable number of prior examinations. FlexLMV-Net explicitly aligns longitudinal feature maps and integrates complementary views through cross-view temporal fusion and a dual-stream attention block. Evaluated on EMBED and CSAW-CC, FlexLMV-Net consistently outperforms existing approaches across breast density, cancer subtypes, and demographic subgroups, highlighting the benefits of explicit longitudinal alignment, multi-view modeling, and flexible use of available screening history. We further introduce (this http URL), an open-source benchmark that unifies state-of-the-art single- and multi-timepoint breast cancer risk prediction models under standardized preprocessing, training, inference, and evaluation protocols, enabling fair and reproducible comparison of existing methods.
[CV-240] Anomaly-Guided Observation and Two-Level Semantic Feedback for Training-Free Video Anomaly Analysis
链接: https://arxiv.org/abs/2610.16324
作者: Hangil Park,Seungheon Kim,Tae-Kyun Kim
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Training-free language-based video anomaly detection (VAD) uses frozen vision–language models (VLMs) and large language models (LLMs) to localize abnormal events without task-specific model-weight optimization. Existing approaches increasingly exploit broader temporal and semantic context, yet generic context does not necessarily foreground abnormal events or specify how video-level semantics should modify local anomaly scores. We introduce a training-free semantic feedback framework that uses VAD-guided active observation for video anomaly understanding (VAU) and returns the resulting anomaly-focused description to refine temporal detection. Starting from an upstream anomaly-score curve, we adapt reasoning-based iterative frame selection: high-score regions seed the observations, and anomaly scores together with VLM relevance assessments guide subsequent temporal exploration. A frozen video-language model generates a single open-ended anomaly description from the selected frames. The description is returned through two complementary feedback levels: video-level feedback determines whether the score distribution should be suppressed, preserved, or sharpened, with selective visual rechecking, while local feedback retrieves event-related windows and promotes scores only after target-local visual verification. All model weights remain frozen, and feedback is applied once without regenerating the description. On the HIVAU video-summary evaluations of UCF-Crime and XD-Violence, VAD-guided observation improves GPT-rated reasonability, detail, and consistency over the matched CLIP-guided variant. For VAD, semantic feedback improves the re-evaluated upstream detector from 84.36% to 86.25% AUC on UCF-Crime and from 68.07% to 72.58% AP on XD-Violence, where the final AUC is 91.74%.
[CV-241] Building age estimation based on widely available multi-source1 geospatial data
链接: https://arxiv.org/abs/2610.16320
作者: Jialu Sui,Hongsheng Zhang,Rui Liu,Jing Ling,Ruisheng Wang,Hao Wu,Gang Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Building age is fundamental for urban infrastructure safety, operation, and planning, as it directly affects structural safety, and sustainable development. Existing methods mostly rely on detailed building information such as street view, function, or materials, which are harder to obtain than geospatial data like satellite imagery and Digital Surface Models (DSMs). However, existing methods based on those geospatial data mainly rely on optical imagery and local texture features, limiting their ability to capture structural information and distinguish buildings from different periods. Moreover, the inherent temporal ordering among building age categories is often overlooked. To address these limitations, we propose the first deep-learning framework for multi-building age estimation using widely available multi-source geospatial data, including remote sensing data and their derived products, as well as building footprint information that estimates age at both pixel and instance levels within a unified pipeline. Specifically, a large pretrained backbone with fine-tuning adapters extracts representative features, and raster-derived and property-derived fusion modules support pixel-wise and instance-wise predictions. Pixel-level estimation employs a pyramid decoder with Cosine-guided Mixture-of-Experts (CMoE) to fuse multi-source raster data across scales, capturing fine-grained building age differences, while instance-level prediction uses attention pooling and Property-Conditional Adaptive Instance Normalization (PCAdaIN) to extract instance-level representations and integrate property attributes. A Temporal Ordinal Constraint Loss (TOCL) enforces the sequential ordering of predicted ages. We also construct a building age benchmark for Hong Kong to enable comprehensive evaluation. The source code and datasets are available at this https URL.
[CV-242] RAR: Rate-Adaptive Rollout for Time-Consistent Interactive World Models
链接: https://arxiv.org/abs/2610.16302
作者: Yaoyang Liu,Dongyu Yan,Yihua Du,Shengju Qian,Zeke Xie,Xin Wang,Ying-Cong Chen,Long Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Video-based interactive world models aim to enable real-time, user-controllable environments. Yet most existing systems are trained and rolled out at a fixed temporal rate, limiting both runtime-rate adaptation and physical-horizon coverage – a limitation we term temporal-rate coupling. Specifically, when runtime generation throughput differs from the learned temporal rate, the generated world evolves unintentionally more slowly or quickly. Moreover, using a single temporal rate imposes an inherent trade-off between temporal granularity and physical horizon: higher rates capture finer motion but cover a shorter duration under a fixed context window. To address both limitations, we introduce Rate-Adaptive Rollout (RAR), a framework that enables interactive world models to roll out at controllable temporal rates. At its core, Stride-RoPE aligns latent frames sampled at different rates on a shared physical-time coordinate, enabling consistent multi-rate rollout, unseen-rate generalization, and online rate switching without discarding history. Meanwhile, RAR uses multi-stride video sampling to enable joint learning of fine-grained dynamics and long-horizon evolution. Across action- and camera-controlled environments, RAR achieves temporally consistent multi-rate rollout and online rate switching while preserving high-rate generation quality and extending physical horizons under matched training budgets. RAR also transfers to pretrained world models while preserving perceptual quality.
[CV-243] Learning Temporal Evidence Selection for Video Anomaly Understanding
链接: https://arxiv.org/abs/2610.16299
作者: Hangil Park,Seungheon Kim,Tae-Kyun Kim
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Video anomaly understanding (VAU) aims to generate language-level interpretations of anomalous events in long videos, requiring visual evidence that supports explanation beyond anomaly detection alone. Existing VAU pipelines commonly rely on anomaly-centric or predefined temporal sampling strategies, which efficiently emphasize suspicious moments but overlook contextual cues, object interactions, and temporally distributed evidence needed for reasoning. Motivated by recent progress in learnable temporal selection for long-video question answering, we revisit VAU temporal sampling as a downstream-aligned evidence selection problem. We introduce the semantic-conditioned Temporal Sampling Q-Former that learns to select temporal visual evidence through the VAU generation objective. Since generic VAU prompts provide weak video-specific guidance, we condition the sampler on CLIP-embedded object tags automatically extracted from the input video. These tags are used only for temporal evidence selection and are not appended to the LLM generation prompt, separating sampler conditioning from prompt augmentation. The proposed sampler operates in visual-token space, computes soft temporal selection weights from frame-level representations, and merges duplicated selections before response generation. Experiments on HIVAU-70K show that our model achieves the best performance among the compared fine-tuned models, while ablations demonstrate that learnable temporal sampling improves over fixed anomaly-focused sampling and that CLIP-embedded object-tag conditioning further improves over generic-prompt conditioning. Our results suggest that temporal sampling in VAU should be aligned with explanation-oriented evidence utility rather than anomaly saliency alone.
[CV-244] ClothAtlas: Joint Garment Identification and State Estimation under Self-Occlusion
链接: https://arxiv.org/abs/2610.16293
作者: Jens Lundell,Alberta Longhini,Lei Shu,Hemal Arora,Evan Robert,Carlota Parés-Morlans,Marcel Büsching,Mårten Björkman,Jeannette Bohg
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: 9 pages, 6 figures, 2 tables. Project page: this https URL
Abstract:Garment manipulation requires inferring both garment identity and complete configuration from partial, self-occluded observations. This creates a chicken-and-egg problem: completing a partial observation requires a garment-specific deformation prior, while identifying the garment requires a configuration that explains the observation. We introduce ClothAtlas, which jointly identifies garments and estimates their complete configurations from partial point clouds and a library of canonical meshes. A shared graph-conditioned flow-matching model samples configurations for each candidate mesh and ranks candidates by observation fit. The model accommodates different mesh shapes, vertex counts, and connectivities, allowing new garments to be added from canonical meshes alone without retraining. On held-out DexGarmentLab garments, ClothAtlas ranks the correct mesh first among 221 candidates for all 22 test garments from a single flat observation. From a single self-occluded observation, it identifies 15/22 garments, compared with 8/22 for a deterministic graph estimator. Persistent vertex identities support grasp-target localization on real garments, reducing mean error from 15.4 to 6.4 cm overall and from 17.0 to 8.3 cm on occluded targets compared with dense descriptor matching. Project page: this https URL
[CV-245] 3D Gaussian Splatting for Fine-grained Urban Scene Reconstruction via Semantic-guided Density Control and Loss Design
链接: https://arxiv.org/abs/2610.16292
作者: Yueling Liao,San Jiang,Xing Zhang,Zhongwen Hu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:3D Gaussian Splatting (3DGS) has been becoming a promising 3D reconstruction technique because of its advantage of high-fidelity rendering effect with high efficiency. However, it still suffers from coarse density control and semantically blind loss functions when reconstructing the fine-grained structures in complex urban scenes. Thus, this study proposes SG-GS, a semantic-guided 3DGS method for fine-grained structure reconstruction. It takes the semantic segmentation deduced by the Segment Anything Model (SAM) as a constraint and systematically integrates it into both Gaussian density control and loss function design. On the one hand, a semantic-constrained adaptive density control mechanism is constructed, which dynamically allocates gradient thresholds to different semantic regions of the scene and introduces an additional boundary-enhanced densification strategy, so that new Gaussians are directed to structurally critical regions. On the other hand, a multi-feature guided loss function is designed, and fine-grained supervision signals are gradually introduced through a phased progressive training strategy, alleviating the artifacts caused by the lack of semantic constraints in the basic loss. Experiments on three datasets show that the proposed method consistently improves reconstruction accuracy. Compared with the original 3DGS, the PSNR increases by approximately 2.7% to 5.2%, while artifacts on linear structures and boundary blurring are effectively mitigated. The method can serve the fine-grained 3D reconstruction of urban scenes.
[CV-246] ProGAN-Based Synthetic Image Generation for Industrial Visual Inspection of Metallic Parts
链接: https://arxiv.org/abs/2610.16287
作者: Hossein Omid Beiki,Sepehr Asgari,Oliver Petrovic,Christian Brecher
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Automated visual inspection is essential for efficient repair and recycling of metallic components. AI-based defect detection is a strong fit. However, it is often constrained by scarce, low-diversity training data. To address this gap, we study Generative AI-driven synthetic image generation using Progressive Growing GANs (ProGAN). We implement a ProGAN pipeline with Wasserstein loss and gradient penalty and evaluate key hyperparameters on two public datasets, MVTec AD (metal nut) and MPDD (tube). As a result, training remained stable and the model yielded realistic images. These images can augment training data, improve robustness, and reduce the cost of collecting rare defects.
[CV-247] DepthTile: Depth-Guided Existing-Image Self-Tiling with Input-Side Tileability Assessment
链接: https://arxiv.org/abs/2610.16278
作者: Rui Wang,Xi Chen,Hui Dai,Piaopiao Yu,Yixue Hao,Long Hu,Min Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Existing-image self-tiling aims to transform a given image into a seamlessly repeatable tile while preserving its original semantics and local details. This task requires weak editing to retain the source but leave periodic seams, whereas strong editing improves tileability at the large risk of content this http URL diffusion-based tiling pipelines alleviate boundary discontinuities, yet lack awareness of which regions are geometrically important to preserve. In this work, we propose DepthTile, a depth-guided self-tiling method whose generation module requires no retraining of the diffusion backbone. It converts monocular depth into a spatial preservation prior and integrates it into the timestep-dependent masking mechanism of Tiled Diffusion. DepthTile converts depth into a spatial editability map and fuses it with the differential diffusion mask. This strengthens source preservation in near-field regions indicated by the monocular depth prior while retaining editing flexibility near compatible background boundaries. To access when self-tiling is feasible, we introduce the Self-Tiling Ability Score (STAS), a lightweight input-side predictor based on RGB, gradient, depth, and depth-gradient cues. On a 335-image benchmark, relative to the default Tiled Diffusion pipeline, DepthTile reduces MAE, LPIPS, and Tiling Score by 12.8%, 26.7%, and 8.4%, respectively. STAS reliably enables difficulty-aware analysis and further supports compatibility-aware pairwise image tiling.
[CV-248] Graph3R: Unifying View Graph Initialization with Feed-Forward Reconstruction
链接: https://arxiv.org/abs/2610.16272
作者: Jingwen Yu,Hengli Wang,Jianhao Jiao,Zhonghang Liu,Ping Tan,Hong Zhang
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注:
Abstract:Feed-forward reconstruction recovers camera poses and dense geometry from image sets whose views are assumed to come from a single scene. In practice, uncurated image collections often contain views from disjoint scenes, and reconstructing these views jointly can corrupt camera poses and geometry. Existing pipelines typically resolve view associations in separate stages before final reconstruction, leaving view graph initialization and reconstruction inference only loosely coupled. We introduce Graph3R, which unifies view graph initialization with feed-forward reconstruction through a shared geometric representation. A lightweight verification head predicts pairwise covisibility from the multi-view features of \pi^3 to construct a view graph. This graph classifies the input views into scene groups. For each group, we mask the shared features before decoding camera poses and dense geometry. Reusing the multi-view features allows Graph3R to reconstruct disjoint scenes separately with a single backbone pass. Across three benchmarks, Graph3R improves average maximum recall at 100% precision by upto 8.99 percentage points over the baseline, while improving reconstruction quality on mixed-scene inputs. Regarding the multi-session mapping task, Graph3R improves cross-session localization and achieves competitive pose estimation accuracy, while delivering a runtime speedup of approximately 35.9\times over the baseline when reconstructing the query group. By integrating learned view association with geometric inference, our approach takes a step toward end-to-end reconstruction from uncurated image collections.
[CV-249] Hardware-Friendly Token-Group Activation Quantization for Low-Bit Mamba Super-Resolution NEURIPS2026
链接: https://arxiv.org/abs/2610.16261
作者: Jinwoo Chung,Jangho Kim
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Accepted to NeurIPS 2026. 32 pages, 6 figures. Code: this https URL
Abstract:Mamba-based super-resolution models are efficient alternatives to Transformer-based restoration backbones, but their low-bit post-training quantization remains challenging because selective scan, Mamba’s input-dependent state-space operator, induces different activation ranges across spatial tokens. Thus, a single layer-wise clipping range poorly matches all tokens. We propose Token-Group Quantization, a hardware-friendly activation PTQ framework that statically approximates a hardware-unfriendly post-scan grouping oracle, where tokens with similar preferred clipping ranges are assigned to the same group. To avoid runtime post-scan grouping in static low-bit inference, a lightweight predictor infers the groups from pre-scan features available before the quantized scan path. During calibration, we form pseudo-labels by clustering post-scan activation descriptors that capture similar rounding–clipping behavior. At inference, the frozen predictor maps pre-scan features to groups and selects entries from offline-calibrated group-wise clipping-bound tables, without runtime statistics collection, dynamic clipping updates, or post-scan group assignment. To address low-bit restoration degradation, we further introduce a Frequency-Preserving Refinement (FPR) objective for clipping-bound calibration. FPR aligns Fourier amplitude and phase between full-precision teacher and quantized student outputs to preserve high-frequency structures. Across PTQ protocols, bit-widths, backbones, and tasks, our method consistently outperforms prior Mamba and SR PTQ baselines, transfers to Transformer SR and Vision Mamba classification, composes with rotation-based quantization, and achieves 2.23x speedup on Jetson Orin Nano over FP16. Code is available here.
[CV-250] EgoWholeView: Camera-Only Head-Mounted Capture of Body and Bimanual Hand Motion for Embodied Demonstrations
链接: https://arxiv.org/abs/2610.16256
作者: Yaoqing Hu,Xuguo He,Daohan Cui,Yukun Shi,Kai Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 17 pages, 9 figures
Abstract:Head-mounted body and bimanual motion capture supports self-avatars, teleoperation, and embodied demonstrations. However, a single egocentric camera is vulnerable to a limited field of view, scale ambiguity, and hand-object occlusion. We present EgoWholeView, a wearable capture system of six synchronized wide-angle RGB cameras with a role-specialized pipeline. The outward views localize the rig in a world frame, while an overlapping downward pair independently predicts SMPL-X body and MANO hand parameters. Both predictors are adapted to egocentric fisheye data. To combine the two, reliability cues measured from the predictions themselves weight the fusion of body and hand joint rotations, while shape is aggregated across views and frames. The pair’s baseline fixes metric scale, and triangulation recovers root translation. Relative to monocular prediction, fusion reduces PA-MPJPE by 3.7% for the body and 11.2% for the hands, and temporal refinement lowers angular-acceleration magnitude by 32.2% and 31.0%, respectively. On our six-camera recording, the pipeline produces body and hand states in a shared world frame using benchmark-fitted gates without retuning, and temporal refinement lowers hand acceleration magnitude by 41.08%. Using only RGB cameras and no third-person view, EgoWholeView offers a low-cost, portable route to egocentric interaction data.
[CV-251] MFlowHOI: Whole-Body Human-Object Interaction Generation via Part-Aware MeanFlow
链接: https://arxiv.org/abs/2610.16252
作者: Qiang He,Yonghao Zhang,Yinda Zhang,Chenyu Meng,Liufei Ouyang,Haoxiang Hu,Cangjun Gao,Xiaoming Deng,Cuixia Ma,Hongan Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 9 pages, 6 figures, 4 tables
Abstract:Whole-body human-object interaction (HOI) generation requires coordinating body and finger motion, object trajectories, and time-varying contacts. Existing methods typically rely on iterative diffusion or test-time optimization, resulting in high inference costs. Direct one-step generation is highly desired in downstream applications due to efficient inference. However, it is challenging because these HOI components differ substantially in structure and scale. We present MFlowHOI, an efficient one-step framework for whole-body HOI generation. A Contact-Aware Multipart RVQVAE organizes body, hand, object, and contact signals into a structured latent space, with a shared decoder preserving their global coordination. An ST-DiT then models temporal and cross-part dependencies in this space and generates a complete interaction in a single MeanFlow evaluation. Predicted object-local contact correspondences further enable geometric refinement and interaction-preserving transfer to photorealistic 3D Gaussian avatars with different body proportions. Experiments demonstrate that MFlowHOI outperforms existing base generators in motion quality and text alignment while substantially reducing inference time.
[CV-252] Seg2Click: Learning Pixel-Level Action Masks for GUI Visual Grounding
链接: https://arxiv.org/abs/2610.16234
作者: Jingchao Wang,Dingjiang Huang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 19 pages
Abstract:Visual grounding is a central challenge for GUI agents, requiring an instruction to be mapped to a target within a crowded screenshot. Coordinate generation with single-point supervision reduces a spatially extended target to one position, despite the presence of many valid click locations. Region-based patch scoring alleviates this ambiguity, but its spatial decisions remain tied to visual-token footprints that may span the target, neighboring controls, and background. To address these limitations, we propose Seg2Click, which reformulates GUI grounding as executable-region prediction: identifying where a click can validly execute the given instruction. Using bounding-box supervision, Seg2Click predicts an executable-region mask and computes the final click point through a deterministic rule. This formulation captures the spatial extent of valid click locations and enables finer localization beyond the visual-token grid, making it well suited to small, densely arranged controls without requiring a verifier or zoom-in pass. Experiments across five backbone configurations from two model families demonstrate the effectiveness of Seg2Click. Seg2Click-9B achieves 92.06%, 94.10%, and 55.22% on ScreenSpot, ScreenSpot-v2, and ScreenSpot-Pro, respectively, while matched-backbone comparisons show improvements over both coordinate generation and patch-based grounding. Code will be available at this https URL.
[CV-253] Look Back to Move Forward: Learning to Revisit for Agent ic Compositional Image Generation
链接: https://arxiv.org/abs/2610.16224
作者: Zixu Wang,Xiaoran Zhao,Yahui Liu,Feng-ao Wang,Junhao Xiao,You Shu,Bing Liu,Yiting Liu,Songlei Jian,Zhiliang Tian,Dongsheng Li
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Text-to-image models have made remarkable progress in visual fidelity and instruction following, yet complex compositional generation remains challenging. Existing agentic approaches address this limitation through iterative generation and editing, adapting their actions based on interaction history. However, they typically fix the continuation state (i.e., the image from which refinement proceeds) to the latest image. Since iterative editing is not necessarily monotonic, the agent may continue refining from the latest image even when it has already degraded. In this work, we introduce GenRevisit, an agentic framework that extends action selection to joint continuation-state and action selection. The agent can not only use interaction history to adjust editing instructions but also select an image candidate to edit, jointly deciding where to continue from and how to refine. We train GenRevisit with SFT followed by agentic reinforcement learning. To address the coarse credit assignment of final outcome reward in GRPO, which assigns the same trajectory-level advantage to all actions, we further propose Normalized Action Credit Assignment (NACA) to redistribute positive advantage according to each action’s contribution to reducing the remaining objective. On GenEval2, GenRevisit achieves GM scores of 84.3 and 74.4 with Qwen-Image-2512 and Z-Image-Turbo, surpassing the strongest baselines using the same generators by 12.7 and 10.9 points respectively. GenRevisit achieves state-of-the-art performance on GenEval, GenEval++, and GenEval2 among baselines.
[CV-254] MP-MNCA: Morphology-Prior Masked Neighbor Cross-Attention for Efficient Spatial Domain Identification
链接: https://arxiv.org/abs/2610.16212
作者: Sidharthan Sivacoumar Chitra,Sanjeev Manivannan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Tissues and Organs (q-bio.TO)
备注: Both authors contributed equally to this work
Abstract:Spatial transcriptomics jointly captures molecular state, tissue morphology, and spatial organization, yet spatial-domain methods are typically compared only at late training checkpoints, obscuring differences in convergence efficiency. We introduce MP-MNCA, a morphology-prior masked neighbor cross-attention framework that preserves the full 3,000 dimensional highly-variable-gene representation while selectively routing molecular information across spatial neighbors, with histological similarity incorporated as a continuous attention prior under a self-supervised masked neighbor-contrastive objective. Across 12 heterogeneous human dorsolateral prefrontal cortex (DLPFC) sections, MP-MNCA is evaluated against established spatial-domain baselines using both convergence trajectories and epoch-wise spatial-domain visualizations. Our results show that MP-MNCA (i) reaches coherent spatial-domain structure in substantially fewer training epochs, as reflected quantitatively by convergence curves and qualitatively by progressive spatial grids, and (ii) maintains strong clustering quality, achieving a mean peak refined ARI of 0.546 versus 0.549 for the strongest baseline (STAIG), while attaining the highest mean refined NMI of 0.657 versus 0.640. Moreover, MP-MNCA outperforms STAIG on 7 of 12 sections in ARI and 11 of 12 in NMI, demonstrating faster convergence without sacrificing spatial-domain quality.
[CV-255] VGGA: Geometry-Grounded Autoencoder for Video Generation
链接: https://arxiv.org/abs/2610.16204
作者: Chaofan Gan,Yian Zhao,Shijie Li,Tieyuan Chen,Zicheng Zhao,Mehrtash Harandi,Weiyao Lin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Autoencoders are the core component of latent video generative models, compressing high-dimensional videos into compact latent spaces for downstream generation. However, existing video autoencoders are primarily optimized for pixel-level reconstruction, providing limited supervision for learning structured latent representations. We argue that videos should not be treated merely as signals to be reconstructed, but rather as 2D projections of dynamic 3D worlds that exhibit explicit and well-structured geometric properties. Building on this insight, we propose \textbfVideo \textbfGeometry \textbfGrounded \textbfAutoencoder (\textbfVGGA), a geometry-aware video autoencoder that explicitly incorporates geometric constraints into the reconstruction process. Specifically, VGGA grounds the learned representations along three complementary geometric dimensions: \emphStructural Depth, \emphMotion Correspondence, and \emphProjective Warping. These objectives are unified within a Video Geometry Decoder, which employs a shared spatio-temporal upsampling trunk to jointly optimize geometric learning and RGB reconstruction. Through these geometry-grounded objectives, VGGA encourages the latent representations to capture the underlying 3D structure and motion of dynamic scenes. VGGA consistently improves video generation across both geometry-centric and human-motion domains, reducing gFVD by 28.1% on RealEstate10K and 37.7% on Taichi-HD. Notably, these substantial generation improvements are achieved while maintaining comparable reconstruction quality, demonstrating that geometry-grounded representation learning yields a more effective latent space for video generation.
[CV-256] DIGS: Distribution-Informed Gaussian Splatting for Training-Free Open-Vocabulary 3D Segmentation
链接: https://arxiv.org/abs/2610.16177
作者: Zhifan Zeng,Wenbo Xiao,Arcot Sowmya,Changming Sun
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Open-vocabulary 3D Gaussian segmentation typically binds semantic features to the per-Gaussian state via per-scene gradient training (1-4 h) or the more recent closed-form distillation. Both commit each primitive to a single feature; therefore, boundary Gaussians spanning multiple 2D instances are forced into a hard decision that degrades retrieval accuracy and yields unstable mask boundaries at render time. We present DIGS, a training-free framework with two stages, class-agnostic mapping and multi-modal encoding, that decouples instance discovery from linguistic naming. At the Gaussian level, our distribution-informed representation maintains an explicit top- M label distribution per primitive, updated by a tempered Dirichlet rule; preservation of multi-peaked label distributions at boundary primitives allows a late-argmax renderer to recover clean instance boundaries. A bidirectional state machine aggregates these distributions into a persistent class-agnostic instance ontology. The instance-level encoding stage then fuses the CLIP visual embedding of each instance’s isolated 3D footprint with the CLIP-text embedding of an LLM-generated attribute description, disambiguating visually similar instances. On a single A100, DIGS completes mapping and instance encoding on LERF-OVS in 1.0-3.2 min per scene, depending on whether LLM fusion is enabled, and serves native 3D queries in 5.4 ms. It reaches 71.95% mean mIoU on LERF-OVS, outperforming both the training-free SFS and training-based LaGa baselines. DIGS also achieves the highest mean mIoU on LERF-Mask and 3D-OVS and surpasses Gaga and ObjectGS on Replica under their respective evaluation protocols. Code is available at this https URL.
[CV-257] EgoAlign: Situated 3D Understanding without Situation Descriptions
链接: https://arxiv.org/abs/2610.16176
作者: Yiyang Li,Tianyi Ma,Zehong Wang,Yanfang Ye
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Situated 3D Question Answering requires models to interpret a 3D scene relative to an agent’s pose. Existing methods commonly convey that pose through a natural-language situation description, which may be unavailable or costly to obtain in real-world deployments. While some methods additionally incorporate numerical pose, we find that randomly shuffling the agent pose causes little performance degradation, indicating limited functional dependence on the numerical pose. To address this gap, we introduce EgoAlign, a simple 3D language model that represents a scene in an agent-centric coordinate frame, together with a dedicated egocentric alignment stage comprising four complementary tasks: object category recognition, egocentric direction prediction, metric distance estimation, and spatial relation prediction. This stage teaches the model to consistently interpret scene geometry relative to the agent’s pose before task-specific fine-tuning. EgoAlign achieves the strongest performance on SQA3D, Real-3DQA, and MSQA among compared methods while remaining highly efficient. More importantly, removing textual situation descriptions yields comparable or better downstream performance, whereas shuffling EgoAlign’s numerical pose causes substantial degradation. These results show that EgoAlign establishes agent-centric spatial context directly from numerical pose without requiring textual situation descriptions.
[CV-258] EventLens: Event-Structure Reinforcement Learning for Video Understanding
链接: https://arxiv.org/abs/2610.16172
作者: Guowen Zhang,Boshen Xu,Zihao Yue,Ziheng Wang,Xiaokun Liu,Xin Tao,Wenyu Qin,Pengfei Wan,Qin Jin
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Multimedia (cs.MM)
备注: 22 pages, 8 figures, 13 tables. Project page: this https URL
Abstract:Recent multimodal large language models (MLLMs) achieve strong results on video question answering, yet often fail to recover the temporal structure that makes a video coherent. They may merge neighboring events, miss single-event progression, or infer implausible relations across events. We study structured temporal understanding as the ability to recover event structure from visual evidence, including event boundaries, single-event progression, and multi-event relations. We argue that this gap arises from a mismatch between current post-training objectives and temporal reasoning, where caption-style supervision allows models to rely on language shortcuts rather than visual temporal grounding. To address this, we propose EventLens, a vision-centric reinforcement learning framework that learns video understanding through event-structure recovery. Instead of reconstructing captions, EventLens trains models to recover temporal structure under controlled perturbations, such as altered temporal granularity, reversed progression, and shuffled event order. We instantiate this framework with a three-level temporal ontology and derive three verifiable task families: event segmentation, progression discrimination, and multi-event relation reasoning. These tasks admit deterministic rewards, enabling task-specialized GRPO training without human preference labels or learned reward models. We further introduce multi-teacher on-policy distillation (MT-OPD) to consolidate specialized policies into a unified model. Experiments on Qwen3-VL backbones show consistent improvements on temporal grounding and event-sensitive reasoning benchmarks, while preserving general video-QA performance. Learning curves further demonstrate that EventLens achieves comparable downstream transfer with significantly reduced training cost compared to direct mixed RL optimization.
[CV-259] Geometric Parrots: Rule Absence vs. Rule Frag ility in Spatial and Physical Reasoning for Vision-Language Models
链接: https://arxiv.org/abs/2610.16170
作者: Bilal Jawaid,Hamza Hanif,Muhammad Maaz Uddin,Anas Zafar,Inad Alqurashi,Aizan Zafar,Osama Farooqui,Sheheryar Khan,Ferhat Sadak,Shaina Raza,Rizwan Qureshi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-language models (VLMs) can answer many geometric and physical reasoning questions correctly, yet still behave inconsistently in interaction, for example, abandoning a correct answer when confronted with a confident but incorrect user assertion. Standard single-turn accuracy cannot distinguish such capitulations from beneficial corrections because it scores each response in isolation. We introduce GRIP, a controlled benchmark with 34 procedurally generated reasoning domains, 100,000 images, and 600,000 questions spanning geometric, spatial, visual, and physical reasoning. Each scene is generated by deterministic code, and every answer is derived directly from the underlying scene state, providing exact ground truth without human annotation. On GRIP, we propose Dual-Loop Evaluation. A closed loop probes each image with a five-step compositional question ladder to localize where reasoning fails. An open loop then re-elicits the model’s answer under controlled challenges, including peer-model responses and unsupported user assertions. Because ground truth is known exactly, answer changes can be signed as corrections (wrong-to-right) or capitulations (right-to-wrong), separating two failure modes: rule absence and rule fragility. The dataset is available at this https URL, with further details at this https URL.
[CV-260] Esports World Model Dataset
链接: https://arxiv.org/abs/2610.16166
作者: Weiyu Ma,Zeyuan Gong,Xuhui Liu,Wenxuan Zhang,Jian Ding,Xinyu Cui,Jian Zhao,Mohamed Elhoseiny
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Interactive world models must capture how multiple participants jointly change a shared environment while observing different parts of it. Competitive games make this challenge concrete: players cooperate and compete, while fog of war, team vision and camera placement determine what each can see. Modeling these interactions requires data that connect individual observations and actions to shared state and events. We introduce the Esports World Model Dataset (EWM), a collection of human gameplay spanning StarCraft II, Dota 2 and Counter-Strike 2, with 3,053 games and maps and approximately 4,880 player-view hours, including estimated hours for the ongoing CS2 extension. It includes 4,628 completed StarCraft II views, all ten perspectives for 303 Dota 2 matches, and up to ten views per round in CS2. Across these views, replay-clock alignment connects rendered video with game-native actions, structured state and events from the same match. The three games expose complementary observation conditions: private fog of war, shared team vision and first-person viewpoints. In StarCraft II, player-specific entity records further distinguish live observations from remembered snapshots and link co-observed entities across opposing views. Together with temporal alignment, these records allow researchers to distinguish legitimate information differences from disagreement about shared state. By connecting human decisions, shared dynamics and player-specific information, EWM supports research toward jointly controlled multiplayer world models and interactive environments populated by teammates and opponents learned from human play.
[CV-261] Unravel: Recovering an Interactive 3D Scene from a Single Image
链接: https://arxiv.org/abs/2610.16146
作者: Hui Shan,Keyang Luo,Binjia Zhou,Ming Li,Kai Li,Xin Wang,Zhenyang Li,Zesong Yang,Weikai Chen,Runze Zhang,Xin Wang,Yanwei Fu,Zhen Chen,Xiangru Huang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:We present Unravel, a training-free framework that turns a single image into an interactive 3D scene, where every object is an independent and movable asset. Existing generators either produce one entangled geometry in which no object can be moved, or predict all objects in one feed-forward pass, where a wrong guess cannot be detected or repaired. Unravel instead takes a different approach: it first takes the scene apart, and then puts it back together. Starting from the input image, an agentic harness that iteratively peels the image layer by layer, and adopts point cloud as major modality for spatial relationship recovery. Specifically, a vision-language model plans which objects are fully visible, an image editor removes them and restores the occluded regions. Every edit is verified before being incorporated into the reconstruction. This process gradually turns occluded parts of the scene into direct observation, allowing objects to be reconstructed when they are best exposed rather than hallucinated from the original image alone. The scene is then rebuilt in reverse removal order, so that the supporting objects are naturally placed before the objects they support. Since a single-view image only reveals part of an object, we treat the observed points as a lower bound on its size: an asset may grow to cover its points, but can not shrink away from them. Finally, we physically settle the reconstructed scene to remove floating objects and interpenetrations, followed by an agentic audit that corrects remaining placement errors. Extensive experiments on diverse indoor scenes show that Unravel recovers more objects, places them more faithfully, and produces scenes that are more physically-plausible than existing methods.
[CV-262] All-St3R SLAM: Real-time RGB Dynamic SLAM with Pixel-wise Motion Masking IROS
链接: https://arxiv.org/abs/2610.16143
作者: Yewon Kim,Paul Hongsuck Seo,Sang Chul Ahn
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2026. Project Page: this https URL
Abstract:SLAM (Simultaneous Localization and Mapping) is a core technology in autonomous systems that enables them to estimate their own position and build a map of unknown environments simultaneously. SLAM systems in dynamic environments are of great practical importance. However, existing dynamic SLAM approaches still have several limitations like over-removal and high computational costs. To address these limitations, we propose a real-time RGB Dynamic SLAM framework that precisely removes moving objects at the pixel level without relying on predefined labels or segmentation models. Our system combines pixel motion predicted by ego-motion using DUSt3R with point tracking from All-Tracker, removing only actually moving pixels by analyzing discrepancies between the two motion sources. Unlike object-level masking methods, our approach eliminates only currently moving regions, allowing temporarily static parts to be utilized for tracking. Operating at approximately 10 FPS, the system achieves real-time performance on standard 30 FPS video streams by subsampling every 3 frames. Relying solely on RGB inputs, this significantly improves upon state-of-the-art dense dynamic SLAM approaches (0.5-3 FPS), presenting a highly practical solution.
[CV-263] ProMoDD: Dataset Distillation for 3D Human Motion via Prototype-Conditioned Diffusion Models
链接: https://arxiv.org/abs/2610.16134
作者: Yijin Wei,Yawen Zou,Wenqi Cai,Guang Li,Hangli Ge,Katsuya Hotta,Chunzhi Gu,Chao Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Dataset distillation (DD) aims to retain the training utility of a large dataset in a much smaller set of examples. For skeleton-based action recognition, this requires capturing discriminative motion variation without disrupting the geometry and temporal continuity of articulated bodies. Directly optimizing joint coordinates can violate these constraints, while plausible motion generation alone does not ensure useful training examples. We propose Prototype-Conditioned Motion Dataset Distillation (ProMoDD), which conditions a motion diffusion model on class-wise prototypes of temporally resolved recognition features. We further tailor diffusion training to motion data, enabling the generator to model both skeletal configurations and their temporal evolution. We then construct each compact training set through budget-aware hybrid selection: representative real sequences cover densely populated regions of the pooled recognition space, while selected generated candidates complement them in less-populated regions. This couples prototype-guided generation with feature-space selection, balancing the reliability of observed motions with the variation of synthesis under a fixed per-class sequence budget. Across five difficulty-stratified subsets of the single-person classes in NTU RGB+D 60, ProMoDD achieves its largest gains under tight distillation budgets, reaching 40.32% top-1 accuracy on Full49 with three sequences per class, 4.57 percentage points above class-wise k-means. These gains reflect the complete framework. The same distilled sets train ST-GCN, MS-G3D, and CTR-GCN without regeneration.
[CV-264] A Concept-informed Interpretable Artificial Intelligence Framework for Breast Ultrasound Diagnosis
链接: https://arxiv.org/abs/2610.16128
作者: Shengzhou Zhong,Ziying Wang,Liang Li,Lihua Yan,Shuochun Chen,Yudie Zou,Jifan Zhang,Wenhao Li,Huiling Zhang,Jiayin Lu,Ziting Xu,Qianjin Feng,Zhenwei Shi,Yan Chen,Yingjia Li,Yu Zhang,Zhenyuan Ning
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 31 pages, 17 figures, 4 tables
Abstract:Most existing artificial intelligence systems for breast ultrasound rely primarily on image-driven prediction, with limited incorporation of the knowledge from the Breast Imaging Reporting and Data System (BI-RADS) and the intermediate assessment processes underlying radiologists’ workflow, thereby constraining their interpretability and alignment with clinical practice. We curated, to our knowledge, the largest multicenter collection of breast ultrasound images paired with structured BI-RADS descriptors, comprising 9,941 patients across 20 independent datasets. Across 14 testing datasets, IBUS demonstrated consistent generalization performance across different clinical centers and imaging conditions. Incorporating BI-RADS descriptors enabled fine-grained risk stratification, while cross-checking predicted BI-RADS grades against malignancy probabilities identified internally discordant predictions that may warrant additional clinical review. In a reader study involving eight radiologists, IBUS assistance increased diagnostic sensitivity among both junior and senior readers, with the potential of guiding radiologists to make targeted revision in risk stratification. Overall, IBUS exhibits its potential as a second-stage decision-support system that complements radiologists by providing additional risk estimates and interpretable evidence for diagnostically challenging cases. The demo video of IBUS platform is publicly available at this https URL.
[CV-265] Geometry-Guided Semantic Reconstruction for 3D Instance Segmentation
链接: https://arxiv.org/abs/2610.16122
作者: Yuanhao Su,Shaofeng Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 5 figures
Abstract:3D instance segmentation requires representations that are both geometrically structured and semantically discriminative. Pure 3D encoders model spatial structure well but often lack strong appearance semantics, while directly lifting 2D priors into 3D by concatenation injects rich semantics without ensuring that they align well with the underlying 3D structure. In this paper, we address this problem by formulating 2D-to-3D semantic transfer as a \textbfmasked semantic reconstruction task. Specifically, we mask a subset of background lifted 2D priors and train the 3D encoder to recover the missing semantic priors, providing explicit supervision for distilling rich 2D semantics into the 3D backbone. We further propose \textbfgeometry-modulated semantic injection, which uses raw 3D geometry to modulate lifted 2D priors before sparse encoding, so that geometry explicitly controls semantic injection and alleviates the distribution gap between 2D semantic features and 3D geometric features. Experiments on two baseline frameworks demonstrate that the proposed method achieves state-of-the-art results in both full-data and few-shot settings on ScanNetV2 and ScanNet200. Code is publicly available at this https URL
[CV-266] Elephant in the Fridge: Constant-Memory Frame Packing for Long Video Understanding NEURIPS2026
链接: https://arxiv.org/abs/2610.16118
作者: Shuo Gao,Chenhao Zheng,Zhongzheng Ren
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Multimedia (cs.MM)
备注: NeurIPS 2026 ( this https URL )
Abstract:Hour-long videos, streaming feeds, and movie-length content are pushing vision-language models (VLMs) past their breaking point – the standard recipe of encoding each frame into visual tokens yields sequences that grow linearly with video length, causing prohibitive memory consumption and inference latency. Existing approaches, such as token reduction and keyframe selection, shorten the token sequence but leave the O(n) scaling intact, failing to resolve the fundamental bottleneck. We propose FramePack-U, a strikingly simple, training-free frame packing framework that compresses arbitrarily long videos into a fixed size visual representation at constant memory cost. FramePack-U requires no fine-tuning or architectural modification – it plugs directly into off-the-shelf VLMs at inference time. Inspired by FramePack from the generation literature, we propose a content-aware compression strategy for visual understanding under a fixed token budget. Specifically, FramePack-U (1) scores each frame by its semantic similarity to the text prompt and ranks frames in a diversity-driven manner, and (2) packs the selected frames at varying resolutions into a fixed-length context window. This concentrates the memory budget on frames that are both instruction-relevant and visually diverse. While FramePack-U can in theory pack arbitrarily many frames into the fixed budget, we identify an empirical instantiation that generalizes robustly across settings. Extensive experiments on LongVideoBench, MLVU, and Video-MME show that FramePack-U consistently outperforms full-resolution baselines and prior token reduction and keyframe selection methods across diverse model families, parameter scales, and frame budgets. Notably, on Qwen3.5-35B-A3B with a 128-frame budget, FramePack-U yields gains of +12.9%, +10.6%, and +9.7% over the official model on the three benchmarks, respectively.
[CV-267] Regime-Aware Router Learning for Multi-Domain Medical Image Generation with Diffusion Experts
链接: https://arxiv.org/abs/2610.16082
作者: Yaoshuai Ma,Wei Yao,Yongxiang Liu,Feifan Wang,Da Chang,Pengxiang Xu,Shixun Zhang,Yonghong Tian
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 10 pages, 4 figures
Abstract:Heterogeneous medical image data motivate domain-specialized diffusion experts, but expert specialization alone does not determine how their predictions should be combined during sampling. Learning a global router is particularly challenging when the server has limited, potentially unrepresentative images. We study two router-learning strategies for a fixed expert pool. Server-Only learns input- and timestep-dependent expert assignments from server images with component labels. Anchor-Delta (AD) uses the same images and frozen experts to obtain additional evidence from denoising losses. It distills targets that combine this evidence with the current router’s assignments, while a proxy gate controls a subsequent refinement step. Component labels describe data origin; expert losses measure empirical prediction performance. We evaluate these distinct signals through routing analyses and paired medical image generation. An AD-trained router improves generation metrics over Server-Only in a heterogeneous four-domain setting, and a separate server-data study shows gains in held-out routing over a corrected-routing comparator. Additional paired evaluations show that the effect of inference-time evidence mixing varies across benchmarks. These findings establish server-data availability and expert performance as distinct considerations in learning routers for multi-domain diffusion models.
[CV-268] PIGEONS: Incremental Long-Video Captioning with Persistent Identity Gallery and Evolving Overarching Narrative Summary
链接: https://arxiv.org/abs/2610.16066
作者: Xinlong Chen,Weihong Lin,Yuanxing Zhang,Xinyu Liu,Yue Ding,Bozhou Li,Bohan Zeng,Yang Shi,Qiang Liu,Pengfei Wan,Liang Wang,Tieniu Tan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Detailed video captioning has become an essential semantic interface for multi-modal understanding and generation. While existing methods mainly target short videos, extending them to long videos remains challenging, primarily due to limited context windows of current video understanding models. A naive solution is to adopt segment-wise captioning followed by concatenation, yet this paradigm often suffers from inconsistent subject references and fragmented long-range narratives. To mitigate these issues, we introduce PIGEONS, an incremental long-video captioning model built upon a Persistent Identity Gallery and an Evolving Overarching Narrative Summary. Before caption generation, PIGEONS scans the entire video to construct an identity gallery that captures key subjects along with their visual references, thereby providing a stable referential anchor. The model then performs clip-by-clip captioning conditioned on the persistent gallery and an evolving overarching narrative summary, which compactly maintains and updates essential long-range contextual information after each captioning step. To enable these capabilities, we develop a three-stage training strategy that integrates supervised fine-tuning and reinforcement learning. Extensive experiments demonstrate that PIGEONS markedly improves long-video captioning performance, achieving superior accuracy, completeness, and subject referential consistency over existing methods, while maintaining competitive performance on short-video captioning.
[CV-269] From Failure Prediction to Component Selection in Semantic Segmentation
链接: https://arxiv.org/abs/2610.16058
作者: Minhui Kang,Yiyao Zhang,Peizhe Long,Jiafeng Qi,Hongyu Jin,Jiayun Zheng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 55 pages total: 29-page main paper with 9 figures and 2 tables, followed by 26-page supplementary material with 5 figures and 20 tables. Evaluation code, experimental specifications, and numerical results: this https URL
Abstract:Reliable scene understanding requires deciding which predictions to trust, retain, or inspect. In semantic segmentation, component failure predictors support these decisions, but their predictive performance does not fully describe the value of the resulting selection. We develop a decision-centered evaluation framework that connects three questions: what an available score predicts, what can be learned from its existing fields, and what its ranking delivers. Across three segmentation systems on Cityscapes and ACDC at 90% retention, adapted MetaSeg reduces retained failure relative to forecast entropy in all twelve architecture-population comparisons, but does not establish a failure advantage over direct probability ranking. Image-nested source experiments reveal a different opportunity: reusing the full field set improves predictive and retained-loss performance under both learners in all three systems, while the raw OCRNet probe improves prediction with unresolved selective gains. Mapping a fixed prediction pair to conditional log-loss risk produces mixed selection effects, which we examine through the components exchanged by the two rankings. These findings distinguish the performance of a fitted score from the value obtainable from its inputs and the consequences of its use. The framework provides a practical basis for deciding whether to adopt an additional score, relearn existing fields, or revise a selection rule, and for explaining the gains and failures of component-level reliability methods.
[CV-270] Removing the Compute Bottleneck for Event-based Visual Place Recognition with Spiking Neural Networks
链接: https://arxiv.org/abs/2610.16022
作者: Somayeh Hussaini,Tobias Fischer,Michael Milford
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 6 figures, under review
Abstract:Neuromorphic sensing and computing promise low latency and high energy efficiency for robotics, yet most event-based visual place recognition (VPR) pipelines reintroduce dense, frame-like computation downstream of an inherently sparse sensor, forfeiting the sensor’s latency advantage. In this paper, we address the problem of resolving an event-based VPR query in tens of microseconds on a commodity CPU. We train a two-layer spiking neural network (SNN) with additive spike-timing-dependent plasticity (STDP), which drives the learned weight matrix towards a strongly bimodal distribution, amenable to aggressive magnitude pruning. At inference, our pipeline exploits this sparsity by reading only the pruned weights corresponding to the query’s active pixels. As surviving weights share a near-uniform magnitude, each neuron’s score reduces to an integer count of matching synapses instead of a weighted sum, lowering per-query latency. This computation is small enough to remain resident in a CPU’s fast on-chip cache. On three event-based VPR benchmarks, our sparse SNN exceeds the recall of LENS, an existing event-based SNN, while reducing per-query latency by 5.5-6.2x to tens of microseconds. It matches the recall of a dense variant of our network and of the sum-of-absolute-differences (SAD) baseline, while reading 10x fewer weights than the dense variant. We further characterise how recall and latency scale with the number of active pixels and reference places. These results suggest that the sparsity of both event-based input and learned weights can already be exploited for low-latency inference on conventional processors.
[CV-271] Adversarial Robustness of Camera-Radar Sensor Fusion for Autonomous Driving
链接: https://arxiv.org/abs/2610.16012
作者: Ce Zhou,Yubo Wu,Khang Nguyen,Liyang Xiao,Thomas Hou,Qiben Yan
类目: Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV)
备注: 25 pages
Abstract:Autonomous driving (AD) relies on multi-sensor perception to interpret the surrounding environment. However, the robustness of camera–radar fusion under adversarial conditions remains largely unexplored. This paper presents a systematic measurement study of the robustness of camera–radar sensor fusion in AD. We construct a unified evaluation pipeline on the nuScenes dataset that simulates attacks on both sensing modalities, including eight camera-image perturbations and three radar point-cloud perturbations. We evaluate five camera–radar fusion algorithms across two perception tasks, 3D object detection and depth estimation, and quantify their performance under clean and attacked inputs using standard metrics. Our evaluation reveals a clear asymmetry in the robustness of fusion systems. Camera perturbations substantially degrade both object detection and depth estimation, and moderate attack strengths are already sufficient to cause severe performance degradation. In contrast, the evaluated radar perturbations produce smaller and task-dependent effects, with object-detection models showing particularly limited degradation. Our analysis attributes this difference to the strong dependence of existing fusion architectures on dense camera features, together with the filtering, association, and confidence-weighting mechanisms commonly applied to sparse radar measurements. These results demonstrate that sensor fusion does not inherently guarantee robustness and that the contribution and reliability of each sensing modality must be explicitly considered under adversarial conditions. Based on four empirical findings, we derive measurement-guided design implications for future robust camera–radar fusion, including modality-specific reliability estimation, cross-modal consistency checking, adaptive fusion, and graceful failover under severe sensor degradation.
[CV-272] Meaningful Moments: A Corpus of VLM-Derived Temporal Importance Labels for Video Action Recognition WACV2027
链接: https://arxiv.org/abs/2610.16009
作者: Cameron Keith,SouYoung Jin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 5 figures. Submitted to WACV 2027. Supplementary material included
Abstract:Not every moment in a video is equally useful for recognizing an action. A few decisive moments often carry the evidence: the instant an object moves, the peak of a dive, or one object placed on another. Existing video datasets rarely provide large-scale, action-conditioned labels for which short segments carry that evidence. This work asks whether vision-language models can identify meaningful moments at scale and whether those labels improve action this http URL introduce Meaningful Moments (MM), a corpus of about 4.58 million VLM-generated per-segment importance scores for nearly 500,000 class-labeled videos spanning Something-Something v2 (SSv2), Kinetics-400 (K400), and Diving-48 across 622 action classes. We freeze the action classifier and vary only temporal sampling. An alpha-parameterized comparison of importance-led selection against an importance-inverted control separates gains from the VLM signal from gains caused by variable-density sampling alone. Across all three datasets, VLM-derived importance is informative: at the hard-cut endpoint, keeping VLM-important segments outperforms keeping VLM-designated filler. Usefulness depends on class structure. On Diving-48, importance-led fast-forwarding preserves full-video recognition while the inverted control collapses by roughly 30 percentage points. On K400, hard-cut selection beats uniform sampling by 1.45 percentage points, rising to 2.05 under threshold selection. On SSv2, the labels carry signal but do not recover full-video performance. Direct scoring matches greedy removal within its noise floor at roughly 25 times fewer oracle calls, and a four-VLM cross-oracle study shows consistent downstream trends. We release the importance labels and evaluation protocol under CC-BY-4.0; SSv2 class strings remain under CC-BY-NC-4.0.
[CV-273] Representational Difference Distillation
链接: https://arxiv.org/abs/2610.15997
作者: Abdulkarim Mugisha,Divin Irakiza
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: 9 pages, 7 figures, 1 table
Abstract:Knowledge distillation transfers representational knowledge from a large teacher network to a compact student. Existing relational and contrastive distillation methods treat all samples uniformly, spending equal gradient on well-matched and poorly-matched regions of the representation space alike. We propose Representational Difference Distillation (RDD), which diagnoses where teacher and student representations structurally disagree and concentrates the distillation signal on those regions. RDD uses two asymmetric affinity matrices that capture opposing directions of misalignment: one identifies samples the student incorrectly considers similar, the other identifies teacher relationships the student has yet to learn. These affinities determine which sample pairs the distillation loss aligns and how strongly it corrects each misaligned sample. On CIFAR-100 with resnet32x4 to resnet8x4, RDD reaches 75.58% test accuracy in our runs, on par with CRD (75.45%) and above KD and RKD. We additionally explore curriculum-based and sampling-based variants, finding that weighting the loss is more effective than filtering the data.
[CV-274] When Trackers Fail: VLM-Guided Verification and Recovery for Robust Video Object Segmentation NEURIPS2026
链接: https://arxiv.org/abs/2610.15929
作者: Valay Bundele,Susmit Agrawal,Mehran Hosseinzadeh, TheNam Nguyen,Hendrik P.A. Lensch
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to NeurIPS 2026
Abstract:Recent trackers built on foundation models such as the Segment Anything Model (SAM) achieve strong performance by propagating object masks through memory. However, this strength is also a key source of failure: under occlusion, reappearance, or distractors, an incorrect prediction can enter the tracking loop and be propagated over time, resulting in persistent identity drift. Existing approaches seek to prevent such failures through improved memory and temporal modeling, but once a failure occurs, the tracker has no explicit mechanism to recognize or correct it. We address this gap with Dual-Agent, a failure-aware tracking framework that complements propagation with explicit verification and recovery. We address this gap with Dual-Agent, a failure-aware tracking framework that complements propagation with explicit verification and recovery. A vision-language verification agent first reasons over spatio-temporal context to determine whether the tracker is still following the target. When a failure is detected, a correction agent uses temporal context to recover the target before tracking continues. The correction agent is trained on real SAM3 failures, augmented with targeted identity-switch and distractor perturbations that expose it to challenging ambiguities during recovery. On LVOS-v2 and SecVOS, Dual-Agent improves SAM3 by 1.2 and 2.7 J\F , respectively. Importantly, the proposed framework transfers without retraining to SAM2.1, SeC, and Cutie, as well as to long-term video object tracking benchmarks including LaSOT-ext, TNL2K, and VOT-LT2022. Our results show that explicitly reasoning about when tracking fails and how to recover provides a complementary path to robust tracking beyond improving propagation alone. Project page: this http URL.
[CV-275] Frequency-Step Cache: Separating Synchronization and Repair for Efficient Autoregressive Video Diffusion
链接: https://arxiv.org/abs/2610.15922
作者: Kerry Xu,Param Misra,Zachary Stenglein,Athri Karunamurthy,Fanjiang Ye,Yuke Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 30 pages, 11 figures, 18 tables
Abstract:Autoregressive (AR) video generation enables long-horizon synthesis but requires repeated forward passes through large diffusion transformers. Training-free caching reduces this cost by reusing intermediate computation, yet existing controllers often couple two decisions: when to synchronize the cache exactly and how much computation to repair between synchronizations. We introduce Frequency-Step Cache (FSC), which separates these roles. FSC schedules exact full-computation anchors on a dense-to-sparse trajectory, while a lightweight latent-space risk based on frequency-resolved drift and token age controls inter-anchor repair through whole-step replay, selective token recomputation, or full repair. On SkyReels-V2, whole-step reuse provides most of the acceleration, while selective token computation is most effective as fidelity repair. A controlled ablation shows that using one reactive signal for both synchronization and repair reduces fidelity at matched latency, with the same separation advantage observed on FramePack and Matrix-Game. Across multiple autoregressive video backbones, FSC improves speed-quality trade-offs. On SkyReels-V2, FSC-Speed reaches 2.36x speedup versus MotionCache’s 2.275x while improving SSIM by 0.019 and PSNR by 0.5 dB. At MotionCache-matched speed, FSC-Quality improves SSIM by 0.031 and PSNR by 1.0 dB and lowers LPIPS by 0.024. Without retuning, FSC-Quality transfers to SkyReels-V2 14B at 2.12x speedup versus 1.79x for MotionCache.
[CV-276] VIEWCRAFT: Evolving What Multimodal Agents See and How They Read It
链接: https://arxiv.org/abs/2610.15915
作者: Binxiao Xu,Ruichuan An,Bocheng Zou,Weihua Du,Xinyu Wei,Hangyu Guo,YiFan Zhang,Renye Yan,Chengqi Duan,Hang Hua
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Reusable skills improve multimodal agents without updating model weights, but their effectiveness depends on both the evidence presented and the procedure used to read it. We introduce ViewCraft, which jointly evolves an executable evidence interface and its reading procedure. The interface selects, transforms, and organizes observations; the reader interprets the resulting views. Failure traces guide revisions, replay and paired evaluation retain useful changes, and interface projection recovers visual edits from eligible rejected joint proposals. Across eight benchmarks, ViewCraft improves over direct execution by an average of 6.1 points in task score, with gains on every benchmark. The advantage persists at a matched input-token budget. On all five multi-source tasks, the learned interface uses 8%-15% fewer input tokens than Initial skill. The learned skills also transfer across executors and outperform direct execution in every evaluated transfer, supporting evidence construction and reading as a coupled, reusable skill.
[CV-277] Gen Through the Lens: Camera-Controllable Video Generation via Cross-Coordinate Conditioning Fusion UAI
链接: https://arxiv.org/abs/2610.15900
作者: Shuai Li,Jinxu Xiang,Xinyuan Li,Hanlei Guo,Xiyang Tan,Alan Zhao,Xiaotong Zhao,Ran Zhang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Graphics (cs.GR); Machine Learning (cs.LG)
备注: 15 pages, 9 figures, 5 tables. Shuai Li and Jinxu Xiang contributed equally; Ran Zhang is the corresponding author
Abstract:We present Gen Through the Lens, a camera-controllable image-to-video generation framework that synthesizes videos from a given reference image while enabling realistic and physically consistent camera operations. Unlike previous models that treat camera motion as abstract conditioning or rely on a single coordinate system, our approach explicitly decomposes and fuses control signals across three coordinate spaces: the camera coordinate (modeled by intrinsic parameters), the world coordinate (represented by spatial geometry hints), and the object coordinate (described by 3D bounding boxes of target subjects). Through cross-coordinate conditioning fusion, these heterogeneous cues are unified to guide the generative process, enabling precise control of camera motion, focus, and framing while preserving temporal and spatial coherence. Extensive experiments demonstrate that our method produces controllable and visually consistent results across diverse camera operations, outperforming existing approaches in both realism and editability.
[CV-278] Learning Multimodal Sensor Fusion with JEPA Representations for Breathing Activity and Pose Recognition
链接: https://arxiv.org/abs/2610.15891
作者: Yunxiang Luo,Constantino Álvarez Casado,Xiaoting Wu,Nhi Nguyen,Le Nguyen,Manuel Lage Cañellas,Miguel Bordallo López
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 6 pages, 2 figures, 3 equations, 4 tables, conference paper
Abstract:Predictive video representations do not by themselves specify how a multimodal system should handle missing sensors. We transfer frozen V-JEPA2 features to a set of heterogeneous sensors comprised by RGB video, depth video, and radar time series. A shared fusion model uses modality queries, fixed sensor slots, and explicit availability to classify breathing activity and pose from all seven non-empty sensor subsets. The contribution is a controlled evaluation of subset supervision, pooling capacity, and representation transfer on 39 OMuSense-23 subjects with five subject-group folds and three seeds. With matched downstream parameter counts, the final model reaches recording-level activity macro-F1 of 75.23%\pm0.60 averaged across seven subsets. Supervising all subsets in each batch improves the matched query-pooling control by 1.923 percentage points (conditional 95% subject-bootstrap interval [0.647, 3.209]), while query pooling shows no clear advantage over matched mean pooling under balanced supervision. Radar probes support transfer beyond pooled wavelet pixels, 64-token inference approximately preserves mean activity performance, and the tested Consensus-JEPA loss adds no measured benefit against its exact control.
[CV-279] Sharper Frames Worse Rooms: Viewpoint Coverag e Dominates Image Quality in Feed-Forward 3D Reconstruction
链接: https://arxiv.org/abs/2610.15854
作者: Alibek Dadajonov,Hikmatillo Umarhojiyev
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 9 pages, 3 figures, 4 tables
Abstract:Grounded spatial reasoning about indoor scenes rests on a geometric reconstruction that the multimodal system does not itself produce. Feed-forward multi-view reconstruction models operate under a fixed view budget, making frame selection a critical upstream decision. We show that under this setting the conventional preference for sharp frames can be actively harmful. Across nine handheld room scans, selecting the sharpest 16 of 200 candidate frames is the worst strategy we test, reaching 48% room dimension error where uniform temporal sampling reaches 3%. We trace this failure to a coupling between image quality and camera motion: sharp frames tend to occur during slow motion and therefore concentrate the selection in a narrow angular region, leaving 192^\circ of the room unobserved in the case we illustrate and over 300^\circ in the worst cells. A 2\times2 factorial experiment that independently manipulates viewpoint spread and per-frame sharpness shows that spread has roughly three times the main effect of quality, and that blurry but well spread selections outperform sharp but clustered selections in 8 of 9 captures. The advantage also diminishes as the frame budget increases and angular coverage saturates, consistent with this mechanism. Finally, we ask whether the failure is detectable at inference: the reconstruction model’s multi-view consistency score increases with reconstruction error, while a vision-language model given the same frames reports confidence unrelated to its own spatial estimation error. Under small view budgets, viewpoint coverage is a more important constraint than per-frame image quality, and neither uncertainty signal available at inference reveals this upstream failure.
[CV-280] Mobility Helps Learning: Unsupervised Model Adaptation for Object Recognition via Movement
链接: https://arxiv.org/abs/2610.15846
作者: Yanan Ma,Yihang Tao,Zhengru Fang,Senkang Hu,Xianhao Chen,Yuguang Fang
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 24 pages, 3 figures
Abstract:Movable agents, such as autonomous vehicles and robots, should be able to autonomously adapt their pre-trained object recognition models to new environments without human annotations. While existing test-time adaptation (TTA) methods address label-free model adaptation, they typically treat test data as independent snapshots, without constructively exploiting a rich, unique source of supervision: the significant variance in prediction quality as an agent observes the same object from different distances and viewpoints. To capitalize on this, we propose \textttMoCaFe, a model-agnostic and hyperparameter-insensitive paradigm that leverages motion-induced prediction variance for unsupervised adaptation. Specifically, we first develop mobility-calibrated filtering of pseudo-labels, which constructs reliable pseudo-labels by fusing cross-view predictions using inverse-variance weights. The filtered pseudo-labels then guide the selection of ``hard samples’’ for effective learning. To further suppress pseudo-label corruption, we introduce \textttp-MoCaFe, an optional extension that uses a public dataset to assign weighting factors to data samples, yielding an estimator of the clean loss with bounded bias. Extensive experiments on autonomous driving (nuScenes, KITTI) and embodied-agent datasets show that \textttMoCaFe significantly outperforms state-of-the-art TTA schemes, achieving up to approximately 20% accuracy gain.
[CV-281] ML-CLAPPS: A Probabilistic Framework for Imbalanced Multilabel Learning under Class Prioritization
链接: https://arxiv.org/abs/2610.15816
作者: Priel Hazan,Nir Almog,Noam Keidar,Yael Yaniv,Assaf Schuster
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (stat.ML)
备注: 10 pages, 5 figures
Abstract:Real-world datasets often exhibit severe class imbalance, degrading the performance of deep learning models, particularly on rare classes. While imbalance is well studied in binary and multiclass settings, it is substantially more challenging in multilabel problems due to complex label co-occurrence. In clinical applications such as ECG classification, this challenge is coupled with a second critical challenge: the need to prioritize high-risk conditions, regardless of their prevalence, to mitigate the high cost of false negatives. We introduce the Multilabel Class-Prioritized Probabilistic Sampler (ML-CLAPPS), a learnable sampler that optimizes a distribution over training instances to control the model’s exposure to different classes. This approach addresses multilabel imbalance while enabling prioritization of significant conditions via a user-defined importance vector (IV). Empirical evaluation on a real-world multilabel ECG image dataset demonstrates that ML-CLAPPS achieves state-of-the-art performance. We show that it aligns class exposure with the specified IV, and that increasing a class’s importance leads to corresponding improvements in its classification performance, enabling explicit prioritization of clinically critical conditions.
[CV-282] ChangeT3: Evaluating Vision Test-Time Training for Remote Sensing Change Detection
链接: https://arxiv.org/abs/2610.15813
作者: Sharukshan Niranjan,Oshadha Samarakoon,Nichula Wasalathilaka,Roshan Godaliyadda,Parakrama Ekanayake,Vijitha Herath
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 2 figures, 6 tables
Abstract:Bitemporal remote sensing change detection requires representations that preserve spatial detail while capturing contextual information relevant to changes between acquisitions. Visual test-time training offers an alternative to explicit token-to-token attention by adapting a compact inner model to each image. This letter describes ChangeT ^3 , a vanilla Siamese configuration combining the existing hierarchical ViT ^3 -Tiny encoder with the multiscale ChangeFormer decoder. The two image streams share learned encoder parameters, while their fast weights are computed independently from their respective spatial features. Each encoder block retains the inherited global simplified SwiGLU and local adaptive depthwise-convolution branches. Temporal comparison is performed by the decoder through learned fusion of paired features at four resolutions, followed by coarse-to-fine aggregation and binary prediction. This separation provides a direct setting for examining image-conditioned inner updates without introducing an additional temporal interaction module. Trained from scratch on LEVIR-CD+ with the same decoder, recipe, and budget as a ChangeFormer control, ChangeT ^3 reaches an F1 of 80.61 \pm 0.90 against 78.84 \pm 0.21 for ChangeFormer (three seeds each); as in the standard protocol for this dataset, checkpoints are selected on the test split. Trained with both inner updates disabled, the same encoder reaches 78.63 \pm 0.20. Neither model transfers zero-shot to SYSU-CD, DSIFN-CD, or WHU-CD, and ChangeT ^3 is the slower of the two at small inputs despite similar parameter and operation counts.
[CV-283] otem: Object-Centric 3D Memory for World-Consistent Video Generation
链接: https://arxiv.org/abs/2610.15807
作者: Zichen Wan,Tianyu Huang,Fan Li,Kai Wang,Hongzhi Zhang,Wangmeng Zuo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 18 pages, 8 figures, including supplementary material
Abstract:Generating coherent videos grounded in a unified underlying world remains a major challenge, particularly under large viewpoint changes, where existing methods often struggle to maintain stable representations of individual objects. To this end, we introduce Totem, an object-centric 3D memory framework for world-consistent video generation. Totem first represents each selected object as a persistent 3D memory aligned to a shared world frame, assigning it a dedicated state. Based on these memories, we design a dual-stream 3D conditioning architecture that combines object memories with scene geometry for target-view synthesis. Moreover, these object memories constrain the model’s attention with 3D structural information, thereby preserving object surfaces and relative spatial layouts. Extensive experiments on world-generation and object-centric benchmarks demonstrate that Totem generates videos depicting a more object-consistent 3D world under challenging camera trajectories, while supporting versatile applications.
[CV-284] Reference-Efficient Multimodal Bitstream-Corrupted Video Recovery via Depth-Conditioned Adaptation
链接: https://arxiv.org/abs/2610.15806
作者: Tianyi Liu,Jiaojiao Wang,Ye Lu,Kejun Wu,Chengtao Cai,Zhengguo Li,Chia-Wen Lin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 11 pages, 4 figures
Abstract:Recovering visually faithful content from damaged video bitstreams is important for reliable multimedia communication and storage. Existing methods exploit local residuals and non-local RGB references of the input video sequence, but both sources can become unreliable after decoding, while processing additional references increases receiver computation and memory. This paper studies multimodal bitstream-corrupted video recovery (MBSCVR) using RGB and depth information, where pre-transmission depth-based geometry aids reconstruction when the received appearance is corrupted. We propose a novel MBSCVR framework that selects and effectively utilizes a single selected non-local RGB reference per recovery window to achieve higher performance. It integrates a reliability-aware multimodal selector that combines valid appearance with depth compatibility before reference encoding. Depth-Conditioned Reference Adaptation (DCRA) introduces selected RGB evidence into feature completion and content generation, while frame-aligned depth conditions completion queries. On the complete BSCV benchmark with oracle masks, MBSCVR uses one selected reference to achieve 32.30 dB on YouTube-VOS and 28.93 dB on DAVIS, surpassing full-reference B2SCVR by 0.27 and 0.33 dB and improving SSIM on both subsets. Matched receiver-forward measurements show 36.8% lower latency and 40.8% lower peak memory. These results demonstrate how compact geometric guidance and adapted appearance processing improve the quality–resource trade-off of multimedia recovery.
[CV-285] Data-Centric Recursive Self-Improvement for Navigation via Policy-Synchronized Language-Action Dual Bootstrapping
链接: https://arxiv.org/abs/2610.15782
作者: Songze Li,Haoming Gu,Ziyang Gong,Zhitong Wang,Gengze Zhou,Zun Wang,Yuhan Zhu,Tianxiang Jiang,Juanzi Li,Qi Wu,Weinan Zhang,Yu Qiao,Wenwei Zhang,Yi Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-language navigation requires agents to follow natural-language instructions to reach target locations, making high-quality language supervision essential. Since human annotation is costly, prior work has explored scalable data synthesis methods. However, existing methods typically either use separate models for navigation and instruction generation or couple the two tasks through task-specific components; most also annotate the full trajectory corpus with a fixed model before updating, which can cause generated supervision to lag behind the evolving policy. Therefore, we introduce RSI-Nav, a data-centric recursive self-improvement framework for navigation via policy-synchronized language-action dual bootstrapping, which unifies vision-language-to-action instruction following and vision-action-to-language instruction generation within a shared model. Specifically, after supervised fine-tuning establishes fundamental capabilities, RSI-Nav progressively processes unlabeled trajectories in small shards. The current model generates instructions for these trajectories, executes them to evaluate goal completion and path efficiency, retains qualified samples, and immediately updates itself before processing the next shard. This shard-wise update keeps newly generated supervision aligned with the evolving policy. We also incorporate spatial understanding data, since stronger spatial reasoning benefits navigation. Extensive experiments show that RSI-Nav reaches 68.6% SR on R2R-CE and 72.4% SR on RxR-CE, while providing stronger instruction-generation utility and modestly improved spatial understanding after recursive evolution.
[CV-286] Hand2Robot: Object-Centric 4D Hand Reconstruction from Human Demonstrations for Robotics
链接: https://arxiv.org/abs/2610.15769
作者: Zetao Du,Weiqing Wang,Hanbing Zhang,Yimiao Zhou,Ye Shi,Liang Wang,Jingya Wang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: 21 pages, including appendix. Project page: this https URL
Abstract:Human demonstrations can capture task-relevant manipulation without using the target robot, but the recorded hand motions cannot directly serve as robot action supervision. Hand reconstruction and motion retargeting provide a direct route to such supervision, yet recovering hand motion alone does not ensure physically consistent hand-object interactions. To address this issue, we introduce Hand2Robot, an object-centric 4D hand reconstruction framework for robot policy post-training. Given videos of humans performing the target task, Hand2Robot extracts interaction clips, reconstructs object geometry, and generates robot action supervision through hand-object interaction optimization and motion retargeting. For object tracking, we combine observation confidence with temporal motion constraints to select a globally consistent object pose trajectory from per-frame pose candidates. For hand refinement, we use object geometry and motion as a reference and combine staged 2D alignment with object-local contact constraints to refine hand trajectories and finger joint configurations. Experiments on DexYCB and HO3D-v3 show improved object pose accuracy and hand-object interaction quality. Experiments on three real-robot dexterous manipulation tasks further show that reconstructed and retargeted human demonstrations can effectively supplement limited robot data for target-task policy post-training.
[CV-287] SlotAVS: Object-Centric Audio-Visual Segmentation via Bi-Modal Slot Attention
链接: https://arxiv.org/abs/2610.15764
作者: Arnav Sharma,Anand Mishra,Renu Rameshan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Audio-Visual Segmentation (AVS) requires predicting pixel-wise masks of sounding objects by grounding acoustic signals in visual space. In multi-source scenes, concurrent sound sources often remain entangled within shared visual representations, leading to merged masks or unstable specialization when using standard alignment-focused transformer architectures. To address this, we present SlotAVS, an object-centric framework driven by competitive bi-modal slot attention. While robust foundation models provide rich semantic priors, SlotAVS shifts the paradigm from global cross-modal fusion to explicit competitive binding. By initializing discrete object slots from temporal audio sequences and enforcing strict slot-wise competition over spatial visual tokens, our method structurally mitigates the blending of distinct acoustic entities. Further, to stabilize instance-level learning, we introduce a Hungarian matching strategy that optimally aligns slots with ground-truth objects using spatial and boundary overlap. Coupled with a numerically stable, confidence-aware mask fusion mechanism, SlotAVS robustly aggregates multi-source predictions. Extensive experiments on the AVSBench dataset demonstrate the efficacy of our approach, achieving 88.6% mIoU on the Single-Source (S4) split and 71.7% mIoU on the Multi-Source (MS3) split. These results show that explicit competitive binding is critical for complex audio-visual scene understanding and effectively leveraging foundation-model priors without relying solely on their capacity.
[CV-288] Frozen History: When Temporal Losses Stop Accumulation in Learned Real-Time Supersampling
链接: https://arxiv.org/abs/2610.15752
作者: Hao Yu,Guoying Zhao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 12 pages for main text
Abstract:Learned real-time supersampling accumulates jittered low-resolution samples along motion vectors. Many methods use a temporal consistency loss to reduce flicker. We show that with accurate registration, this loss can suppress accumulation: on static content, the network copies its warped history at almost no temporal cost. Camera jitter in motion vectors raises this cost and can hide the effect. We identify four registration errors that training can conceal: jitter-carrying motion vectors, one motion field for both resolution lattices, jittered depth, and an unregistered skip path. A model-free oracle shows that correct registration makes freezing two orders of magnitude cheaper on static pixels. Controlled training shows that registration reverses the temporal weight’s effect on PSNR. Under exact registration, increasing the weight reduces pixel variance at the expense of accuracy and perceived quality, with a 0.7 to 0.9 dB loss on static frames at weight 0.25. This reversal recurs across three network families and two scales; the loss also reduces accuracy on unjittered STSS data. With a small temporal weight and a normalised sample accumulator, a plain registered network improves the best published QRISP PSNR by 0.34 to 0.62 dB at \times 2, \times 3 and \times 4 upscaling to 1080p. At \times 2, its static pixel standard deviation is one quarter that of DLSS 2. Without retraining, it outperforms classical temporal upsampling by at least 4.4 dB on Arm’s jittered Bistro capture. Our most accurate \times 4 model runs in 2.03 ms per 1080p frame on an H200. Code: this https URL
[CV-289] SpikeOcc: Towards Resilient Multimodal 3D Occupancy Perception via Spiking Experts
链接: https://arxiv.org/abs/2610.15745
作者: Yiding Sun,Zijie Xu,Hongqiang Lin,Jiajun Cao,Dongxu Zhang,Kai Tang,Jihua Zhu,Wenxuan Liu,Zhaofei Yu,Tiejun Huang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal 3D occupancy prediction underpins safe autonomous driving through comprehensive scene understanding. However, existing methods have largely pursued higher accuracy through sophisticated fusion, leaving resilience to missing sensors underexplored. In this paper, we investigate spiking neural networks (SNNs) as a means of building more resilient multimodal perception. Still, the discrete nature of spiking neurons makes optimization challenging. To overcome this difficulty, SpikeOcc is introduced with two complementary components, a Spiking Mixture-of-Experts (SMoE) and Sensor Availability Learning (SAL). Specifically, SMoE organizes RGB, LiDAR, and joint fusion into specialized pathways, enabling full-width unimodal representations while retaining the original multimodal features via a bounded residual. Beyond sensor dropout, SAL employs a staged curriculum with explicit routing supervision, progressively improving robustness under severe sensing conditions. Remarkably, SAL also yields notable gains for conventional models, highlighting its broader potential for resilient multimodal learning. Extensive experiments demonstrate that SpikeOcc maintains competitive full-sensor accuracy while markedly raising the performance floor under sensor outages. SpikeOcc illustrates the promise of neuromorphic computing, pointing toward more reliable and energy-efficient autonomous driving.
[CV-290] RSAJEV: Training-Free Abstaining Prototype Memory for Eliminating Silent Hallucination in JEV-Based Remote Sensing Classification
链接: https://arxiv.org/abs/2610.15742
作者: Man Chen,Antai Chen,Shuai Xu,Chengjie Gu,Huaping Zhou,Kelei Sun,Tao Wu,Yongjie Huang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Scene classification of remote sensing imagery with multimodal large language models (MLLMs) is formulated as a single-forward-pass decision problem: candidate class names are lettered options in the prompt, a registered answer slot is conditioned on the prompt, and classification is read out by scoring letter tokens with the language-modeling head. We show that although this letter binding is highly effective within its training vocabulary (97.71% OA on UCM), it behaves erratically out of vocabulary: zero-adaptation accuracy across four cross-dataset directions fluctuates between 7.35% and 96.40%, and under out-of-vocabulary weights its confidence fails inversely on open-set evaluation, assigning higher confidence to unseen than to seen classes (AUROC 0.38), a failure we term silent hallucination. We retain the slot representation and replace the readout with RSAJEV, a training-free abstaining prototype memory that performs retrieval-based prototype matching: K reference images per class are cached as prototypes via their slot representations under the same prompt, and classification becomes nearest-neighbor retrieval over the prototypes, refined by a shrinkage-Mahalanobis geometric correction. This readout is mechanistically isomorphic to retrieval-augmented generation (RAG) but operates on discriminative decisions; it is the first retrieval-based readout for MLLM slot representations. The retrieval readout reaches 89.43% on the UCM cross-dataset gate with 16 references per class and zero gradient updates, and 96.90% on AID. Its ID-ness scores attain 0.91-0.99 AUROC on open-set evaluation over two datasets, exceeding symbolic confidence (0.38-0.96) in all four open-set settings, restoring a calibratable abstention channel and supporting class addition/removal at zero extra inference cost via offline rescoring of the cached matrix.
[CV-291] ScenePlacer: Reformulating Scene-Level Object Pose Estimation as Visibility Field Learning
链接: https://arxiv.org/abs/2610.15734
作者: Ken Deng,Jiarui Liu,Guibiao Liao,Kunming Luo,Mochu Xiang,Hongyu Yan,Yebin Liu,Shenghua Gao,Yizhou Yu,Ping Tan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 20 pages
Abstract:We present ScenePlacer, a framework for scene-level pose estimation of objects generated by state-of-the-art native shape generation models, placing generated 3D objects into scenes given a partial geometric observation without modifying the underlying generative pipeline. Existing scene generation approaches either generate objects directly in scene coordinates, compromising their canonical-space priors, or regress poses from complete generated shapes in the shape generation process - the latter being a poorly conditioned problem that is sensitive to scale ambiguity and partial observations. ScenePlacer decouples object generation from pose estimation by reformulating alignment as conditional visibility-field learning. Given an observed partial point cloud from reconstruction and a generated mesh, it predicts a visibility field defined in object space, then estimates the rigid transformation by aligning the partial observation to visible parts on the mesh. We propose a pretrained 3D shape VAE as our backbone, preserving its mesh prior while conditioning it on a separately encoded observation. We further introduce visibility-aware, radius-constrained supervision and synthesize training pairs using only object-level meshes and rendered views, eliminating the need for scene-level pose annotations. On 3D-FUTURE, ScenePlacer consistently outperforms existing placement and pose-refinement baselines under both ground-truth and monocular depth estimation. With generated objects, it achieves a Chamfer distance of 0.0077 and an F-score of 92.40 under estimated depth and improves to 0.0023 and 98.27, respectively, under ground-truth geometry.
[CV-292] Visually Legible but Adversarial: Watermark-Constrained Attacks on End-to-End VLMs for Document Understanding
链接: https://arxiv.org/abs/2610.15728
作者: Qingzhe Han,Fan Dang,Shengqi Chen,Xinyi Sun,Haoyu Xiong,Lingkun Li,Di Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Visible watermarks convey document ownership, confidentiality, and status, but can also carry adversarial signals. We introduce Auto-Regional Adversarial Watermarking (AR-AWM), a model-aware white-box attack on end-to-end visual language models (VLMs) for document understanding. AR-AWM constructs a small, spatially distributed watermark-text support and optimizes only its pixels to disrupt visual representations. The attack is untargeted and question-agnostic: generation uses no questions, answers, parsing prompts, language-model gradients, or language-model outputs. Each attacked page is reused across downstream queries. On Qwen2.5-VL-7B, a watermark support covering 7.98% of page pixels under the reporting threshold reduces DocVQA exact match from 0.899 to 0.150 and average normalized Levenshtein similarity (ANLS) from 0.939 to 0.212. Selected qualitative cases retain visually legible source fields. The same support without optimization and with matched random perturbations retains ANLS of 0.932 and 0.933, respectively, identifying adversarial optimization as the primary cause of degradation. Layout-matched optimized rectangles yield ANLS of 0.276, showing an incremental benefit from text-shaped geometry. AR-AWM also degrades all reported text, formula, reading-order, and table metrics on OmniDocBench. Separate white-box optimization on InternVL2.5-8B reduces ANLS from 0.922 to 0.046, while weak cross-model transfer reveals strong encoder specificity. These results identify conventional watermark-shaped supports as an adversarial surface for document VLMs in the evaluated white-box this http URL white-box setting.
[CV-293] PhySimCode: A Benchmark and Evaluation Method for Physics Video to Code Generation
链接: https://arxiv.org/abs/2610.15722
作者: Souraja Kundu,Aditya Gupta,Joel Julin,Yizhou Zhao,Liuyue Xie,Laszlo A. Jeni
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 34 pages, 8 figures, 20 tables. Project page: this https URL
Abstract:Multimodal large language models (MLLMs) are good at video understanding and code generation, but it remains unclear whether they can recover the physical laws underlying videos, a capability essential for converting videos into executable simulations and for providing structured physical supervision in robotics. Captioning motion video does not imply that a model has inferred its governing dynamics, estimated quantitative parameters, or can regenerate the observed scene from code. Existing video-physics benchmarks evaluate plausibility, property estimation, or future-state prediction, but offer limited insight into whether models can translate visual dynamics into runnable simulation code. We introduce PhySimCode, a benchmark for physics video-to-code generation: given only a raw video, a model must infer the relevant physics, recover parameter values, and generate runnable Python that reproduces the input video. The dataset contains 160.6K procedurally generated samples spanning 162 physics phenomena across analytic SciPy simulations and PyBullet-rendered rigid-body, articulated, fluid, and contact-rich scenes; each sample bundles the simulation code, physical parameters, rendered video, and a structured chain-of-thought covering observations, laws, ODEs, numerical methods, and validation checks. We evaluate ten state-of-the-art closed- and open-source MLLMs along physical-law correctness, simulation equivalence, code executability, and parameter recovery. Models reliably state physics laws (4.27-4.93/5) yet score only 1.51-2.47/5 on inferring them from video, and the best system (Claude Sonnet 4.6) recovers just 7.7% of ground-truth parameters within \pm 20% , exposing a substantial video-to-simulation gap. The 2,430-video evaluation set is available at this https URL and the evaluation code at this https URL
[CV-294] Rethinking Human Intent-to-CAD: Parametric CAD Model Generation via Cooperative Multi-Task Alignment and Spatial-Aware Reinforcement Learning ICML2026
链接: https://arxiv.org/abs/2610.15681
作者: Qingwang Zhang,Jiahao Li,Xiangdong Zhou
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted at ICML 2026. Published in Proceedings of Machine Learning Research (PMLR), Volume 306
Abstract:Parametric Computer-Aided-Design (CAD) modeling from human intent remains challenging, particularly during the conceptual design stage, where design goals are expressed through incomplete and unstructured modalities (e.g., hand-drawn sketches and textual descriptions). In this work, we rethink the human intent-to-CAD pipeline and propose a unified method that directly maps multi-level human intents to executable codes, without assuming the prior existence of target CAD models. To support our study, we construct HiCAD, the first large-scale dataset aligning hand-drawn sketches, textual descriptions, and parametric CAD codes. Based on this, we introduce HiCAD, a two-stage framework comprising Cooperative Multi-Task Alignment to bridge the representational gap between heterogeneous inputs, and Spatial-Aware Reinforcement Learning to enforce geometric and topological consistency. Extensive experiments demonstrate that our method significantly outperforms existing baselines across multiple tasks, validating its effectiveness and robustness in transforming heterogeneous human intents into high-fidelity parametric CAD models. Our project page: this https URL.
[CV-295] Recovering Low-Bit Failures in Vision-Language-Action Policies via Quantization-Conditioned Reinforcement Learning
链接: https://arxiv.org/abs/2610.15675
作者: Seongmin Park,Jaewon Lee,Yujun Heo,Yoonsik Kim,Seheon Choi,Hyoung-rock Kim,Jungwook Choi
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Quantization lowers deployment costs of vision-language-action (VLA) policies but can compromise task completion. Across five VLA models and 4,050 real-world trials, we systematically reveal that quantization-aware training (QAT) generally improves on post-training quantization (PTQ) but leaves substantial gaps to full precision (FP) on long-horizon tasks and under visual and semantic shifts, while its imitation-based objective does not explicitly use rollout success or partial completion. To exploit these outcome signals, we introduce Quantization-Conditioned Reinforcement Learning (QCRL), an offline RL method coupling progress-based advantage estimation with quantization-aware source conditioning. FP rollouts supply reference experience beyond action-level distillation, progress-scaled terminal penalty distinguishes partial completion, and quantization-aware conditioning selects advantageous reference behavior within the quantized policy. We evaluate weight-only INT4 W4A16 on \pi_0.5 and GR00T N1.5 and joint weight-activation NVFP4 W4A4 on \pi_0.5 . Using only in-distribution data, QCRL reaches 99.3% of FP’s mean score across three visual-shift conditions for \pi_0.5 with INT4 W4A16, alongside 42.9% model-memory savings. With NVFP4 W4A4, it reaches 100.7% of the same FP baseline, alongside a 1.73\times speedup using NVFP4/FP8 deployment. Hardware gains are relative to FP16 on Jetson Thor.
[CV-296] ProtoPur: Prototype-Purity Pseudo-Labeling for Weakly Supervised Instance Learning in Pathology
链接: https://arxiv.org/abs/2610.15669
作者: Chen Shu,Chengyang Zhang,Boyu Fu,Ting Yin,Wenchuan Zhang,Jie Chen,Yuhao Yi,Hong Bu
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Accepted to IEEE BIBM 2026
Abstract:Weak bag-level labels are widely available in computational pathology, whereas reliable instance-level annotations for whole-slide images remain expensive and ambiguous. Weakly supervised multiple instance learning (MIL) achieves strong bag-level classification, but instance-level supervision from attention scores, model predictions, negative bags, or spatial refinement may fail to identify class-specific local evidence. We propose ProtoPur, a prototype-purity pseudo-labeling method for weakly supervised instance-level learning in pathology whole-slide images. ProtoPur estimates class-balanced local purity over prototype regions constructed from frozen foundation-model features, and assigns pseudo-labels to high-purity regions strongly associated with a target class. This purity-first design separates class-specific evidence from negative or shared morphology, with coverage used only to ensure sufficient pseudo-label assignment. We provide a PAC-style analysis showing that the risk of an instance classifier trained with ProtoPur pseudo-labels can be decomposed into a standard uniform-convergence term and a ProtoPur-specific pseudo-label noise term, comprising selected-region mismatch and a coverage residual. Experiments on three public whole-slide image datasets with two pathology foundation encoders show that ProtoPur consistently improves instance-level performance across multiple MIL backbones while maintaining competitive bag-level performance. Pseudo-label quality comparisons on two public datasets further demonstrate that prototype-purity selection provides stronger instance supervision. These results suggest that the foundation-model features contain exploitable instance-level supervision signals, and ProtoPur is a way to reveal and use these signals from weak bag labels without requiring dense instance annotations.
[CV-297] Sparse All-Layer Connector for Domain Generalised Semantic Segmentation DATE NEURIPS2026
链接: https://arxiv.org/abs/2610.15660
作者: Ruoyu Guo,Xin Kun Lin,Maurice Pagnucco,Yang Song
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at NeurIPS 2026. The final version will be updated soon
Abstract:Domain generalised semantic segmentation (DGSS) has recently benefited from combining multiple foundation models, such as vision-language and vision foundation models, to leverage their diverse representations for generalisation. Existing multi-model approaches share a common design choice: cross-model fusion is restricted to depth-aligned layers, implicitly assuming that mutually beneficial information resides at the same depth across models. However, foundation models pretrained under different objectives may develop distinct representation hierarchies, making the optimality of depth-aligned fusion questionable. Moreover, we observe that the foundation models exhibit different depth-wise domain sensitivity. Motivated by these observations, we propose a Sparse All-Layer Connector (SALC) that enables each layer to access information from preceding layers of another model. As the number of accessible layers grows with depth, dense aggregation may hurt generalisation. SALC therefore learns to adaptively select a sparse subset of informative layers. We further introduce a candidate dropout regularisation that strengthens sparsity and encourages SALC to explore diverse layers, leading to more robust selection. Across four foundation model combinations and three DGSS evaluation settings, the proposed design demonstrates stronger generalisation than conventional depth-aligned fusion. The code is available at this https URL.
[CV-298] SurgCast: Action-Conditioned Future Skeletons for Controllable Surgical World Models
链接: https://arxiv.org/abs/2610.15645
作者: Wanhao Liu,Rulin Zhou,Liangjing Shao,Zhaocheng Lin,Dongyue Li,Jinsong Lin,Guoheng Ma,Hangheng Yang,Zhiqing Tang,Jing Chen,Panshuo Li,Hongliang Ren
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 9 pages, 7 figures
Abstract:Surgical world models must turn prospective robot actions into visually faithful, structurally consistent instrument motion. We introduce SurgCast, which predicts future instrument skeletons from current structure, robot state, and future actions, then uses the forecast to control video generation. Dual-path injection combines skeleton geometry with derived, instrument-specific motion descriptors through geometric residuals and conditioning modulation. Both paths complement native action conditioning and are retained in a four-step causal student. On SutureBot and SRTH-Porcine-Cholei, the teacher and student improve all seven reported quality and instrument-consistency metrics over their respective direct-action baselines; teacher FVD decreases by 37.6% and 67.3%. Ablations show that future actions improve skeleton prediction and that combining both injection paths performs best across all six metrics under predicted skeletons. These results support using predicted articulated structure and its motion semantics as a control interface for surgical video prediction. Project videos and examples are available at this https URL.
[CV-299] Robust Detection of Directional Adversarial Attacks in Deep Neural Networks for Radiological Imaging CVPR2026 CVPR
链接: https://arxiv.org/abs/2610.15628
作者: Tristan Malatyński,Joanna Jaworek-Korjakowska
类目: Computer Vision and Pattern Recognition (cs.CV); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
备注: Published in CVPR 2026 Workshops (SAIAD); open access: this https URL
Abstract:Deep learning is now central to radiology, helping detect changes on X-rays, CTs, and MRIs. However, these systems are highly vulnerable to adversarial attacks - small, crafted perturbations that mislead models while appearing unchanged to humans. Such errors risk missed cancers or false positives. Studies show over 90% attack success on medical scans. Almost 30% of US hospitals use AI in imaging in 2023. Thus, an effective method of detecting such attacks and their directionality is crucial to ensure the credibility and reliability of medical systems based on DNNs. We propose a novel detection framework that leverages subsequent attacks using random noise to identify adversarial perturbations. Our approach uses the analysis of variations in the Sigmoid function to distinguish genuine medical images from adversarially manipulated ones. Specifically, we compare prediction differences between clean and attacked images, as well as between different adversarially altered versions, to improve detection accuracy. We evaluated our method on three popular medical datasets including ‘Chest X-Ray Images for Classification’, ‘Retinal Fundus Multi-disease Image Dataset’ and ‘Brain Tumor Dataset’ under various attack scenarios, including random noise. Our framework achieved up to 99.8% ACC in identifying adversarial directional attacks, significantly outperforming existing defense mechanisms. It consistently identified adversarial samples across varying attack strengths while keeping false positives low, showing strong reliability and potential for clinical use. Furthermore, our proposal highlights its potential for real-world deployment.
[CV-300] 1Step-WAM: Your World Action Model but Faster using 1-Step Action Generation
链接: https://arxiv.org/abs/2610.15625
作者: Rundong Su,Pengtao Chen,Haohan Geng,Sha Zhang,Zhihang Yuan,Yicheng Rong,Chong Yu,Hao Huang,Tao Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:World Action Models (WAMs) unify visual world modeling and action generation, yet their real-world deployment is heavily bottlenecked by iterative action denoising and the sequential execution of visual prefilling and action prediction. We demonstrate that the action-denoising trajectories in WAMs are approximately straight, rendering the first-step velocity a strong estimate of the final action direction. Building on this observation, we introduce 1Step-WAM, a framework that transforms multi-step action denoising into one single forward. Specifically, we employ a lightweight calibrator to directly correct the first step velocity, and use bandwidth-constrained action decoding to further reduce errors in the calibrated one-step actions. Compressing action generation into a single step enables it to execute in parallel with visual prefilling, completely breaking the sequential bottleneck of the traditional imagination-then-action paradigm. Evaluations on RoboTwin and LIBERO show that 1Step-WAM largely preserves multi-step task success with only about 13.4M trainable parameters, while achieving inference speedups of 7.44\times , 5.98\times , and 24.84\times for FastWAM, ImageWAM and Kairos, respectively. Furthermore, real-world robotic deployments using the FastWAM backbone confirm a consistent 7.44 \times speedup while maintaining close success rate to the multi-step baseline. Code is avaliable at this https URL.
[CV-301] Lesion-Aware Attention for Mammography Segmentation with Heterogeneous Supervision
链接: https://arxiv.org/abs/2610.15612
作者: Di Dai,Bo Liu,Liwei Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Accurate lesion segmentation in mammography is important for computer-aided breast cancer diagnosis, but dense pixel-level annotations remain costly and limited. Existing public segmentation datasets usually cover only a small set of lesion types with manually delineated masks, whereas larger mammography datasets often provide only image-level diagnostic labels. This annotation mismatch makes it difficult to train segmentation models that can generalize to rare or weakly annotated lesion categories. To address this problem, we study open-vocabulary mammographic lesion segmentation under a mixed-supervision setting, where pixel-level masks are available for base lesion categories and image-level labels are available for a broader set of categories. We propose a two-branch framework that jointly learns from segmentation data and classification data. The segmentation branch learns lesion boundaries from annotated masks, while the classification branch introduces additional image-level semantic supervision. To better transfer image-level supervision to lesion-relevant regions, we further introduce a Lesion-wise Attentive Classifier that maintains category-specific feature prototypes and generates lesion-aware attention maps for foreground feature pooling. This design reduces the influence of background tissue and encourages more compact lesion representations. Experiments on multiple mammography datasets show that the proposed method consistently improves open-vocabulary segmentation baselines on both base and novel lesion categories, while reducing the gap to fully supervised models that require dense target-domain masks. Qualitative results further indicate that our method produces more complete lesion localization for small, low-contrast, and visually subtle mammographic findings.
[CV-302] Squeeze-Aware-YOLO: Early Channel Recalibration and Multi-Scale Fusion for Real-Time Traffic Sign Detection
链接: https://arxiv.org/abs/2610.15604
作者: Mizbaul Haque Maruf,Syed Samin Sadaf,Syed Tamzid Bakht,Md. Hasanul Kabir,Sabbir Ahmed
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at DICTA 2026 (International Conference on Digital Image Computing: Techniques and Applications, Sydney, Australia). 8 pages, 7 figures, 6 tables
Abstract:Traffic sign detection plays a vital role in intelligent transportation systems, supporting the safe operation of advanced driver assistance systems and autonomous vehicles. Despite significant progress in real-time object detection, accurately identifying traffic signs under challenging conditions remains difficult. Variations in lighting, adverse weather, motion blur, and partial occlusion often reduce detection performance, especially for small or visually degraded signs where important spatial information can be lost. To address these challenges, we propose Squeeze-Aware-YOLO, a lightweight traffic sign detection framework that introduces a multi-branch bottleneck module combining convolutional kernels of different sizes with a Squeeze-and-Excitation mechanism to improve feature extraction and adaptive channel weighting. In addition, we integrate the Multi-Branch Auxiliary Feature Pyramid Network and Re-parameterized Heterogeneous Multi-Scale modules to strengthen multi-scale feature representation and feature fusion while preserving real-time inference efficiency for practical deployment. The proposed approach is evaluated on the Bangladeshi Traffic Sign Detection Dataset, the CSUST Chinese Traffic Sign Detection Benchmark 2021, and the German Traffic Sign Detection Benchmark, achieving mAP@0.5 scores of 74.7%, 79.4%, and 84.0%, respectively, while consistently outperforming recent state-of-the-art YOLO-based traffic sign detectors. These findings demonstrate that combining multi-scale receptive fields, channel attention, and enhanced feature fusion enables more accurate and robust traffic sign detection while maintaining the computational efficiency required for real-world deployment.
[CV-303] FireNav: Decision-Oriented Reconstruction Evidence Routing for Vision-Language Navigation in Indoor Fire Environments
链接: https://arxiv.org/abs/2610.15595
作者: Hongbo Ding,Xiang Chen,Jiangxin Dong,Jinshan Pan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: this https URL
Abstract:In real-world indoor fire environments, visual degradation severely impairs instruction grounding, waypoint prediction, and long-horizon exploration in vision-language navigation (VLN). Although image restoration offers a natural means to mitigate such visual degradation, existing methods primarily optimize reconstruction fidelity, which does not necessarily translate into better navigation decisions and may even corrupt task-critical cues when applied indiscriminately. In this paper, we present FireNav, a decision-oriented reconstruction evidence routing framework that aligns visual restoration with navigation utility in VLN. To advance this task, we first contribute FireVLN-Bench, a paired indoor fire navigation benchmark with clean references preserving task semantics, safety supervision, and matched settings for continuous navigation and robot execution with simulated dynamics. Our FireNav retains native visual tokens and treats latent reconstruction features as auxiliary evidence, where a lightweight router leverages visual features, task context, and prior execution states to determine whether such evidence should be admitted and how strongly it should contribute to the visual tokens. To learn when reconstruction is beneficial, we further introduce decision utility calibration, which compares action prediction losses with and without candidate evidence under the same state, thereby supervising evidence selection according to its contribution to waypoint, heading, and arrival prediction. Experiments demonstrate that FireNav consistently outperforms state-of-the-art baselines in both navigation success and path efficiency across evaluation settings.
[CV-304] Ellipsoids from Tangents
链接: https://arxiv.org/abs/2610.15580
作者: Efrat Dahan,Michael Werman
类目: Computer Vision and Pattern Recognition (cs.CV); Computational Geometry (cs.CG)
备注: 8 pages, 7 figures. Written in 2016; posted in 2026 with revised exposition. Results unchanged
Abstract:This paper presents an algorithm for detecting and fitting multiple ellipsoids in 2D, 3D and higher-dimensional data. The representation is based on tangents to the ellipsoid together with their contact points, rather than on points alone. A tangent contributes d independent equations where a point contributes one, so each hypothesis needs fewer samples, which in turn reduces the number of iterations required to recover every ellipsoid. The problem is posed as sets of linear equations whose solutions – the ellipsoid parameters – are recovered robustly by clustering in parameter space. The algorithm works in all dimensions and outperforms the state of the art in 2D. The approach applies in general to the robust solution of linear equations, and so has utility beyond ellipsoid fitting.
[CV-305] runcation Samplers Are Nested: Comparing Decoding Rules Is a Measurement Problem
链接: https://arxiv.org/abs/2610.15570
作者: Rasul Khanbayov,Hasan Kurban
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Settling one comparison between two truncation samplers has cost six thousand A100-hours, and nothing says in advance how far apart two samplers can be. We observe that almost every deployed truncation rule keeps a prefix of the probability-sorted vocabulary, so any two such rules are nested at every decoding step. Nesting yields a closed-form per-step distance and a coupling under which both samplers emit identical text until they first disagree. Accumulated over a generation, this gives a separation certificate that bounds the difference between two samplers on any bounded outcome, is computable from logs a comparison already produces, and equals their sequence-level total variation distance exactly when neither rule ever overtakes the other. Two consequences follow without measurement: exactness is fragile, and certifying a small difference requires the two retained masses to agree to very high precision at every step. On six open-weight models over four math and science benchmarks, the certificate for every hyperparameter grid we tested is at least 0.998, so a bound alone rarely settles a comparison, although two members of one family can be far closer. What remains useful is the coupling, which gives an unbiased paired estimator that cuts variance 1.2 - 3.8\times against prompt-level pairing. Read against benchmark sizes at the default pair’s variance and our point estimate of its between-prompt component, resolving a one-point difference needs 8 to 32 samples per prompt on GSM8K, MATH-500 and GPQA-Diamond and about 210 on AIME 2025; at that component’s upper confidence bound GPQA-Diamond needs more than 32 and AIME 2025 cannot be resolved at any sample count. Choosing a truncation sampler is a measurement problem at least as much as a design one. The code is available at this https URL.
[CV-306] Genesis: Visual Augmentation Primitives for Multimodal Chain-of-Thought Reasoning
链接: https://arxiv.org/abs/2610.15565
作者: Fan Yang,Lifeng Chen,Yipu Wang,Han Wang,Feilong Chen,Yidong Ma,Zhiyang Chen,Chaoyang Zhao,Lingxiang Wu,Hanqi Jiang,Yibo Chen,Xu Tang,Yao Hu,Ming Tang,Xiaojie Jin,Xiaolong Zheng,Jinqiao Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Recent multimodal large language models achieve promising results on visual reasoning tasks, approaching human-level performance on existing STEM benchmarks. However, these capabilities degrade under complex real-world scenes. We identify that chain-of-thought reasoning remains largely confined to the linguistic space, rarely anchoring each step in visual evidence. The main cause is the absence of such data during training. In this work, we introduce a training data augmentation pipeline, \textbfVisual Augmentation Primitives (VAP), which leverages state-of-the-art vision generative models to synthesize diverse photorealistic training samples and corresponding reasoning text. VAP introduces some basic operators and applies them to a seed image and gets a variant. It also constructs coherent multimodal reasoning chains throughout the generation process. Such training data force models to perceive visual content and perform logical reasoning according to them at every step. We further introduce multimodal chain-of-thought rubric-based reinforcement learning to enhance this. It rewards not only text reasoning but also visual reasoning by scoring intermediate reasoning steps for logical validity, completeness, and consistency. Moreover, we construct the UNREAL benchmark to systematically evaluate visual reasoning on hyperrealistic scenes across controllable difficulty levels. Experiments show that a 4B model built on this framework surpasses 8B baselines, with performance improving consistently as the model scales to 32B, validating the scaling law for visual reasoning.
[CV-307] A Plausible Future Is Not a Causal Future: Can Video Models Generate How the World Unfolds?
链接: https://arxiv.org/abs/2610.15562
作者: Jialiang Li,Zhengning Xu,Zongyang Ma,Shaobo Cui
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 81 pages, 17 figures
Abstract:Large-scale video generation models are increasingly viewed as a promising path toward world models, yet their ability to capture causally coherent physical dynamics remains poorly understood. Existing evaluations largely emphasize visual quality, prompt adherence, or broad physical plausibility, leaving open whether a model can faithfully realize a complete causal cascade, in which an initiating event propagates through causally dependent interactions to a terminal outcome. We introduce CausalCascade, a simulator-grounded benchmark centered on 50 event-rich Blender scenes with structured event graphs, graph-grounded diagnostic VQA, three levels of future-event specification, and 2,832 controlled counterfactual variants, and use it to evaluate nine image-to-video models with complementary motion-, event-, and counterfactual-level diagnostics. Across the evaluated models, generated motion is more reliably localized spatially than realized with correct spatiotemporal evolution, while richer future-event specification improves event recovery but leaves physical realization limited. Counterfactual evaluation further reveals weak selective responsiveness to physical interventions, while most audited failures are foundational, arising at scene consistency or event occurrence. Together, these results position causal cascade generation as a process-level test of world modeling and highlight the need to jointly recover the intended future, faithfully realize its physical evolution, selectively respond to physical interventions, and maintain a coherent physical world over time. Code and dataset are available at this https URL .
[CV-308] ELECT: Settling Identity Before Rendering for Scalable Story Visualization
链接: https://arxiv.org/abs/2610.15555
作者: Mehrajul Abadin Miraj,Md. Mahmudul Hasan,Meherun Farzana,Aniket Joarder,Mahmudul Hasan,Abir Chakraborty Partha,Md. Ahasanul Alam,Redwan Ahmed Rizvee,Md. Fahim Arefin,Md Mahmudur Rahman,Md. Tanvir Alam,Md Mosaddek Khan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Training-free story visualization without reference images must preserve recurring entities across frames while scaling to long narratives. Existing methods typically maintain consistency by conditioning each frame on other frames of the same story, coupling generation time, VRAM, or both to story length. We introduce ELECT, which instead settles cross-frame identity before final rendering through a language-guided scene-entity mapping and stationary anchors elected from the model’s own drafts. This makes the per-frame cost of the final rendering pass independent of story length. Joint attention exceeds 96 GB of VRAM at 32 scenes, whereas ELECT remains at 8.3 s per frame and under 20 GB VRAM from 10 to 1000 scenes. On the Story50 long-story benchmark, ELECT reaches 0.833 average Character Consistency Similarity (aCCS), compared with 0.826 for ten-round refinement, at one seventh of its generation cost, with lower text alignment (0.327 vs. 0.340). At 1000 scenes, to our knowledge, the first evaluation of training-free story visualization at this length, ELECT exceeds ten-round refinement by 0.017 aCCS and the nearest flat-cost baseline by 0.091, while also surpassing refinement in text alignment at one twenty-fourth of its generation time. These results show that long-horizon consistency need not require increasingly expensive inter-frame conditioning as stories grow.
[CV-309] UMAS3D: Unified Mask-Free 3D Editing with Region-Adaptive Source Preservation
链接: https://arxiv.org/abs/2610.15533
作者: Xianlin Zeng,Xiaolong Shen,Ruijie Quan,Yi Yang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 27 pages. Project page: this https URL
Abstract:3D editing aims to modify an existing 3D asset according to user intent while preserving geometry and appearance outside the intended edit region. Existing methods often rely on task-specific training, source-3D inversion, or external 3D masks, and lack a unified framework for diverse editing conditions and tasks. We propose UMAS3D, a unified 3D editing framework that is training-free, inversion-free, and free from external 3D masks. UMAS3D converts text, bounding boxes, reference images, and their combinations into common target-view guidance, after which addition, deletion, modification, appearance, and pose editing share the same downstream 3D generation process. Our key insight is to couple online edit-region estimation with region-aware source preservation. Mask-Free Adaptive Region Modeling (MARM) initializes an Edit/Keep prior from source-target visual differences and progressively refines it as the target structure evolves, while Source-Guided Preservation Steering (SGPS) constrains the current Keep-region states to retain source geometry and appearance without suppressing target-driven changes. Experiments show that UMAS3D supports addition, deletion, modification, appearance, and pose editing while effectively preserving unedited source content. These advances make UMAS3D a unified framework for flexible and source-preserving 3D editing across diverse editing conditions.
[CV-310] GNFormer: Gaussian Neural Field Transformer for Efficient Camera-based Occupancy Prediction
链接: https://arxiv.org/abs/2610.15522
作者: Guangcheng Chen,Li He,Hong Zhang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: The code will be released in this https URL
Abstract:Camera-based occupancy prediction provides geometric and semantic scene understanding for robot perception and has been adopted in autonomous vehicle and UAV perception systems. Gaussian primitives offer a promising alternative by potentially representing only occupied regions, thereby concentrating computation and representational capacity where needed and reducing wasted computation and memory in empty space. However, existing Gaussian-based methods suffer from poor primitive utilization, as ineffective training objectives leave some Gaussians in empty space, while the limited expressiveness of Gaussian functions restricts their ability to model complex shapes. To address these limitations, we propose a novel 3D representation, Gaussian Neural Field (GNF), which can accurately map voxel positions to semantic occupancies. In GNF, Gaussians and a neural network jointly parameterize the field, where the Gaussians define its spatial support and provide features for neural decoding, while the neural network enables more expressive occupancy modeling than Gaussian functions alone. We further develop the Gaussian Neural Field Transformer (GNFormer), a streamlined and flexible method that predicts GNFs for camera-based occupancy prediction. Across Occ-ScanNet, SSCBench-KITTI-360, and nuScenes-SurroundOcc benchmarks, GNFormer achieves the highest IoU and mIoU and the fastest inference among the compared single-frame methods. In the experiments, GNFormer can represent indoor and outdoor scenes with only 256 and 1024 Gaussians, respectively, substantially fewer primitives than prior Gaussian-based methods. The code will be released.
[CV-311] Semantically Steerable Image Tokenization
链接: https://arxiv.org/abs/2610.15515
作者: Danilo de Goede,Mohammad Mahdi Derakhshani,Yuki M. Asano,Cees G. M. Snoek
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Adaptive image tokenizers compress an image into a variable number of tokens matched to its complexity, and recent work makes this allocation controllable: conditioned on a target reconstruction error, the tokenizer itself decides how many tokens an image needs. However, the fidelity measure guiding this allocation shapes which aspects of the image are preserved under compression. The pixel-wise error used in prior work is blind to the spatial structure that carries an image’s meaning, and under heavy compression such tokenizers are prone to abruptly discarding tokens that encode essential content. We introduce GIST (Guided Image Semantic Tokenizer), an adaptive tokenizer that can be steered toward a requested semantic fidelity. Its token allocation is jointly controlled by pixel-level and semantic target errors, with semantic error measured in the feature space of a frozen DINOv3 encoder. This joint control enables semantic protection: with the semantic target fixed at zero, the reconstruction target can be relaxed to discard pixel detail while preserving what the image depicts. In a controlled comparison on ImageNet-100, GIST outperforms its pixel-controlled counterpart KARL at every token count, with up to 30% lower LPIPS and 37% lower DreamSim. On average, lowering its semantic target reduces LPIPS 2\times faster per token than lowering the reconstruction target, and up to 5\times in the low-token regime. Semantic fidelity is thus an effective control signal for adaptive tokenization: it lets a tokenizer see, and so preserve, what an image depicts.
[CV-312] Structured Stateful Relational Scripts with a High-Quality Dataset for Multi-Shot Audio-Visual Video Generation
链接: https://arxiv.org/abs/2610.15512
作者: Chen Zhao,Lehan Shen,Qilong Huang,Hongyu Li,Zhuoliang Kang,Xiaoming Wei,Jian Yang,Ying Tai
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Coherent multi-shot audio-visual video generation requires maintaining consistent identities while depicting evolving states and coordinating sounds across shot boundaries. Although dense captions describe individual shots in detail, they rarely specify which entities persist, how their states change, or which entities and actions produce sounds. To address this limitation, we propose Structured Stateful Relational Scripts (SSRS), a textual supervision representation that separates persistent entity and voice identities from shot-specific visual and auditory states. SSRS explicitly encodes sound-source bindings and temporal relations, while a global audio-visual event sequence connects local actions and sounds, including events that continue across shot boundaries. We further introduce MSAV-Script-100K, a large-scale dataset of curated multi-shot audio-visual videos paired with detailed SSRS annotations, enabling models to learn identity persistence, state transitions, and event continuity from structured supervision. Experiments demonstrate improvements in cross-shot consistency, adherence to specified state changes, and audio-visual alignment, highlighting the effectiveness of our SSRS and MSAV-Script-100K for coherent multi-shot audio-visual generation. Our proposed dataset is available at \hrefthis https URLMSAV-Script-100K.
[CV-313] Semantic Saliency from Multi-Modal Large Language Model Scene Understanding Maps
链接: https://arxiv.org/abs/2610.15467
作者: Ziqi Wen,Shravan Murlidaran,Jonathan Skaza,William Wang,Miguel P. Eckstein
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Image-computable low-level image saliency has shaped perceptual psychology and computer science, but is limited by its inability to capture human high-level cognitive factors that influence attention and eye movements. Here, we propose a new fully automated method to compute semantic saliency by quantifying each object’s contribution to scene comprehension without fitting any parameters to human data: the Automated Scene Understanding Map (AUTO-SUM). AUTO-SUM automatically segments and removes individual objects from a scene, uses multimodal large language models (MLLMs) to describe the original and modified scenes, and measures the semantic similarity between these descriptions using text embeddings. We show that AUTO-SUM can predict which object is most important in a human-measured scene understanding map (H-SUM) estimated using human scene descriptions. We also show that AUTO-SUM can predict the object most fixated by human observers during free viewing and scene description tasks better than a low-level saliency model (GBVS, Itti-Koch), a saliency model trained based on human meaning ratings (Meaning Map), and comparable to DeepGaze, which is trained on human fixation data. We contend that AUTO-SUM can be used as a semantic saliency model that complements lower-level saliency models. Code are available at this https URL .
[CV-314] EmoTrace-Bench: Diagnosing Emotional Expression in Text-to-Video Generation through Controlled Cues and Intensity
链接: https://arxiv.org/abs/2610.15452
作者: Yeheng Wang,Liudan Shi,Yi Zheng,Jingke Li,Shiyuan Liu,Xin Tao,Yuanxing Zhang,Guoyu Sun
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 30 pages. Code and data: this https URL
Abstract:Text-to-video (T2V) models are increasingly used to generate videos with targeted emotional expressions, yet existing evaluation methods mainly rely on holistic video-text alignment and provide limited insight into whether failures arise from target-emotion expression, intensity control, or specific affective cues. We introduce EmoTrace-Bench, a fine-grained diagnostic benchmark that controls emotion intensity and selected affective cues in prompts and evaluates generated videos at multiple levels. EmoTrace-Bench contains 448 controlled prompts covering eight emotions, two intensity levels, and diverse cue conditions. Using 12 T2V models, we generate 5,376 videos and evaluate target-emotion presence, perceived intensity, and ten audiovisual dimensions of emotional expression with an automated evaluator calibrated and validated against human annotations on a subset of the data. Results reveal substantial differences across models and prompt designs. Prompts that explicitly describe a character’s emotional reaction are followed more reliably than prompts that describe only events or scenes, whereas adding more affective cues or using more explicit cinematic audiovisual language does not consistently improve emotional generation. These results illustrate how aggregate alignment scores may overlook localized differences in target-emotion expression, intensity control, and cue realization. EmoTrace-Bench provides a traceable and interpretable framework for diagnosing emotional video generation without modifying existing generation pipelines, and we release the prompts and the automated evaluation rules.
[CV-315] Rethinking Supervision in Document Unwarping: A Self-consistent Flow-free Approach
链接: https://arxiv.org/abs/2610.15445
作者: Shaokai Liu,Hao Feng,Wengang Zhou
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by TCSVT 2023
Abstract:In recent years, the proliferation of smartphones has led to an upsurge in the digitization of document files via these portable devices. However, images captured by smartphones often suffer from distortions, thereby negatively affecting digital preservation and downstream applications. To address this issue, we introduce DRNet, a novel deep network for document image rectification. Our approach is based on three key designs. Firstly, we exploit the intrinsic geometric consistency inherent in document images to guide the learning process of distortion rectification. Secondly, we design a coarse-to-fine rectification network to leverage the representations derived from the distorted document image, thereby enhancing the rectification result. Thirdly, we propose a unique perspective for supervising the learning of rectification networks, where undistorted document images are employed for supervision, which is free of warping mesh as ground truth in existing methods. Technically, both low-level pixel alignment and high-level semantic alignment jointly contribute to the learning of the mapping relationship between deformed document images and distortion-free ones. We evaluate our method on the challenging DocUNet Benchmark dataset, where it sets a series of state-of-the-art records, demonstrating the superiority of our approach compared to existing learning-based solutions. Additionally, we conduct a comprehensive series of ablation experiments to further validate the effectiveness and merits of our method.
[CV-316] Harnessing SPAD Noise: Test-Time Gated Attention for Indoor Depth Estimation
链接: https://arxiv.org/abs/2610.15431
作者: Yiquan Gao
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Robotics (cs.RO); Signal Processing (eess.SP)
备注:
Abstract:Monocular depth estimation (MDE) has emerged as a cost-effective alternative to hardware-dependent 3D sensing modalities, yet predicting precise metric depth remains a persistent challenge due to intrinsic scale ambiguity, edge bleeding, and limited zero-shot generalization in unseen environments. Recent multi-modal approaches incorporate 1D/3D Single-Photon Avalanche Diode (SPAD) data to resolve these ambiguities, but they mainly treat SPAD noise as a harmful artifact required for denoising or leave the noise ignored. In this paper, we present an essentially different insight: native SPAD noise is an informative prior that implicitly encodes deterministic environmental characteristics. Driven by this paradigm shift, we introduce a novel, lightweight Test-Time Gated Attention framework by Harnessing SPAD Noise, therefore dubbed HSN. Tailored for indoor scenarios, HSN leverages the synergistic combination of 1D SPAD histograms and Gate Adapters (GAs). Through three complementary optimization phases, it generates channel-wise gated attention vectors to dynamically adjust encoder feature maps at inference time, effectively resolving ambiguous scale and boundary challenges without the heavy computational overhead or dense 3D modalities that other TTA methods suffer from. Extensive experiments on challenging real-world indoor benchmarks show that HSN achieves state-of-the-art metric depth accuracy in in-domain settings, while exhibiting even greater potential under zero-shot generalization to unfamiliar environments. Furthermore, interpretability analysis reveals that the proposed gated attention mechanism adaptively shifts attention toward deeper semantic layers under intensifying SPAD noise. Code will be made publicly available.
[CV-317] OmniRSI: Taming Latent Diffusion for Unified Multimodal Remote Sensing Image Generation
链接: https://arxiv.org/abs/2610.15414
作者: Hebaixu Wang,Jing Zhang,Haonan Guo,Di Wang,Jiayi Ma,Bo Du
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Under review. Code and datasets will be provided in the this https URL
Abstract:Multimodal remote sensing image generation emerges as a promising paradigm for recovering complete and complementary observations from heterogeneous and partially observed remote sensing data. However, existing methods predominantly depend on sensor-specific fine-tuning of natural image generation models, limiting unified multimodal generation capacity due to the absence of shared representations and robust conditioning mechanisms. To address these challenges, we propose OmniRSI, a unified multimodal remote sensing image generation model. Specifically, we develop a remote sensing-specific variational autoencoder that captures multimodal priors and promotes a locally smooth latent space for unified representation and robust reconstruction across diverse observed data. Building upon that, we fully adapt Stable Diffusion 3.5 with a ControlNet branch to support unified multimodal generation and translation conditioned on textual and visual cues. Besides, we devise a training-free cyclic latent alignment mechanism that iteratively enriches incomplete visual conditions with generated multimodal information, progressively improving guidance reliability while preserving observed measurements. We also construct RSI-MM, a large-scale dataset comprising 11.5M multi-resolution images with text annotations and partially paired data for training and evaluation. Extensive experiments verify that OmniRSI achieves superior performance in multimodal reconstruction, generation and translation.
[CV-318] A Decade of PhotoMappers: A Longitudinal Study of Crowdsourced Disaster Photos Geolocalization with Vision-Language Models and Geospatial Reasoning
链接: https://arxiv.org/abs/2610.15403
作者: Hao Li,Wenping Yin,Fabian Deuser,Jia Jia,Ziqi Liu,Levente Juhász,Fan Zhang,Filip Biljecki
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Crowdsourced disaster imagery provides timely, ground-level observations when conventional disaster mapping is delayed or incomplete, yet its geographic value depends on where observations are produced, how they are interpreted, and whether they can be reliably located as actionable Geospatial Intelligence (GEOINT). Using more than a decade of PhotoMappers, a crowdsourced disaster-mapping initiative, this study examines the changing geography of disaster information and asks how emerging GeoAI models compare with human geographic interpretation. We combine longitudinal spatial analysis with cross-view geolocalization using street view imagery (SVI) and remote sensing imagery (RSI), and evaluate vision-language models (VLMs) reasoning through 30 indicators spanning geographic cues and reasoning evidence. The results reveal a geographically and temporally uneven archive shaped by changing disaster types, Community Lifelines, and operational priorities. Integrating VGI, SVI, and RSI with Sinkhorn re-ranking achieves 40.90% R@1, 61.10% R@5, and 69.45% R@10, while yearly R@1 varies markedly from 25.00% to 65.07%. Geographic reasoning is similarly uneven: Qwen3.6-27B achieves an overall agreement of 67% compared with 51% for InternVL3-8B, with both models showing greater difficulty with localized and place-specific evidence than broad spatial configurations. We argue that the value of GeoAI lies less in replacing volunteered geographic interpretation than in complementing it, scaling candidate retrieval while retaining human judgment for ambiguous and consequential cases. This provides a pathway toward more human-centered and responsible GeoAI for disaster mapping. All data and code are openly available at this https URL.
[CV-319] From Acoustic Sensing to Physical Actions: A Comprehensive Survey of Security Risks in Edge-Deployed Audio LLM -Based Agents
链接: https://arxiv.org/abs/2610.15390
作者: Binhong Wang,Haidong Kang,Yanfeng Zhang,Hong-Ning Dai,Guo Yu,Shangce Gao
类目: Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 20 pages, 8 figures
Abstract:Audio Large Language Models (Audio LLMs) are embedded in agents for smart homes, vehicles, wearables, and embodied systems, where acoustic inputs can propagate from perception into persistent state and consequential execution. This setting extends security analysis beyond model behavior to the system conditions that determine how acoustic inputs can influence agent behavior and external actions. Existing surveys examine acoustic security, LLM-based agents, and edge deployment in isolation, leaving their security interactions in deployed Audio LLM-based agents insufficiently characterized. This survey introduces an Acoustic Sensing-to-Action (S2A) system model that traces these security implications along the path from acoustic sensing to external execution, while accounting for edge-cloud runtime and lifecycle transformation. We use the model to synthesize attacks, defenses, and evaluation methods across the acoustic, agent, and edge-system literature. The synthesis shows that security gaps can emerge within components and at the transitions between stages, where protections developed for one stage do not necessarily provide end-to-end assurance in the deployed system. These findings motivate security evaluation that follows acoustic influence from sensing to action across persistent state, system boundaries, and changing deployment conditions in real-world agents.
[CV-320] VideoEvoHarness: Self Evolving Agent Harness from Unlabeled Videos for Video Understanding
链接: https://arxiv.org/abs/2610.15389
作者: Weixin Xu,Yifan Liu,Yi Wang,Mingfei Rong,Zhenyu Yang,Zhisheng Zhong,Shengsheng Qian,Changsheng Xu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Video understanding agents rely on a harness to acquire evidence, use tools, and reason, but improving it requires substantial effort. Recent self evolution methods seek to optimize harnesses through execution feedback to reduce human intervention. However, existing methods often rely on annotated video questions and answers or focus on updating model parameters, leaving harness evolution from unlabeled videos underexplored. VideoEvoHarness jointly evolves a Questioner and a Solver harness from unlabeled videos with frozen model parameters. The Questioner constructs questions and computes reference answers directly from video memory records. Question quality assessments and Solver execution trajectories guide question generation policy revisions. The Solver supports evolving skills, strategies, hyperparameters, and tools using execution feedback. Executable strategies define workflows in code, reducing repeated interpretation of natural language instructions. Each round, candidate harnesses are compared on identical validation questions, with validated improvements retained for later use. Experiments on benchmarks and a dedicated benchmark question set, using videos disjoint from evolution videos, show higher overall accuracy than initially and transfer gains across backbones. Overall accuracy on this set also improves progressively across rounds. This supports our foundational framework for harness evolution from unlabeled videos, opening a promising direction for self evolving video agents. Code is available at this https URL.
[CV-321] UniForge3D: Joint 3D Reconstruction and Generation in a Shared Anchor-Guided Space
链接: https://arxiv.org/abs/2610.15357
作者: Wenjue Chen,Wenyue Chen,Heliang Zheng,Rongfei Jia,Yuan Liu,Meng Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project webpage: this https URL
Abstract:Generating complete 3D assets from sparse, unposed images requires preserving observed geometry while completing unseen regions. Joint reconstruction and generation offers a way to address both goals: reconstruction recovers the geometry observed in the input images, while generation contributes learned shape priors. However, combining camera-relative reconstruction with canonical-space generation introduces a coordinate mismatch that makes direct geometric interaction more difficult. Moreover, conditioning each 3D location on all views without accounting for their relevance may mix useful geometric evidence with unrelated image content, hindering local detail recovery. We present UniForge3D, which jointly models reconstruction and generation in a shared anchor-guided space, resolving their coordinate mismatch. This space follows the anchor view’s azimuth while keeping the object upright. The two branches exchange features bidirectionally to predict per-view point maps and a complete sparse voxel structure. We further condition point-map prediction on the generated structure to establish pixel-to-voxel correspondences and select the most informative input view for each voxel. Features from the selected views then guide detailed shape generation, helping preserve observed geometric details. Our method achieves state-of-the-art appearance and geometry results on Toys4K and GSO, generating complete 3D assets with improved fidelity. Project webpage: this https URL
[CV-322] Video-Based Sexual Harassment Detection Using Fine-Tuned Video Transformers
链接: https://arxiv.org/abs/2610.15354
作者: Jubayer- Al- Mahmud,Md Shamimul Islam,Md Nur Hossain
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Sexual harassment represents a pervasive societal challenge, with a substantial proportion of incidents remaining unreported due to insufficient evidence and delayed intervention mechanisms. Smart surveillance systems capable of detecting such incidents in real-time could assist law enforcement and serve as a deterrent. However, video-based harassment detection poses significant challenges, including variations in clothing, illumina- tion, partial occlusions, low resolution, and diverse camera view- points. Despite its social importance, this problem has received limited research attention, primarily due to the scarcity of large- scale annotated video datasets. To address this gap, we extend the Sexual Harassment Video (SHV) dataset to 600 short video clips sourced from YouTube and annotated into harassment and non-harassment categories using consistent labeling guidelines. We fine-tune three state-of-the-art video transformer models, VideoMAE, ViViT, and TimeSformer, pretrained on Kinetics- 400, for binary harassment classification. Experimental results show that all three models achieve competitive classification performance, with TimeSformer attaining the highest test accu- racy of 92.27%, outperforming VideoMAE 91.71% and ViViT 91.16% on the extended SHV test set. Our analysis reveals that performance is notably influenced by dataset size, video quality, frame sampling strategy, and model architecture. This work presents a comprehensive evaluation of modern video transformers for this sensitive real-world task.
[CV-323] Across Viewpoints and Time: Trajectory-Generalizable Physical Attacks on World Models and Video Understanding Models
链接: https://arxiv.org/abs/2610.15343
作者: Song Xia,Meiwen Ding,Chenqi Kong,Wenhan Yang,Xudong Jiang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Project page: this https URL
Abstract:World models and video understanding models are increasingly deployed in dynamic visual environments, yet their robustness to physical adversarial threats remains largely underexplored. We study whether a single physical adversarial texture can mislead the victim model across different real-world scenarios where observing viewpoints vary along uncontrolled object–camera trajectories. We identify that heterogeneous video semantics and trajectories induce poorly aligned gradients for optimizing the adversarial texture. To address that, we propose the Trajectory-aware adversarial Alignment and Refinement Attack (TARA), which incorporates trajectory-aware optimization to capture the evolving geometric variations of a printed texture under real-world scenarios. To improve cross-instance optimization compatibility, TARA incorporates trajectory-aware contrastive alignment to encourage attacked representations from diverse video–trajectory instances to approach a shared adversarial region and then performs trajectory-aware task refinement to translate the aligned representation into task-level prediction errors. For reproducible evaluation, we develop a geometry-grounded tracking pipeline that provides physically plausible adversarial regions with temporally consistent trajectories on EPIC-KITCHENS-100 and Something-Something V2. We further collect real-world print-and-recapture videos to validate attack effectiveness under physical deployment conditions. Together, the annotated benchmarks and real-world captured data form a reproducible testbed for future evaluation. Notably, our results reveal that a single printable adversarial texture can persistently disrupt world models and video understanding models under physical conditions, highlighting the fundamental security challenge.
[CV-324] StimFlow: Synthesizing Brain Activity from Images via Stimulus-conditioned Flow Matching
链接: https://arxiv.org/abs/2610.15341
作者: Bui Van Hung,Dang Thanh Vu,Hoang-Son Vo,Mai Duc Minh Huy,Mai Tien Dung,Soo-Hyung Kim
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:The inherent variability of fMRI responses across trials encourages a generative view of the stimulus-to-response mapping. Prior generative encoders often operate in compressed latent spaces and condition predominantly on semantic features, which provide limited low-level cues for early visual cortex. We propose StimFlow, which synthesizes fMRI directly in native voxel space using conditional flow matching with a linear path. This design aligns training and inference distributions, enabling competitive performance with a single sample and reducing reliance on averaging. To match the visual hierarchy, StimFlow combines structural DINOv2 features with semantic CLIP features, which prove complementary along the cortical processing stream. With a single sample, StimFlow achieves a per-voxel Pearson correlation of 0.371, matching baselines that average five draws. Generation in voxel space allows direct ROI analysis and concept localization without a decoder. Furthermore, it recovers over half of full-data performance from just one hour of scans, supporting efficient cross-subject adaptation.
[CV-325] BaT-SAM: Introducing Inductive Bias into SAM via a Mixture of Convolutional Experts for Brain Tumor Segmentation
链接: https://arxiv.org/abs/2610.15330
作者: Sumiaya Jannat,Md Mahedi Hasan,Md Shamimul Islam,Md Nur Hossain
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Accurate brain tumor segmentation from magnetic resonance imaging (MRI) is essential for diagnosis and treatment planning; however, manual delineation is labor-intensive and time-consuming. Existing approaches are either task-specific with limited generalization or rely on user-provided prompts during inference. We present BaT-SAM, a parameter-efficient adaptation framework that enables fully automatic brain tumor segmentation with the Segment Anything Model (SAM). BaT-SAM injects local inductive bias into SAM’s frozen Vision Transformer (ViT) encoder via Conv-LoRA, where image features are processed by a sparsely gated mixture of multi-scale convolutional experts. This design compensates for the limited translation-equivariant inductive bias of vanilla ViTs while fine-tuning fewer than 5M parameters (less than 1% of the full model). To remove the need for user prompts, we further introduce a lightweight self-prompting box regressor that is jointly optimized with the adapter using scheduled sampling and predicts tumor locations directly from image embeddings. On the BRISC2025 benchmark, BaT-SAM achieves a weighted mean IoU of 0.832 while outperforming state-of-the-art benchmarks. It further demonstrates strong cross-dataset generalization, achieving Dice scores of 0.702 on Figshare and 0.784 on the LGG MRI Segmentation dataset.
[CV-326] CIViC: Evidence Gates for Audio-Conditioned Visual Prediction on Proxy Tasks
链接: https://arxiv.org/abs/2610.15327
作者: Mengzhe Geng,Max W. Y. Lam,Yuewen Cao,Shoukang Hu,Jing Xu,Marc Tessier,Xu Li
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Sound (cs.SD); Audio and Speech Processing (eess.AS)
备注:
Abstract:Removing audio can shift a visual predictor’s output without showing whether the task uses the intended audio information. We introduce Conditional Input and Visual Interpretation Checks (CIViC), a four-gate protocol that links an interpretation to an intervention, held-out task comparison, or endpoint control. We demonstrate CIViC on actor-disjoint face-raster tests, a synthetic audio-feature mixture, and source-video-disjoint SpeakerVid proxy tasks. In the static face-raster test, cross-actor audio replacement meets the registered detection rule with a small effect; on a separate temporal endpoint, constant-audio fits score better than audio-trained fits in four matched seeds. In the synthetic mixture, the condition-augmented fit has higher 1,000-step RMSE than the audio-only fit (0.1807 vs. 0.1707), with the gap shrinking during training. In the SpeakerVid proxy, the five-feature-zero fit (five low-level audio summaries zeroed; identity, dialogue, and condition inputs retained) has lower whole-frame grayscale RMSE than aligned audio across four held-out conditions and five paired seeds, while aligned audio has lower segment-drift error. The results show that output activation, held-out error, and endpoint controls support different interpretations across these proxy tasks.
[CV-327] mpoKV: Predictive KV Cache Quantization for Autoregressive Video Generation
链接: https://arxiv.org/abs/2610.15324
作者: Jiahui Hu,Weile Mo,Peiliang Cai,Jiacheng Liu,Jinkui Ren,Xiantao Zhang,Xiaobing Tu,Linfeng Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Autoregressive (AR) video diffusion models enable streaming generation of long videos but suffer from significant memory overhead due to the rapidly growing KV cache. Existing KV cache quantization methods are primarily designed for LLMs and transfer poorly to video models, where activations exhibit different statistics and strong spatiotemporal redundancy. We observe that the video KV cache is largely predictable over time, with a clear asymmetry. Keys are highly predictable from their temporal history, while values depend primarily on the current key and need a short history of preceding values. Based on these observations, we propose \textbf\method, a training-free predictive KV cache quantization framework for AR video generation. \method predicts keys and values according to their distinct dependencies and stores only the residuals at low precision. Both predictors are lightweight, calibrated once offline by linear regression on a small set of BF16 samples, and remain fixed across prompts and generation runs. Experiments on Self-Forcing, Causal Forcing, and LongCat-Video-13B show that \method consistently achieves higher generation quality than existing KV cache quantization methods at comparable or higher compression ratios, reaching up to 6.71\times KV cache compression with the best overall quality–efficiency tradeoff. Our code will be released on GitHub.
[CV-328] Channel Match: Calibrating Shape Latents for Few-Step 3D Generation
链接: https://arxiv.org/abs/2610.15321
作者: Di Yang,Zhihao Li,Yanhai Xiong,Yufei Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Recent native 3D generators achieve high-quality asset creation by modeling structured 3D latents with flow-based transformers, but their high-resolution generation remains computationally expensive due to costly iterative sampling. In this work, we introduce a few-step distillation framework for native 3D generation, enabled by continuous self-consistency training with a custom JVP FlashAttention kernel and an analytic velocity regression objective for sparse structured 3D latents. To stabilize joint optimization, we further introduce an adaptive loss weighting strategy that automatically balances them without manual tuning. Beyond acceleration, we reveal a previously overlooked failure mode of few-step native 3D generation: sampling-induced latent distribution drift preserves geometric correctness but disrupts downstream PBR texture synthesis, causing texture artifacts on geometrically valid assets. To address this issue, we propose Channel Match, a training-free adaptive per-channel latent calibration method that restores the latent distribution without modifying either generator. We validate our approach on TRELLIS.2 at the 1024^3 resolution setting. Our two-step model achieves 4.5\times faster geometry sampling while maintaining geometry quality comparable to the many-step teacher, and Channel Match further improves rendering quality by recovering the latent interface between geometry and appearance generation.
[CV-329] HGA-Quant: Heterogeneity-Aware Quantization for AutoRegressive Visual Generation Models
链接: https://arxiv.org/abs/2610.15297
作者: Xinmeng Liu,Xianglong Yan,Kaicheng Yang,Weisheng Dong,Yulun Zhang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Recently, AutoRegressive Visual Generation (ARVG) models have emerged as a competitive paradigm for high-quality image synthesis. However, their substantial computational and memory demands make them difficult to deploy on resource-constrained edge devices. Post-Training Quantization (PTQ) provides a promising solution to reduce the computational and memory costs of ARVG models, yet existing methods often suffer from substantial image quality degradation under low-bit quantization. In this paper, we propose a training-free PTQ framework for ARVG, consisting of Tail-Aware Bounded-init Grid Refinement (T-BGR) and Coarse-to-Fine Activation Precision Scheduling (CAPS). T-BGR introduces a tail-aware reconstruction objective for weight quantizer calibration, assigning greater importance to distribution-tail coefficients and preserving fine-grained information without introducing additional parameters or high-precision branches. CAPS performs coarse-to-fine activation precision scheduling by jointly considering the importance and computational cost of different generation stages, enabling efficient mixed-precision allocation under a given bit budget. Extensive experiments on multiple ARVG backbones demonstrate that our framework consistently improves robustness under aggressive low-bit quantization. Notably, on Infinity, our framework achieves near full-precision image generation quality at an effective W4A5.6 activation precision, while existing methods at W4A6 still exhibit noticeable quality degradation or inconsistent performance. Code is available at this https URL.
[CV-330] Co-DSL: Human-LLM Co-design of Reusable Spatial Concepts for Layout Generation
链接: https://arxiv.org/abs/2610.15290
作者: R. Kenny Jones,Jiajun Wu,Maneesh Agrawala
类目: Graphics (cs.GR); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this http URL
Abstract:We present Co-DSL, a framework in which users and LLMs co-design reusable spatial concepts for layout generation. In the first stage, our system allows users to create a library of personalized operations. To add a concept, a user provides a description, which the LLM converts into an initial implementation. The user can then observe the concept’s execution behavior over example scenarios and provide lightweight feedback to the LLM, which revises the concept’s implementation logic. In the second stage, given a library, an LLM can author a layout program that meets a new layout request from a user. Executing a layout program instantiates a concept-graph, a declarative composition of library concepts. In this way, we require the LLM only to specify which concepts should hold, while we use a solver to search for a layout that satisfies them. Across five spatial layout domains, we find that Co-DSL libraries let LLMs capture layout patterns that are challenging for prior fixed-operation systems, and demonstrate that our concept-graph representation is needed for LLMs to reliably compose concepts together. We further find that Co-DSL lets LLMs author generative layout programs that expose high-level parametric controls users can manipulate to explore a distribution of related layouts without re-prompting.
[CV-331] ASM-6D: Fast Certified Pose-Shape Inference with Complete Correspondence Flow Maps
链接: https://arxiv.org/abs/2610.15287
作者: Ho Jin Choi,Minku Kim,Nadia Figueroa
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: 18 pages, project page: this https URL
Abstract:Category-level 6D pose and shape estimation requires recovering an object’s pose and geometry from partial observations without an exact instance model. We present ASM-6D, which couples generative complete-geometry prediction with explicit category-level pose-shape inference through shared correspondences. Given an RGB-D image, ASM-6D generates alternative complete 3D correspondence hypotheses whose point identities are shared with a category-level Active Shape Model (ASM), including occluded points. Each hypothesis is fitted directly for rotation, translation, metric scale, and instance shape without a separate correspondence-assignment stage. We analytically eliminate translation and shape coefficients, reducing fitting to a nine-variable scaled-rotation optimization problem that enables efficient batched evaluation of multiple hypotheses in real time. ASM-6D achieves state-of-the-art performance on ROPE and HouseCat6D benchmarks. Ablations demonstrate the importance of generative correspondence prediction and active shape variation. The representation also supports shape completion, shape-dependent pose ambiguity, tracking, and robotic grasping at 6-38 Hz, depending on the inference budget.
[CV-332] Protein Crystallization Segmentation and Classification Using Subordinate Color Channel in Fluorescence Microscopy Images
链接: https://arxiv.org/abs/2610.15286
作者: Truong X. Tran,Marc L. Pusey,Ramazan S. Aygun
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:The accuracy of detecting protein crystals for fluorescence microscopy images is very critical for high throughput and automated systems. Although the trace fluorescent labeling method could highlight protein crystals, reflection and emission from the fluorescence dye is not always due to crystal regions. Therefore, the analysis of the peak wavelength in the emission spectra of a fluorophore may not always yield effective results. In this paper, we show that using the subordinate color intensity corresponding to longer wavelengths than the peak wavelength of the emission spectra could improve the accuracy of protein crystal detection. Hence, we have built a segmentation method based on the percentile intensity of the subordinate color for trace fluorescently labeled (TFL’d) protein crystallization trial images. Compared to using the dominant color channel, our segmentation method on subordinate color channel was able to reduce the misclassification rate of likely-leads or crystals as non-crystals by the percentage of from 9.71% to 2.02% depending on the classifier. Similarly, the accuracy of classifiers were increased by the percentage of from 1.77% to 5.53%. Our method reached around 94% accuracy while keeping misclassification of likely-leads and crystals as non-crystals below 1%. Moreover, to evaluate the generalizability of our method, we have conducted new wet lab experiments on two proteins, Concanavalin A (Con A) and Ab inorganic pyrophosphate (AbIPPase), and the misclassification rate was below 1%. Our experiments show that using the subordinate channel may be more helpful for TFL’d protein trial image classification.
[CV-333] GeoBridge: Evolving Transferable Decoder Topologies for Geometric Vision
链接: https://arxiv.org/abs/2610.15280
作者: Rui Wang,Haoyu Zhang,Ran Cheng
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Depth estimation and Gaussian reconstruction share a need for multi-scale feature processing, yet use different prediction interfaces and training objectives. Whether a decoder’s internal computation can be optimized once and reused across these systems remains a structural design question. Search over primitive configurations alone leaves the organization of transformations fixed. To address this gap, we propose GeoBridge, an evolutionary framework for transferable decoder topologies. First, GeoBridge makes stage-internal topology searchable through residual, sequential, and parallel composition of configurable primitives. Shape-preserving composition retains the stage interfaces of a fixed multi-scale scaffold. Second, a functional-tree encoding supports stage-aware crossover and hierarchical mutation, while a gradient sensitivity-alignment proxy evaluates candidates without training. Multi-objective evolution balances this proxy with computation and parameter cost. Third, selected topologies are instantiated and trained in target systems with task-specific prediction interfaces. A single source search takes approximately 12 GPU hours. GeoBridgeNet-A3 achieves the best depth accuracy among four adapted decoder baselines under a common encoder and training protocol on NYU Depth V2 and KITTI. The same topology reaches 27.11 dB PSNR in Gaussian novel-view synthesis on RealEstate10K, and its refined representations support open-vocabulary segmentation on LERF. These results support decoder topology as a reusable design object across the evaluated geometric vision systems. Code and pretrained weights are available at this https URL.
[CV-334] ExeWorldBench: Benchmarking Code-Based Executable 3D World Modeling
链接: https://arxiv.org/abs/2610.15166
作者: Xinge Peng,Zhen Fang,Ruibing Wang,Yongxuan Lv,Haodong Li,Guohui Zhang,Xiaoxiao Ma,Yiting Lu,Lin Chen,Zehui Che,Feng Zhao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Coding agents are increasingly evolving beyond reproducing what a world looks like toward constructing executable worlds that capture how a world works. This shift calls for executable world modeling, which integrates two core capabilities: world understanding, reasoning about the underlying structural, mechanical, and causal principles of a world, and executable realization, translating that understanding into a code-based system whose behavior emerges from physical interactions. However, existing benchmarks provide only fragmented evaluations centered on physical reasoning, visual fidelity, or behavioral correctness, leaving the integration of these capabilities insufficiently assessed. To bridge this gap, we introduce ExeWorldBench, to our knowledge the first benchmark to comprehensively evaluate world understanding and executable realization through code-based 3D world construction. ExeWorldBench comprises three progressively challenging tasks: Mechanism Executability, which evaluates the realization of individual mechanisms from visual observations; Compositional Executability, which extends to multi-stage systems with spatial, mechanical, and causal dependencies; and Intent-Driven Executable World Synthesis, which requires constructing executable worlds from high-level goals and physical requirements. Extensive experiments on frontier models reveal substantial limitations in executable world modeling, offering insights into the development of more physically grounded and causally coherent world construction systems.
[CV-335] RIG: Toward Measuring Texture Realism in Image Generation
链接: https://arxiv.org/abs/2610.15164
作者: Zhiwei Xu,Jiali Cui,Zhiqiang Lao,Heather Yu,Liang Peng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 22 pages, 19 figures
Abstract:This paper studies the backbone used by metrics for evaluating the quality of generated images. Although generative models have advanced rapidly, some outputs still exhibit unrealistic texture characteristics, and these differences are difficult to measure. Widely used evaluation metrics provide limited insight into texture realism, since they often rely on a backbone trained for another objective, such as image classification or image-text alignment. Such objectives typically encourage invariance to low-level variations that carry texture, so the resulting features primarily encode semantic and global structure. As a result, texture-specific discrepancies may not be explicitly captured. We propose TRIG, a metric directed at the texture realism of generated images. Its backbone is instead the discriminator of a generative adversarial network pre-trained on a texture dataset, which provides a learned prior for the local statistics of natural textures. A plausibility head is fine-tuned to identify local texture degradations, while preserving natural defocus and sensor noise as positive cases. On the features of this backbone, we compute a standard distributional distance between a reference set of real images and a set of generated ones. We apply TRIG to several generative models and observe closer agreement with human realness ratings than baseline metrics.
[CV-336] RootSLAM: Square Root Bundle Adjustment for Visual-Inertial SLAM
链接: https://arxiv.org/abs/2610.15157
作者: Tomás Santucci,Mateo de Mayo,Daniel Cremers,Taihú Pire
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Efficient perception is a prerequisite for the small, low-power devices targeted by robotics, mixed reality, autonomous driving, and drone applications, all of which rely on Simultaneous Localization and Mapping (SLAM) to estimate their motion within their surroundings. Learning-based SLAM systems achieve state-of-the-art accuracy but are demanding in compute, whereas classical geometry-based methods retain a strong efficiency advantage with little to no loss in accuracy. In this work we present RootSLAM, a fully-featured visual-inertial SLAM system for multi-camera setups built on Basalt. RootSLAM achieves accuracy competitive with the state of the art while running up to 20 times faster, facilitating the deployment of perception systems in resource constrained environments. We adapt the square-root bundle adjustment formulation from structure from motion into the multi-camera visual-inertial SLAM setting through a novel reformulation that leverages the system’s pose-graph, letting the global optimizer run in single-precision floating point, thus doubling its performance. We provide a thorough benchmark against the state of the art together with an ablation study of our contributions. Our new implementation, including its map management and loop detection module, pose-graph optimizer, and global mapping components, will be open-sourced.
[CV-337] RIECNN: Real-Time Image Enhanced CNN for Traffic Sign Recognition
链接: https://arxiv.org/abs/2610.15127
作者: Reem Abdel-Salam,Rana Mostafa,Ahmed H. Abdel-Gawad
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Published in Neural Computing and Applications, Volume 34, pages 6085-6096 (2022)
Abstract:Traffic sign recognition plays a crucial role in the development of autonomous cars to reduce the accident rate and promote road safety. It has been a necessity to address traffic signs that are affected significantly by the environment as well as poor real-time performance for deep-learning state-of-the-art algorithms. In this paper, we introduce Real-Time Image Enhanced CNN (\our) for Traffic Sign Recognition. \our is a real-time, novel approach that tackles multiple, diverse traffic sign datasets, and out-performs the state-of-the-art architectures in terms of recognition rate and execution time. Experiments are conducted using the German Traffic Sign Benchmark (GTSRB), the Belgium Traffic Sign Classification (BTSC), and the Croatian Traffic Sign (rMASTIF) benchmark. Experimental results show that our approach has achieved the highest recognition rate for all Benchmarks, achieving a recognition accuracy of 99.75% for GTSRB, 99.25% for BTSC and 99.55% for rMASTIF. In terms of latency and meeting the real-time constraint, the pre-processing time and inference time together do not exceed 1.3 ms per image. Not only have our proposed approach achieved remarkably high accuracy with real-time performance, but it also demonstrated robustness against traffic sign recognition challenges such as brightness and contrast variations in the environment.
[CV-338] Drive3R: Block-wise Streaming Geometric Foundation Models for Autonomous Driving
链接: https://arxiv.org/abs/2610.15112
作者: Tianyi Yang,Haojun Yang,Jipeng Xiao,Xuesong Chen,Shaoshuai Shi,Li Jiang,Zhipeng Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 11 pages, 6 figures, 4 tables. Project page: this https URL
Abstract:High-fidelity closed-loop simulation for autonomous driving increasingly relies on dense multi-view 3D reconstruction. While recent feed-forward geometric foundation models have revolutionized per-scene optimization, applying them to continuous, long-horizon driving sequences is computationally intractable due to GPU memory limits and the quadratic complexity of global attention. Existing scaling strategies are fundamentally limited, as causal streaming suffers from compounding scale drift and trajectory deviation, while rigid “chunk-then-align” heuristics fragment global context and produce trajectory discontinuities. To address these limitations, we introduce Drive3R, a block-wise streaming geometric foundation model. Its block-wise causal attention resolves the scalability bottleneck of fully bidirectional attention while avoiding the loss of local future context in purely causal streaming. Its confidence-aware re-anchoring further mitigates long-range drift by detecting unreliable streaming states, re-initializing the KV cache, and stitching refreshed segments through SE(3) alignment. Extensive experiments demonstrate that Drive3R establishes a new state-of-the-art on nuScenes, KITTI, and Waymo. Specifically, on nuScenes, it improves over the recent DVGT-2 baseline from 0.1806 to 0.1081 in AbsRel, from 0.6750 to 0.5344 in point-cloud Accuracy, and from 1.5812 to 0.9092 in ATE.
[CV-339] IDE: Trajectory-Informed Skill Distillation for Efficient Text-to-Video Prompt Optimization
链接: https://arxiv.org/abs/2610.15098
作者: Chengxing Zhou,Yuliang Huang,Jingwei Song
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Text-to-video (T2V) models are highly sensitive to prompt specificity, while iterative prompt optimization typically improves generation quality only through repeated video generation and vision-language model evaluation at inference time. We propose TIDE, a framework that distills this costly search process into an interpretable and reusable prompt-rewriting Skill. TIDE formulates prompt rewriting as a set of source-conditioned rule modules, each targeting a distinct semantic dimension. Once distilled, the frozen Skill activates the relevant rules and generates both positive and negative prompts in a single LLM call, followed by only one T2V generation, without test-time visual search or backbone fine-tuning. Evolved using only 100 prompts over 14 hours, TIDE achieves prompt-optimization performance comparable to computationally expensive search-based methods. Moreover, when transferred to previously unseen video-generation models, TIDE outperforms existing approaches specifically designed or trained for those models. Code is available at this https URL.
[CV-340] VLCQ: Post-training quantization for deep neural networks using variable length coding
链接: https://arxiv.org/abs/2610.15092
作者: Reem Abdel-Salam,Ahmed H. Abdel-Gawad,Amr G. Wassal
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Performance (cs.PF)
备注: Published in Future Generation Computer Systems, Volume 166 (2025), Article 107654
Abstract:Quantization plays a crucial role in efficiently deploying deep learning models on resource-constrained devices. Post-training quantization does not require access to the original dataset or retraining the full model. Current methods that achieve high performance close to baseline results typically require INT8 fixed-point integers. However, achieving higher model compression with lower bit-widths often leads to significant performance degradation. In this paper, we propose VLCQ, a post-training quantization method that relaxes the fixed-point encoding constraint, enabling more flexible exploration of quantization techniques through variable-length encoding. This approach achieves competitive or improved performance while reducing bit-widths, without requiring access to training data or model fine-tuning. Extensive experiments on image classification, segmentation, and object detection demonstrate that VLCQ improves model compression and accuracy compared with state-of-the-art post-training quantization approaches. For per-channel quantization, VLCQ surpasses FP32 accuracy and Piece-Wise Linear Quantization (PWLQ) on most evaluated models, achieving up to 6x compression compared with FP32 and up to 1.7x compared with PWLQ. When prioritizing compression with limited accuracy loss, VLCQ achieves up to 12.25x compression compared with FP32 with less than 4% performance degradation. For per-tensor quantization, VLCQ is competitive with Data-Free Quantization (DFQ), while providing greater flexibility to achieve lower bit rates across different tasks and models.
[CV-341] Region Aware Squeeze-and-Excitation Networks for Image Segmentation
链接: https://arxiv.org/abs/2610.15079
作者: Jack Roberts,Praneeth Tummala,Kavi Sarna,Jeova Farias Sales Rocha Neto
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Squeeze-and-Excitation (SE) modules improve convolutional networks by modeling channel-wise dependencies, but global average pooling collapses spatial structure into a single per-channel descriptor, limiting their effectiveness for dense prediction tasks such as image segmentation. We generalize SE by replacing global pooling with spatially guided pooling, enabling region-aware channel recalibration through four modules with complementary cost profiles. BASE formalizes Tiled Squeeze-and-Excite using a regular grid; SPASE pools over SLIC superpixels for image-adaptive regions; MAPSE obtains regions from semantic maps produced by a frozen, general-purpose network; and PLEASE learns soft region assignments end-to-end through a lightweight assignment network. Each channel receives one gate per region, so attention at each pixel reflects its region membership, bridging channel and spatial attention. PLEASE also produces its soft assignments as a byproduct of the forward pass, providing structured spatial interpretability without post-hoc analysis. The modules offer complementary computational tradeoffs: some add essentially no parameters, while others add no measurable latency. Across binary segmentation on CUB, DUTS, and ECSSD, semantic segmentation on PASCAL VOC 2012 and ADE20K, and classification on mini-ImageNet, our modules outperform SE and related spatial-channel attention variants on most benchmarks, achieving the majority of best results.
[CV-342] BrainAid: Efficient Brain Tumor Detection Using Lightweight CNNs and Explainable AI for Resource-Constrained Environments
链接: https://arxiv.org/abs/2610.15077
作者: Sai Sriram Gonthina,Ross Gore,Bikash Chandra Singh,Pronaya Bhattacharya,Sachin Shetty,Abhishek Phadke,Sandip Roy
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:A brain tumor is an unusual growth of cells in the brain that can be cancerous or not, affecting normal brain function and causing serious risk to health. Existing solutions for brain tumor classification, such as expert manual evaluations of magnetic resonance imaging (MRI) scans, are generally inconsistent as different professionals have different criteria and can be resource-intensive, leading to delayed diagnoses. Additionally, while deep learning models have demonstrated potential, they typically require significant computational resources, motivating the need for more efficient and accessible diagnostic tools that can be deployed in real-time healthcare settings. To address these limitations, this study presents BrainAid, which makes efficient use of Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) for effective classification of four-class classification of glioma, meningioma, pituitary tumor, and no-tumor MRI scans, outperforming all the standard models such as VGG16, Inception-V3, ResNet-50 and so, while integrating Explainable AI (XAI) to improve interpretability. The BrainAid is trained to classify tumors with a high precision of 98%, achieving an F1 score of 98% and an accuracy of 98% using publicly available dataset. The model is deployed on a Raspberry Pi, demonstrating efficiency in low-resource environments. This system shows the feasibility of real-time, edge-based tumor classification, providing quick diagnostic support for medical professionals and enhancing patient care in resource-limited healthcare facilities and underserved areas.
[CV-343] Global Disaster Risk Assessment Using World Risk Index Data: A Geospatial and Visual Analytics Approach
链接: https://arxiv.org/abs/2610.15064
作者: Malithi Nawarathne,Nadeesha Herath Mudiyanselage,Lihini Sangeetha
类目: Computer Vision and Pattern Recognition (cs.CV); Computers and Society (cs.CY)
备注: 16 pages
Abstract:The world has seen several threats that include resource depletion, food security, poverty and inequality, difficulties associated with urbanization, migration, and environmental effects. The World Risk Index (WRI) is a metric that is used to measure the susceptibility of nations against natural disasters and related risks that is reported by the Humanitarian Data Exchange. The WRI is comprised of factors such as susceptibility, vulnerability to natural hazards, and coping mechanisms. The WRI index has allowed to recognize regions in advance in order to face threats and situations beforehand. This study was conducted by analyzing the data that was related to years from 2000 to 2025 using data visualization techniques. Python related libraries of Matplotlib, Seaborn, and Geopandas were used due to their efficiency and usability in analyzing WRI data. As per the analysis, it was identified that countries of Philippines, Mexico, Indonesia, and India were recorded with high WRI index values that showed these nations should be prepared for natural disasters and situations. Hence, this study has provided insights using data visualization techniques in understanding natural disasters and threats so that necessary measures can be taken
[CV-344] Deep Learning-based Semantic Segmentation of Dental Radiographs: Comparative Evaluation of PSPNet DeepLabv3 UNet and UNet
链接: https://arxiv.org/abs/2610.15061
作者: Malithi Nawarathne,Nadeesha Herath Mudiyanselage Lihini Sangeetha
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 11 pages
Abstract:Accurate and efficient analysis of dental radiographs has become increasingly important for disease diagnosis and clinical decision making. In particular, dental radiographs segmentation plays a critical role as it is challenging due to several reasons, such as low contrast, noise in the images, variations in the image quality, etc. Previous studies have primarily addressed these challenges using convolutional neural networks (CNN)-based models. However, direct comparison of these segmentation models is still challenging due to the variations among the experimental settings. Therefore, we mainly compared four segmentation models, including Pyramid Scene Parsing Network (PSPNet), DeepLabV3, UNet, and UNet++, through a fair, controlled experimentation setting under varying conditions such as image sizes and batch sizes. For this, we used 1269 dental radiography images, and prior to training, we implemented a series of preprocessing techniques. To assess the models, we used several metrics, including Dice coefficient (DICE), Intersection over Union (IoU), accuracy, precision, training and validation losses. According to our results, UNet++ demonstrated the highest IoU score of 0.6719, and a DICE value of 0.7869, showing superior results in terms of segmentation, specifically when configured with an image size of 300x300 and a batch size of 16. In addition, we also explored the effect of image resolutions and batch sizes on model performance and accuracy under different architectures. Therefore, this study provides insights into the strengths of different segmentation models and the potential of deep learning-based semantic segmentation methods for improving automated dental radiograph analysis.
[CV-345] Use Of Convolutional Neural Networks In Detection Of Branch Dieback Disease In Pequi Trees
链接: https://arxiv.org/abs/2610.15049
作者: Ronilson Rodrigues Pinho,Elua Ramos Coutinho,Nilza de Lima Pereira Sales,Sildenir Aalves Ribeiro,Sergio Manuel Serra da Cruz
类目: Computer Vision and Pattern Recognition (cs.CV); Computational Engineering, Finance, and Science (cs.CE); Databases (cs.DB)
备注: 17 pages, 8 figures and 7 tables
Abstract:This study aims to evaluate and compare convolutional neural network (CNN) architectures, specifically AlexNet, Inception, LeNet, VGG16, and MobileNet, for diagnosing branch dieback (seca-dos-ponteiros) in pequi trees (Caryocar brasiliense). This disease represents a severe threat to tree development and fruit yield, causing substantial economic and ecological losses. The methodology implements computer vision techniques using Python libraries, including Scikit-learn, NumPy, TensorFlow, and Matplotlib, to process and evaluate visual data. The CNN models analyze images of leaves, branches, and flower spikes to detect pathological patterns, such as alterations in texture and coloration. Ultimately, this comparative analysis identifies the most effective architecture for accurately distinguishing healthy plants from diseased ones.
[CV-346] What Really Works for Universal Multimodal Embeddings? A Controlled Study of Contrastive Training Strategies
链接: https://arxiv.org/abs/2610.15035
作者: Duo Li,Haojun Pan,Zuhao Yang,Xintong Wang,Chengfu Huo,Shijian Lu,Liang Ding
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 16 pages, 3 figures
Abstract:Contrastive learning has become the standard paradigm for multimodal representation learning, aligning semantically matched pairs while separating mismatched ones in a shared embedding space. Although various contrastive training strategies have been proposed across negative selection, batch construction, and gradient modulation, their actual effectiveness under controlled, reproducible conditions remains poorly understood. This motivates a fundamental question: Do sophisticated contrastive strategies consistently bring meaningful downstream gains in multimodal embedding learning? To address this, we systematically evaluate representative strategies under a unified benchmarking framework on MMEB. Surprisingly, direct evaluations reveal that most existing strategies offer marginal gains over a standard aligned baseline, with several even degrading overall performance. To uncover the underlying failure modes, we introduce an MLLM-as-a-Judge diagnostic framework that dissects top-1 retrieval behaviors into labeled positives, false negatives (unlabeled valid matches), and hard/easy negative errors. Our systematic analysis yields three key insights: (1) Gradient distortion: Upweighting gradients from high-similarity negatives consistently hurts downstream generalization; (2) Distribution illusion: Wider separation between positive and negative similarity distributions does not necessarily translate to better retrieval performance; and (3) Synergistic training: While false-negative mitigation or hard-batch construction yields limited gains in isolation, integrating hard-negative mining, false-negative mitigation, and hard-batch construction achieves optimal performance, delivering an average improvement of 6.0% on MMEB with Qwen2-VL-2B. Overall, our findings demystify popular contrastive strategies and provide actionable recipes for training universal multimodal embedding models.
[CV-347] Appearance Compensation in Volume Rendering
链接: https://arxiv.org/abs/2610.15030
作者: Jianghong Zhou
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 139 pages, 18 figures
Abstract:Can a volume that reproduces every external RGB view still have an incorrect density? We study this question in emission-absorption rendering with an independently parameterized, physically bounded effective source. Appearance compensates a density change by replacing the radiance that it removes from the rest of each ray. We characterize the surviving density information with a shared spatial appearance field. Our first result is an optimal density-only gain after finite-degree spherical-harmonic source refitting. A coupled tensor gauge and physical localized packets identify the sharp limiting nonlinear inverse coefficient as its reciprocal. Our second result realizes positive gain at every fixed finite degree for a residual set of isotropic spatial sources at one fixed density. At the same references, growing angular capacity hides a fixed nonzero density error while retaining physicality and any prescribed finite set of smoothness bounds. Local identifiability and the absence of a capacity-uniform recovery modulus therefore coexist. Our third result constructs an exact physical SH2 ambiguity: central density doubles and relative density L^2 error exceeds 14.52% , yet every external RGB ray is unchanged. We connect these continuum results to shared finite representations, continuous pixel measurements and density-sensitive acquisition. Complete-class certificates separate SH0, SH1 and SH2 fits at one learned density. Finite-camera results expose the distinct role of spatial capacity, and transmission measurements separate otherwise identical RGB fields. Controlled training tests and captured-smoke profiles supply mechanism evidence with explicit observation scopes. The results show why rendering accuracy alone cannot certify density: recovery also depends on appearance capacity, spatial resolution and measurement design.
[CV-348] CowCare: A Unified Visual Framework for Viewpoint-Robust Cattle Tracking and Spatiotemporal Heat-Stress Analysis
链接: https://arxiv.org/abs/2610.15012
作者: Xin Liu,Weiran Li,Minghao Zhu,Mina Han,Zhenbo Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Reliable individual monitoring is essential for dairy cattle welfare and early heat stress detection in Precision Livestock Farming (PLF). However, translating raw multi-camera videos into automated diagnostics remains challenging due to disconnected tracking and recognition pipelines and complex physical environments. Dense mutual occlusions and cross-camera viewpoint variations fragment identities, while environmental noise, including camera vibrations, and limited agricultural data obscure subtle heat stress cues such as rapid breathing and flank undulations. To bridge low-level visual features and high-level behavioral diagnostics, we propose CowCare, a unified framework combining viewpoint-robust tracking with spatiotemporal heat stress recognition. Its tracking module, CowTrack, maintains cross-camera identity consistency through a View-Decoupled Aggregation Network (VDA-Net). Instead of relying on static global representations, VDA-Net aggregates localized textures under heavy occlusion and uses a Viewpoint Decoupling Transformer (VDT) to suppress transient viewpoint variations. CowTrack further incorporates a causal online self-supervision mechanism for continuous adaptation to dynamic farm scenes. The resulting tracklets are subsequently processed by TSDF-Net, a fine-grained video understanding network. Its Temporal Stabilization and Aggregation Module (TSAM) suppresses high-frequency jitter, while a Diffusion Semantic Prior (DSP) uses frozen pre-trained diffusion semantics to enhance subtle behavioral signals and mitigate overfitting. Extensive evaluations on two real-world farm benchmarks demonstrate the framework’s effectiveness: CowTrack achieves 83.48% HOTA and 87.72% IDF1 for global multi-object tracking on MVCow, and TSDF-Net achieves 87.72% Top-1 accuracy on HSCow. These results establish CowCare as a robust end-to-end framework for automated livestock health monitoring.
[CV-349] Retraining-Free Depth Pruning of Vision Transformers: A Cross Domain Study of Layer Redundancy
链接: https://arxiv.org/abs/2610.15005
作者: Alessandro Viespoli,Loris Nanni
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Transformer architectures have achieved strong results in natural language processing and computer vision, but their deployment is limited by high computational, memory and storage costs, motivating research into model compression techniques, such as pruning, while preserving performance. Prior work has shown substantial redundancy in large language models (LLMs), and redundancy in vision transformers has also been investigated, but it remains less systematically analyzed than in LLMs. In this work, we extend similarity-based redundancy analysis to vision models by measuring cosine similarity between layer inputs and outputs to identify transformations with minimal impact. We study redundancy patterns across several transformer architectures and find that deep vision transformers such as DINOv2 and SwinV2, like language models, exhibit significant redundancy. We then evaluate how this redundancy can be exploited to improve efficiency, analyzing the trade-off between inference speed and performance degradation under different pruning strategies. Our results show that moderate pruning can yield substantial acceleration with limited impact on accuracy, and that 20% of the DINOv2 blocks can be removed without any retraining, for a 1.24x throughput gain and a 20% lower energy per image at a relative accuracy loss below 1% on ImageNet-1K. We further propose a model selection pipeline that automatically identifies the best pruned configuration for a given dataset, reducing FLOPs by 13% to 27% on average across architectures with an average accuracy change between -0.3 and +1.0 points.
[CV-350] Metric-Driven Reinforcement Learning for Object Hallucination Reduction in LVLM-Based Image Captioning
链接: https://arxiv.org/abs/2610.14976
作者: Thanh Nguyen,Hee Suk Yoon,Tung Minh Luu,Eunseop Yoon,Chang D. Yoo
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 20 pages, IEEE ACCESS 2026
Abstract:Large Vision-Language Models (LVLMs) have excelled in joint visual and language understanding, particularly in image captioning. However, LVLMs remain prone to object hallucination, describing non-existent entities in the given image. Previous remedies, including dataset-enhanced supervised training and reinforcement learning from human feedback, show promise but rely heavily on costly human involvement, limiting scalability. To address this, we propose a reinforcement learning framework to mitigate object hallucination in LVLM-based image captioning, driven entirely by natural language processing (NLP) metrics. While optimizing reinforcement learning solely on NLP metrics offers scalability, it also introduces pitfalls: reward hacking, sparse feedback, and costly training. Our framework sidesteps these by introducing an Object Consistency Score (OCS) capturing both correctness and coverage of objects, incorporating KL-regularization for denser token-level guidance, and designing a lightweight PPO variant that fine-tunes adapters atop a frozen LVLM backbone, achieving efficiency and stability. Extensive experiments on the InstructBLIP baseline demonstrate the effectiveness and generalizability of our framework, achieving a substantial hallucination reduction of up to 41% on the COCO dataset and 29% on the Visual Genome. Furthermore, the framework can enhance caption quality by incorporating complementary evaluation metrics such as BERTScore and METEOR. These results reveal a broader insight in multimodal learning: LVLMs can effectively self-correct hallucinations using metric feedback, motivating deeper exploration into scalable reinforcement frameworks that significantly reduce human effort.
[CV-351] AD-GS: Anatomical Density-guided 3D Gaussian Splatting for Sparse-View Dental CBCT Reconstruction
链接: https://arxiv.org/abs/2610.14963
作者: Zonglin Li,Kexian Xie,Feiran Li,Chu Zhou,Yixing Huang,Ruiyang Chen,Zhanyu Ma,Heng Guo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Sparse-view Cone-Beam Computed Tomography (CBCT) reconstruction is essential for reducing radiation dose and scan effort, but remains underconstrained due to limited projection information. Recent 3D Gaussian Splatting (3DGS)-based CBCT reconstruction methods rely on gradient-driven densification, which becomes unstable under sparse acquisition and often leads to over-smoothed structures. We propose AD-GS, an Anatomical Density-guided 3D Gaussian Splatting framework that formulates Gaussian densification as an attenuation-aware Markov Chain Monte Carlo (MCMC) sampling process. By interpreting Gaussians as samples from an underlying anatomical attenuation distribution, AD-GS adaptively concentrates representation capacity in high-attenuation regions, enabling sharper structural boundaries under sparse projections. To further constrain reconstruction, we incorporate a semantic regularization based on zero-shot features from Segment Anything in Medical Images, enforcing multi-view consistency through semantic priors. Experiments on the ToothFairy3 dental CBCT dataset demonstrate that AD-GS achieves superior structural fidelity and boundary preservation compared with existing 3DGS-based approaches under sparse-view settings within the dental and maxillofacial setting. Code is available at this https URL.
[CV-352] CatPortrait: Taming Video Generative Models for Portrait Animation via In-context Conditioning
链接: https://arxiv.org/abs/2610.14957
作者: Ziqi Ni,Zhou Fang,Shijian Wang,Rui Li,Yi Zhou
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to PRCV 2026
Abstract:Recent video generative models have shown strong capability in synthesizing high-quality and temporally coherent videos. Yet, adapting them for controllable portrait animation remains nontrivial, as existing methods typically rely on architectural changes or extra conditioning modules, increasing complexity and reducing compatibility with pretrained models. We present CatPortrait, a simple and effective framework that enables controllable portrait animation via 3D face guidance as in-context conditioning. CatPortrait preserves the original video generation backbone and injects structured 3D facial guidance into the input space through spatial concatenation, allowing pretrained image-to-video models to animate portraits without sacrificing their pretrained priors. CatPortrait supports both video-driven and audio-driven portrait animation within a unified framework. Extensive experiments demonstrate that CatPortrait achieves competitive or superior results across both settings.
[CV-353] DSM-Free UAV Visual Localization via Latent-Depth Variable Projection and Structure-Adaptive Coupling
链接: https://arxiv.org/abs/2610.14945
作者: Yibin Ye,Xichao Teng,Shuo Chen,Xiaokai Song,Yuli Sun,Qifeng Yu,Zhang Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 15 pages, 8 figures
Abstract:Visual localization enables unmanned aerial vehicles (UAVs) to maintain absolute positioning in GNSS-denied environments. A common approach matches UAV images with geo-tagged orthographic maps and estimates camera pose using Perspective-n-Point (PnP) with elevations from Digital Surface Models (DSMs). However, accurate and up-to-date DSMs are often unavailable in practice. Orthographic-perspective geometry (OPG) enables DSM-free localization in scenes with varying elevations, yet existing methods typically estimate pose indirectly through intermediate matrix estimation and decomposition. To address this limitation, we propose Latent-Depth Variable Projection (LDVP), which treats the unknown scales along perspective viewing rays as latent depths. By analytically eliminating these depths and solving for the horizontal camera center in closed form given the rotation, LDVP reduces nonlinear estimation to a rotation-only problem, enabling direct recovery of metric horizontal position without assuming a planar scene. To improve localization robustness across different scene structures, we further introduce Structure-Adaptive Coupling (SAC), which uses relative scene structure inferred from monocular depth estimation to select LDVP when the matched points are not dominated by a single plane, and Planar-PnP otherwise. Our local observability analysis reveals how scene elevation variation affects the geometric constraints of OPG, providing a basis for this adaptive strategy. Extensive experiments show that LDVP consistently outperforms existing OPG-based DSM-free methods, while SAC further improves robustness across diverse scenes and substantially narrows the gap to DSM-based localization. Code will be released at this https URL.
[CV-354] xt-Guided Medical Image Segmentation via Residual Aggregation and Dense Alignment Representations NEURIPS2026
链接: https://arxiv.org/abs/2610.14944
作者: Jie Gui,Hang Tu,Yunfeng Dou,Wen Sha,Xiuquan Du,Shuo Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at NeurIPS 2026 (poster)
Abstract:Specifying segmentation targets with natural language provides flexible semantic conditioning for medical image segmentation. However, existing methods still struggle to reliably ground high-level textual semantics into accurate pixel-level spatial prompts, especially under low-contrast lesions, ambiguous boundaries, and visually similar anatomical contexts. In this paper, we propose RADAR (Residual Aggregation and Dense Alignment Representations), a parameter-efficient framework for text-guided medical image segmentation. RADAR explicitly decouples and synergizes the medical semantic anchoring capability of BiomedCLIP with the prompt-driven mask decoding ability of MedSAM2, while enabling low-cost task adaptation via LoRA. Specifically, Guided Residual Depth Aggregation adaptively aggregates cross-layer features under semantic guidance to alleviate delayed textual injection and deep semantic dilution. Structured Dense Representation Alignment leverages a frozen BiomedCLIP to construct crop-level target, boundary, and context anchors, improving the local discriminability of dense features and the reliability of automatic prompts. Multi-modal Prompt Generation further converts cross-modal semantic responses into composite text, point, and mask prompts for MedSAM2 decoding. Experiments on three public medical image segmentation datasets show that RADAR achieves competitive performance with only 0.74M trainable parameters, demonstrating its effectiveness in parameter-efficient adaptation, semantic grounding, and dense representation enhancement. The source code is available at this https URL.
[CV-355] Engram Neuron Inspired Adaptive Memorization for Lifelong Learning
链接: https://arxiv.org/abs/2610.14940
作者: Suiyang Jiang,Bomin Wang,Hangqi Zhou,Jianfeng Feng,Xiahai Zhuang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Lifelong learning enables artificial systems to adapt to changing environments. To achieve this, an effective artificial learner should dynamically reorganize past and new experiences, maintaining stability while enabling flexible knowledge integration. However, existing efforts primarily mitigate catastrophic forgetting, lacking the ability to adaptively update memory akin to biological systems. Here, by integrating the full memory life cycle of engram neurons into artificial learners, we propose a framework that constructs a structured memory from gradient representations, which regulates parameter updates and guides memory management. Through comprehensive evaluations on four benchmarks, our approach shows promising lifelong learning capability, maintaining stability over long task sequences and supporting flexible adaptation. Beyond the performance gains, our approach offers a computational perspective on robust and flexible lifelong learning, echoing mechanisms observed in biological memory systems.
[CV-356] Scalable Multi-Human Cooperative Manipulation via Commander-Executor Coordination
链接: https://arxiv.org/abs/2610.14911
作者: Buzhen Huang,Xingchen Wu,Yuanwang Yang,Chenyi Xie,Kun Li
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multi-human cooperative object manipulation is a highly challenging topic in humanoid control. While existing methods can handle simple collaborative carrying tasks, they struggle with role-dependent, long-horizon, and sequential cooperative manipulation. To address this issue, we propose a hierarchical commander-executor framework that enables multiple agents to accomplish complex collaborative tasks in distinct ways. Specifically, a commander combines instructions and scene information for joint semantic and geometric planning. It then decomposes the task into role-specific subtasks and generates shared object-centric coordination plans, specifying execution timing, target positions, and manipulation actions for each agent. Conditioned on the commander’s high-level strategy and each agent’s local observations, the executor learns a low-level physics-based policy through a solo-to-cooperative curriculum to realize coordinated human-object interactions. The planned object trajectory provides a shared coordination target, while root positions and grasping instructions specify role-dependent execution cues. We evaluate our method on a variety of object manipulation tasks, demonstrating its ability to execute long-horizon tasks that combine different types of collaboration and to scale from two-agent to multi-agent cooperation. Extensive experiments further show that our approach achieves higher success rates compared to state-of-the-art methods. Code will be made publicly available at this https URL.
[CV-357] MORE: Learning Motion-Oriented Registers for Driving World Models and Planning
链接: https://arxiv.org/abs/2610.14871
作者: Yingdong Gu,Haoyu Wang,Shaocheng Yan,Zhenjun Zhao,Hao Han,Pengcheng Shi,Xianfei Li,Qinghua Tan,Wenlong Liao,Tao He,Jiayuan Li,Pai Peng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:By predicting future states, latent world models provide rich self-supervision for end-to-end driving, but often inherit dense features from pretrained visual encoders as their world-state interface. This creates a mismatch: visual pretraining produces information-rich, spatially dense features, but their adaptation for motion-conditioned world modeling and planning is deferred to downstream training. The dense grid also makes future prediction more computationally expensive as visual resolution increases. We therefore argue that video pretraining should directly learn state representations suited to downstream world modeling and planning, rather than relying on downstream training to adapt inherited visual features. We propose MORE, a V-JEPA framework that learns Motion-Oriented REgisters during video pretraining as a shared state interface for driving world models and planning. MORE injects temporally aligned ego motion into a small set of registers and uses them as the sole context for masked patch prediction, channeling dense patch-level supervision through a compact state bottleneck. A complementary blockwise register-masking objective further shapes predictive structure in the register space. During downstream training, the same register space defines the future states predicted by the world model and provides shared scene memory for trajectory generation and scoring. On NAVSIM-v2, MORE outperforms the state of the art by 1.6 in EPDMS, while achieving superior efficiency with only 1/10 the inference latency of DriveLaW. These results support learning the state interface for world modeling and planning directly during video pretraining.
[CV-358] ScalLoc: Continuous Oblique Thermal UAV Geo-Localization with a 2D Satellite Map
链接: https://arxiv.org/abs/2610.14866
作者: Xiaoran Zhang,Yu Liu,Jinyu Liang,Huaxin Xiao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 14 pages, 8 figures
Abstract:Cross-modal Thermal Geo-localization (TG) provides an illumination-robust positioning solution for Unmanned Aerial Vehicles (UAVs) in Global Navigation Satellite System (GNSS)-denied environments. However, existing thermal geo-localization methods mainly focus on near-nadir observations or depend on map-side elevation priors, limiting their applicability to continuous oblique thermal localization with only geo-referenced 2D satellite maps. To address this problem, we propose ScalLoc, a scale-aligned local-to-map localization framework for continuous oblique thermal UAV geo-localization. ScalLoc follows a reconstruction–refinement–anchoring–propagation design: it reconstructs local window geometry from thermal sequences, refines camera poses using track-based pose optimization, adaptively anchors each window to the satellite map, and propagates validated states to recover frame-level global trajectories. To support systematic evaluation, we construct Thermal-Oblique, a real thermal UAV sequence dataset containing 216 urban and rural day–night trajectories and 28,112 thermal frames with frame-level annotations. Extensive experiments demonstrate that ScalLoc reduces the mean localization error from 112.69 m to 26.97 m compared with the VGGT-only configuration, corresponding to a 76.07% reduction, and achieves the lowest frame-weighted mean position error among the compared methods. These results demonstrate the feasibility of continuous oblique thermal geo-localization using only lightweight 2D satellite maps, without requiring digital surface models or preconstructed 3D maps. Code and dataset are available at this https URL.
[CV-359] CAST: Real-Time Motion Capture for Any Skeleton Topology
链接: https://arxiv.org/abs/2610.14856
作者: Pu Pang,Zhongfei Qing,Ling-Hao Chen,Songnan Li,Hua Huang,Bo Hu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 39 pages. Code repository: this https URL
Abstract:This paper studies the problem of recovering motion from monocular video, even when the target skeleton topology differs from the subject observed in the video. Although previous methods have made efforts to resolve this problem, they remain vulnerable to joint-rotation errors that are coupled across the skeleton and can propagate along its kinematic hierarchy. To mitigate these errors, we present CAST, a topology-aware motion capture framework that combines learned local rotations with geometric correction guided by predicted global bone directions. Our observability analysis characterizes which rotational components can be constrained by outgoing bone directions. Guided by this analysis, GRACE uses reliability-gated alignment to correct composed orientations in parent-to-child order, leaving unconstrained components to the learned rotation predictor. The motion predictor and correction module are trained jointly with motion-derived supervision, using a frozen image encoder. Experiments under the TopoCap evaluation protocol show that the proposed framework improves motion recovery across different target topologies, including unseen structures. Reference-based cross-skeleton evaluation on a curated Mixamo benchmark and a user preference study further assess transfer to different target structures. A causal CAST-B variant achieves over 51 FPS on a single NVIDIA A100 GPU, with timing covering image encoding and motion prediction. Code repository: this https URL.
[CV-360] Getting Space Right: Agent ic Rewards for Image Generation
链接: https://arxiv.org/abs/2610.14852
作者: Xiaofeng Tan,Jiaxin Lin,Haobo Yuan,Tiancheng Shen,Zihan Su,Rui Chen,Hongsong Wang,Ming-Hsuan Yang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 15 pages, 7 figures
Abstract:Reinforcement learning (RL) is promising for spatial text-to-image generation, but existing work largely focuses on policy optimization, assuming accurate rewards and effective online rollouts. We observe in-domain gains with out-of-domain degradation, exposing two limitations: unreliable spatial rewards in open-world settings and rollouts lacking basic-to-complex progression. We propose a curriculum-guided RL framework. First, we introduce SpatialCritic, an agentic reward model equipped with a spatial perception harness that combines reusable verification guidance with specialized vision tools. We then construct spatial preference datasets through targeted counterfactual perturbations and perform agent-harness alignment to improve tool selection and evidence integration efficiently. Second, curriculum learning organizes shared foundational capabilities and their compositional dependencies in a capability graph for mastery-guided rollouts, periodic review, and targeted remediation. On SpaRW-Eval, SpatialCritic achieves 95.1% verification accuracy, outperforming 320B GLM-5.3-Flash with a 40x smaller 8B backbone and raising average baseline scores from 78.4 to 83.9 on SD3.5-Medium and 83.2 to 86.5 on BAGEL. Project page: this https URL
[CV-361] owards Privacy-Preserving Thermal Human Perception from Dataset to Deployment
链接: https://arxiv.org/abs/2610.14846
作者: Kailai Sun,Yue Cui,Xinwei Wang,Zijie Gao,Baoshen Guo,Zhiqing Hong,Irfan Qaisar,Xu Yang,Yueting Li,Dian Fan,Qianchuan Zhao
类目: Computer Vision and Pattern Recognition (cs.CV); Signal Processing (eess.SP)
备注:
Abstract:Indoor human perception is the foundation for smart homes and healthcare. However, RGB cameras raise privacy concerns in private indoor environments, while low-resolution thermal cameras provide limited spatial and semantic information. Large-scale indoor thermal datasets and real-world deployments remain limited. Here, we introduce Indoor RGB-T, a large-scale paired RGB–thermal dataset for indoor human perception. The dataset covers 20 indoor rooms and contains 90,576 80x62 thermal frames, 87,016 temporally matched RGB–thermal pairs, and 115,557 annotated human bounding boxes. We establish benchmarks for robust thermal occupant-state detection with 0.75 mAP and occupancy estimation with an accuracy of 0.847 in unseen rooms. We further evaluate RGB-to-thermal and thermal-to-RGB translation, achieving a temperature MAE of 0.654 C and an R2 of 0.923. Finally, we deploy the thermal perception model on a commercial sleep-monitoring system in the real world. Field testing with 100 participants and deployment of approximately 1,000 devices further demonstrate its practical feasibility. We identify practical challenges that motivate future research directions, including environmental thermal interference, sensing range and multi-person occlusion. The code, trained models and partial dataset are available at this https URL and this https URL.
[CV-362] [Re] Boosting the Visual Interpretability of CLIP via Adversarial Fine-Tuning
链接: https://arxiv.org/abs/2610.14832
作者: Anton Nuzhdin,Andrei Gruzitski,Balázs Egyed,Aleksandr Raudvee
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:This paper presents a reproducibility study of ``Boosting the Visual Interpretability of CLIP via Adversarial Fine-Tuning’', published at ICLR 2025, which proposes an unsupervised adversarial fine-tuning (AFT) method with norm regularisation to enhance the visual interpretability of CLIP’s image encoder. We attempt to reproduce the key claims regarding improved saliency map quality, increased concept alignment, transferability to out-of-distribution datasets, and the trade-off with zero-shot accuracy. Beyond reproduction, we propose a saliency-guided regularisation extension that introduces an Energy Pointing Game loss, directly supervising the spatial alignment of Simple Gradient saliency maps with target objects. We evaluate our extension across a range of saliency-loss weights and show that explicit saliency supervision \revsubstantially improves localisation metrics at a modest cost: the reduction in zero-shot accuracy and adversarial robustness is small at the lowest saliency-loss weight ( w_s=0.2 ) and moderate at the strongest setting ( w_s=0.6 ). We restructure the codebase for configurable, systematic experimentation.
[CV-363] YOUDrive: Driving Style as a Composable Direction for End-to-End Autonomous Driving
链接: https://arxiv.org/abs/2610.14802
作者: Guantian Zheng,Lidong Yu,Tianyu Gao,Jiaxing Chen,Jiashu Li,Chen Lv
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: 12 pages, including appendix
Abstract:End-to-end driving policies can achieve strong task performance while offering limited control over their driving style. Driving style describes how a vehicle accelerates, brakes, and maintains distance. Policies trained without explicit preference conditioning lack a direct mechanism for adjusting these behaviors at inference. We propose YOUDrive, a vision–language–action planner for continuous and composable driving-style control. Each teacher’s behavior is distilled into a task vector, a learned parameter update that induces a style direction in the planner’s representation. These directions can be scaled and combined at inference without retraining the backbone. Conditioned on these directions, a flow-matching decoder models a multimodal trajectory distribution and generates kinematically feasible motion alternatives. We introduce the Style Alignment Score (SAS) to describe human-referenced kinematic behavior after safety and task gating. On NAVSIM navtest, a single coefficient monotonically controls assertiveness across six teacher-derived personas, while PDMS remains above 0.90 across the tested individual and blended configurations, reaching 0.954. Further analyses reveal systematic policy-specific kinematics after controlling for scene context, whereas the evaluated discrete style labels provide limited additional explanatory power. Code, checkpoints, and the metric will be released.
[CV-364] BridgeWAM: Bridging Video and Action Experts via Latent Bridge Queries
链接: https://arxiv.org/abs/2610.14746
作者: Jiaxin Fang,Ruitong Liu,Zeyu Lai,Xingyu Wang,Mingyuan Jia,Pingyue Sheng,Weikang Wan,Mengdi Xu,Yuxiang Gao
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:World Action Models (WAMs) commonly couple video and action experts through dense, layer-wise Mixture-of-Transformers (MoT) attention, exposing action prediction to irrelevant visual features while enforcing matched expert depths. This coupling hinders selective visual-feature use and introduces redundant computation. We introduce Bridge-of-Experts (BoE), instantiated as BridgeWAM, which aggregates visual, language, and proprioceptive priors through 32 Latent Bridge Queries (LBQs) into a shared, limited latent space. Joint video-action supervision lets action gradients guide the Video Expert toward representations better suited for action prediction, while the Action Expert learns to selectively retrieve task-relevant information through cross-attention. Decoupling expert depths and reusing LBQs during action denoising yield a 5.06B model with a 30-layer Video Expert and a two-layer Action Expert, achieving 98.25% success on LIBERO and 72.24% pooled success on LIBERO-Plus, demonstrating effective task execution and strong generalization with an ultralight action head. Ablations support joint representation adaptation and unified LBQ conditioning. Relative to Fast-WAM, BridgeWAM reduces single-sample inference FLOPs by 28.72% in simulation profiling and real-robot latency to 77.1 ms (a 2.47-fold speedup). Qualitative experimental results further indicate preferential attention to target objects and task destinations across perturbations.
[CV-365] Japanese Handwriting Generation Based on Human Preference Feedback Using Direct Preference Optimization
链接: https://arxiv.org/abs/2610.14743
作者: Wakana Kuwata,Ryota Mibayashi,Masanori Tani,Hiroaki Ohshima
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted for publication at BESC 2026
Abstract:We propose a Japanese handwriting generation framework that integrates human preference feedback using Direct Preference Optimization (DPO) to align generative models with human cognitive and stylistic preferences. While font generation technologies leveraging Generative Adversarial Networks and diffusion models have advanced significantly, applying these approaches to handwriting generation presents unique challenges. Font generation is a deterministic process yielding consistent styles. Handwriting generation, on the other hand, is inherently non-deterministic. In handwriting, glyph shapes never match exactly even when the same writer writes the same character. We define this shape difference as handwriting variability. Although this variability is essential for conveying the natural quality of human writing, it complicates model training and substantially increases data collection costs to capture diverse distribution patterns. To address these limitations, we incorporate human perceptual feedback into the training loop of handwriting generation. While collecting extensive handwritten data is costly, evaluating pairwise preferences between two generated samples remains very easy for humans. Applying DPO, an efficient RL-free preference optimization method, to our generation model enables high-quality handwriting generation without relying on large-scale dataset collection. Experimental results across four distinct Japanese handwriting styles demonstrate that DPO-based training enhances overall structural accuracy. However, in highly complex styles with significant variability, such as brush pen writing, the model fails to capture style-specific features. These findings highlight a critical challenge in extending DPO to complex handwriting styles.
[CV-366] Align-VAE: Aligning Attention and Geometry in VecSet VAEs for 3D Reconstruction
链接: https://arxiv.org/abs/2610.14734
作者: Huaizhi Qu,Mingxin Yang,Xintong Han,Fangneng Zhan,Chunchao Guo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Set-based variational autoencoders (VecSet VAEs) represent 3D shapes as compact sets of latent tokens. Although each token originates from a query at a sampled surface location, whether these latents truly align with geometry remains unclear. We investigate this alignment and identify four issues: (i) attention-sink-like tokens that attract queries from across the shape; (ii) spatially periodic repetitions in decoder attention responses; (iii) mismatched normal information in encoder and decoder query features; and (iv) missing supervision of the intrinsic relation between signed distance fields and surface normals. To address these issues, we introduce Align-VAE with four corresponding modifications: (i) a shared bank of learnable decoder tokens provides a workspace for global computation; (ii) distance-biased cross-attention guides retrieval toward nearby predicted spatial anchors; (iii) normal-free encoder queries align the two stages’ access to normal information while retaining normals in the encoder keys and values; and (iv) a screened Poisson-inspired regularizer supervises surface values and normal-direction derivatives. These changes align latent retrieval and field prediction with geometry without enlarging the per-shape code. On the ABO dataset, Align-VAE reduces Chamfer distance by 26.6% compared to the standard VecSet VAE.
[CV-367] LightVAE: Towards Compact and Efficient Video Autoencoders
链接: https://arxiv.org/abs/2610.14726
作者: Shiqiao Gu,Yang Yong,Fengjun Zhong,Ruihao Gong,Chengtao Lv,Xiangyu Fan,Lei Yang,Xianglong Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 24 pages, code is available at this https URL
Abstract:Latent diffusion models enable high-quality video generation, but video VAE encoding and decoding remain computationally expensive. Existing methods use stage-specific lightweight operators, yet either constrain spatial and temporal computation jointly or remove temporal computation altogether, without distinguishing their compression sensitivities in pretrained Conv3D. To address these limitations, we present LightVAE, a compression framework that exploits redundancy in temporal modeling and network depth. LightVAE exactly reparameterizes eligible causal Conv3D layers into an aggregate spatial response and a difference-driven temporal correction. Based on their distinct compression sensitivities, temporal low-rank compression (TLR) keeps the aggregate spatial kernel dense and applies layer-specific low-rank approximation only to the temporal correction. An algebraically equivalent folded implementation converts the arithmetic savings into practical acceleration. LightVAE further uses width-preserving block pruning to reduce depth and probe-guided feature alignment, guided by feedback from a frozen teacher suffix, to recover quality after compression. The framework also extends to video encoders. Across three video VAE backbones and two GPU platforms, LightVAE consistently accelerates decoding. For the Wan VAE family, it achieves up to 2.24\times and 1.11\times the decoding throughput of state-of-the-art efficient VAE decoders on NVIDIA H100 and RTX~5090, respectively, with higher DAVIS PSNR and lower LPIPS. On MiniMax-H3, it achieves up to 2.97\times the original decoding throughput while retaining 99.51% of the original model’s overall VBench-2.0 score. Our code is available at this https URL.
[CV-368] PixARG: Pixel-Asynchronous Retrieval-Augmented Diffusion for Fast and Realistic Image Generation
链接: https://arxiv.org/abs/2610.14713
作者: Zewei Xin,Qinya Li,Chaoyue Niu,Fan Wu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to ACM Multimedia 2026 (MM '26) as a poster presentation
Abstract:While diffusion models achieve impressive image generation capabilities, they suffer from high computational costs, implausible artifacts, and limited adaptability to dynamic or private data due to static, centralized training. To address these, we propose PixARG, a novel retrieval-augmented generative framework that integrates dynamic real-world image retrieval into the diffusion pipeline to simultaneously enhance efficiency and visual realism. Specifically, PixARG constructs a pixel-level affinity map between the user’s text prompt and the retrieved image to guide a heterogeneous noise initialization. By modulating noise intensity across spatial regions, this mechanism preserves semantically relevant structural priors while perturbing less relevant areas, thereby significantly accelerating convergence. To accommodate this spatially non-uniform noise and ensure synchronized generation, we introduce a pixel-wise asynchronous denoising scheduler and a spatiotemporal denoiser conditioned on the affinity map. Furthermore, we derive an affinity-adaptive training objective that elegantly balances deterministic restoration with stochastic regeneration across pixels. Experiments on LAION and COYO datasets across offline, online, and user-specific settings show PixARG enhances realism and semantic alignment while drastically reducing diffusion steps.
[CV-369] PURE: Refining Feed-forward 3D Reconstruction with Unlabeled Post-training
链接: https://arxiv.org/abs/2610.14698
作者: Yaolong Zhu,Peng Li,Jiahao Lu,Chengfeng Zhao,Mengfei Li,Chao Wang,Wenyue Chen,Zekai Gu,Chengwei Ren,Ronggang Wang,Yuan Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by ACMMM 2026. Project Page: this https URL Github: this https URL
Abstract:Feed-forward 3D reconstruction models have achieved remarkable progress through large-scale training and strong learned priors. However, they remain fragile when applied to dynamic in-the-wild videos, where non-rigid motion, occlusions, and illumination changes are prevalent. A key limitation lies in their reliance on expensive depth and camera pose annotations, which hinders scaling to diverse real-world dynamic data and restricts generalization. To address this issue, we propose \textbfPURE, an unlabeled post-training refinement framework that improves depth and camera estimation without requiring ground-truth supervision. Instead of relying on annotations, PURE elegantly distills robust structural knowledge from state-of-the-art 2D priors into 4D feed-forward architectures. PURE leverages relative depth constraints, affine-invariant geometry distillation, and pose refinement based on feature correspondences to stabilize geometric predictions under challenging dynamic conditions. Extensive experiments on both synthetic and real-world dynamic benchmarks demonstrate that PURE delivers consistent improvements across multiple state-of-the-art architectures. Furthermore, our data-efficient paradigm effectively mitigates domain shifts, successfully unlocking the architectural limits of these base models in challenging, in-the-wild dynamic scenes.
[CV-370] Demiurge: Forging Navigable 4D Worlds via Autoregressive Video Rerendering
链接: https://arxiv.org/abs/2610.14685
作者: Hanxin Zhu,Jianqi Ma,Cong Wang,Zhihao Hu,Junyao Gao,Tianyu He,Junyan Lin,Jiayin Xiao,Jiateng Shou,Sen Liang,Weidong Zhang,Zhibo Chen,Jun Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Recent advances in video diffusion enable novel-view video synthesis from a single input video, supporting world modeling, embodied intelligence, and robotic simulation. However, existing methods process camera trajectories independently, lacking the cross-trajectory consistency needed for coherent, navigable 4D worlds. We introduce Demiurge, an autoregressive video rerendering pipeline built on a dynamic 4D point-cloud memory cache. It promotes cross-trajectory consistency through four components: (1) an appearance video diffusion model that synthesizes photorealistic novel-view videos conditioned on the 4D memory; (2) a memory-consistent video depth completion model that aligns known-region depths with cached geometry and completes missing regions, mitigating scale ambiguity and discontinuities during memory updates; (3) memory-aware dual-stream warping that provides high-fidelity visual guidance and occlusion cues for novel-view synthesis; and (4) geometry-aware video refinement that enhances perceptual quality. Extensive experiments demonstrate state-of-the-art performance and three key capabilities: coherent spatiotemporal lattice construction over dense multi-view trajectories; large-parallax video rerendering under challenging camera motions, including 180-degree rotations with no input-view overlap; and geometry-faithful outputs that can be directly optimized into high-quality 3D Gaussian Splatting representations with minimal artifacts. Project page: this https URL.
[CV-371] A Hybrid Kinematic GCN and Elastic Transformer for Efficient Indian Sign Language Recognition
链接: https://arxiv.org/abs/2610.14680
作者: Pushp Garg,Himangshu Sarma,Mrinmoy Ghorai
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Sign Language Recognition (SLR) presents complex challenges due to variable signing speeds and spatial intricacies. Specifically, Indian Sign Language (ISL) requires precise modeling of its inherently bimanual nature and is bottlenecked by the scarcity of large-vocabulary benchmarks and the inability of standard models to handle irregular signing velocities. To address this, a structurally-aware hybrid framework for efficient, gloss ISL recognition is proposed. To minimize computational redundancy, an entropy-driven SVD sampler that dynamically extracts informative keyframes is used. Because this subsampling creates irregular physical time gaps, the proposed architecture employs a Continuous Dual Positional Encoding paired with a distance-aware Elastic Transformer to preserve natural signing rhythm. Spatially, a Kinematically-Conditioned Adaptive GCN explicitly models bimanual geometry using unified bone-joint features. To validate our framework, a 120-class, 9,453-video native ISL dataset is created. Augmented by Physical Node Dropout and ArcFace optimization to mitigate subject-specific memorization, our framework achieves a competitive 89.8% accuracy, establishing a strong baseline for complex sign language modeling.
[CV-372] HumanPhys: A Human-Centric Physical-World Perception Benchmark for Real-World Visual Reasoning
链接: https://arxiv.org/abs/2610.14679
作者: Yiming Zhang,Enzhi Zhou,Junan Lin,Dulan Zhou,Weiqiang Wang,Yuge Huang,Jian Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We introduce HumanPhys, a human-centric benchmark for real-world physical perception, and a training corpus emitted by the same pipeline. Each question targets a physically grounded quantity related to people, including depth, inter-person distance, gaze angle, displacement, velocity, heading-step trajectory, blink count, and contact-deformation strength. The rendered probes additionally test illumination inference from body and face cues and camera ego-motion, both with and without a person in the frame. All labels are derived programmatically from sensor calibration, source metadata, or renderer state, rather than written by annotators or produced by estimators on the benchmark frames. HumanPhys systematically probes six dimensions of human-centric physical perception: spatial state, kinematics, blink events, deformation, illumination, and camera pose. To the best of our knowledge, no existing MLLM benchmark-and-training corpus jointly supervises this set of human-centered physical quantities under direct metadata- or renderer-derived labels. The six dimensions appear only partially dissociable across the tested MLLMs. We benchmark a range of open- and closed-source MLLMs on HumanPhys and find that all six capability dimensions remain challenging for current models. We hope HumanPhys will support future research on multimodal systems that perceive and reason about people in the physical world.
[CV-373] CAST-ASD: Causal Attention over Speakers and Time for Online Active Speaker Detection
链接: https://arxiv.org/abs/2610.14676
作者: Pratham Gupta,Haotian Qi,Tushar Sandhan,Gabriel Skantze
类目: Computer Vision and Pattern Recognition (cs.CV); Sound (cs.SD)
备注: 5 pages, 1 figure, 6 tables
Abstract:We present CAST-ASD (Causal Attention over Speakers and Time), a causal joint model for active speaker detection (ASD). Recent ASD models improve performance by jointly processing all faces in a scene, but operate offline, with the full clip available before any frame is labelled. Low-latency causal detectors instead process each face independently. CAST-ASD combines joint face reasoning with fully causal inference. It applies scene-level attention across faces within the current frame and causal temporal attention over each face’s history, enabling information sharing across speakers with zero lookahead. Its parameter count does not grow with the number of faces. On MSDWild, where most negative face-frames contain a visible competing speaker, CAST-ASD reduces competing-speaker false alarms by 32% relative to the causal per-face detector it extends, with larger gains as scene occupancy increases. It outperforms all causal baselines on both validation splits while using only one-tenth the inter-speaker parameters of a three-block fixed-roster convolution. Code and models are available at this https URL.
[CV-374] Competition Domains Matter: Selective Head Globalization for Training-Free Sparse Attention in Video Diffusion Transformers
链接: https://arxiv.org/abs/2610.14672
作者: Tiantian Liu,Meng Wan,Zhiqiang Liang,Jue Wang,Ningming Nie,Yangang Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Training-free sparse attention is a practical route to efficient video diffusion transformer inference, yet existing methods typically fix the competition domain at each query row for every attention head. Prior work primarily addresses how to estimate token-level importance within this fixed domain; whether the domain itself should adapt to head heterogeneity remains under-specified. We introduce EG-Sparse, a training-free framework that routes low-entropy heads to a grouped cross-head pool while retaining per-row local selection for high-entropy heads, using only scores already materialized by the sparse pipeline. A controlled diagnostic fixes the ranking score and varies only the competition rule: grouped cross-head selection reduces low-entropy-head reconstruction error by 9.8%, whereas forcing high-entropy heads into an all-head global pool increases it by 7.9%. Across four configurations, EG-Sparse achieves higher video quality while reducing attention density and GPU denoising latency compared with strong training-free sparse-attention baselines. Mechanism ablations further show that selective domain adaptation outperforms uniform, random, and fixed domain-assignment policies. These results support competition-domain adaptation as a distinct design axis within centroid-guided training-free sparse attention.
[CV-375] SPARE: Shared Perception and Adaptive Retrieval of Evidence for Streaming Video Understanding
链接: https://arxiv.org/abs/2610.14671
作者: Senbin Xu,Zijun Zhang,Xinhua Wu,Li He,Jubo Feng,Li Qiao,Zhen Gao,Yan Zhang,Jixiang Luo,Rongrong Ji
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Streaming video understanding requires answering questions from the observed video, with evidence drawn from recent observations or earlier events. Historical context can improve recall but may interfere with judgments based on recent frames. Its usefulness depends on both when it is introduced and which clips are selected, while relevance alone does not guarantee sufficient evidence for an answer. We propose SPARE, a framework that combines adaptive historical retrieval with visual encoder sharing. A lightweight gate predicts whether recent observations provide sufficient evidence to answer the question. When additional evidence is predicted to be necessary, a multimodal embedding model retrieves relevant historical clips to supplement the current context. LoRA instruction tuning trains the understanding model to use recent observations and retrieved history jointly and abstain when the evidence remains insufficient. To reduce redundant visual parameters, the understanding and retrieval branches share a visual encoder while retaining separate language backbones. Four linear adapters align the shared features with the retrieval model through distillation at the visual token and final retrieval embedding levels. With visual-only inputs, SPARE achieves 77.78% on the evaluated Real-time and Backward task groups of OVO-Bench and 73.31% on StreamingBench. Compared with Qwen3-VL-8B-Instruct evaluated at 2 fps, SPARE improves accuracy by 20.68 percentage points on OVO-Bench and 14.15 percentage points on StreamingBench.
[CV-376] ETRIS-4DGS: Efficient Post-Training Compression of 4D Gaussian Splatting ICME2026
链接: https://arxiv.org/abs/2610.14663
作者: Chengzhen Wu,Kelei Liu,Sicheng Li,Lu Yu,Yiyi Liao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ICME 2026
Abstract:4D Gaussian Splatting (4DGS) has emerged as a powerful dynamic scene representation, supporting real-time high quality rendering but suffers from prohibitive storage overhead. Existing methods typically attempt to solve this by treating compression as a per-scene optimization problem, limiting generalizability and applicability where low-latency compression is required. In this work, we leverage existing image and video codecs and propose a generalized post-training compression framework for 4DGS to address these issues. Specifically, we apply importance-based adaptive quantization to Gaussians according to their importance levels, effectively refining bitrate allocation while remaining fully compatible with existing codecs. To further improve compression efficiency, we utilize the statistical properties of 4DGS and propose motion- and importance-aware packing to reorganize attributes into codec-friendly 2D maps. Our approach can achieve over 40x compression rate in seconds while maintaining high quality, facilitating the practical deployment of 4DGS.
[CV-377] Guide: Faster Autoregressive Video Diffusion via Historical Trajectory Guidance
链接: https://arxiv.org/abs/2610.14657
作者: Zhuo Chen,Bin Zou,Fanyue Wei,Lixin Duan,Angela Yao,Wen Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 23 pages, 5 figures. Project page: this https URL
Abstract:Can autoregressive video diffusion be accelerated further when only four denoising steps remain? Existing feature caching approaches struggle in this regime, as large feature changes between denoising timesteps undermine both direct reuse and lightweight predictors. We observe that features across video chunks remain strongly correlated at matching denoising timesteps, offering a complementary source of guidance. Building on this insight, we introduce TGuide, a framework that accelerates autoregressive video diffusion through a lightweight Anchor-Transport-Correct (ATC) predictor guided by historical sampling trajectories. Our ATC combines two complementary references: the current chunk’s preceding-step features preserve its spatial structure, while the previous chunk’s same-timestep features provide information at the target denoising stage. By integrating these references through spatial retrieval and adaptive correction, ATC improves velocity prediction accuracy despite the large gaps between denoising steps. To mitigate error accumulation across both denoising steps and video chunks, we train the predictor on its own autoregressive rollouts, allowing it to adapt to states shaped by its own earlier predictions. An optional confidence-aware scheduler further adjusts the quality-efficiency trade-off. Experiments on Self Forcing, Causal Forcing, and HY-WorldPlay show that TGuide preserves generation quality while achieving denoising speedups of up to 1.89\times over the original 4-step generators. On Self Forcing, TGuide generalizes from 81-frame training videos to 333-frame generation, approximately four times the training length, while exceeding the original model’s VBench score.
[CV-378] Prospective Forcing: Learning to Look Ahead for Autoregressive Long Video Diffusion
链接: https://arxiv.org/abs/2610.14654
作者: Yifan Shen,Yewei Xia,Gongxu Luo,Peiyuan Zhu,Boyang Sun,Zhenhao Chen,Guangyi Liu,Guangyi Chen,Kun Zhang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Autoregressive video generators synthesize long videos through chunk-wise rollouts, yet they are typically trained to predict only the immediate future. This limited prediction horizon makes it challenging to maintain video quality over extended rollouts. To address this problem, we introduce Prospective Forcing, a multi-head prediction framework that injects long-range predictive signals into autoregressive video representations. Built on a rolling-window generator, Prospective Forcing attaches lightweight, offset-specific heads that share the same conditioning information and independently predict chunks at different future temporal offsets. By extending supervision beyond the next chunk, these heads encourage the shared history representation to retain information useful for longer-horizon prediction, supporting temporally consistent generation. At inference time, we accelerate sampling by using accepted future-chunk predictions to skip full-window backbone executions. We equip the head with learnable embeddings to generate multiple candidates for the next chunk, and a learned verifier either selects a candidate or rejects all and falls back to standard autoregressive generation.
[CV-379] AG: Calibrating SAM3 for Universal Image Segmentation without Dense Supervision
链接: https://arxiv.org/abs/2610.14651
作者: Hao Zhu,Heyu Sun,Shuo Jin,Jiayu Xiao,Feng Dai
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Under review
Abstract:The recent Segment Anything Model 3 (SAM3) extends powerful promptable segmentation from spatial cues to concepts, raising the prospect of a single foundation model for universal image segmentation. However, this promise often fails to translate into reliable performance on downstream segmentation tasks. In this work, we empirically identify concept recognition, rather than mask quality, as the primary bottleneck: SAM3 accurately delineates the objects it recognizes yet struggles to robustly determine which concepts are present. Driven by this, we present \textit\underlineText \underlineAnchor re-\underlineGrounding (\TAG), a simple yet effective framework that rectifies the concept recognition of SAM3 on the target domain without dense annotations. Specifically, \TAG introduces an \textitAnchor Residual Pathway comprising image-conditioned, low-rank adapters to re-ground text anchors in the target distribution and modulate subsequent text-conditioned interactions. Guided solely by image-level category labels, \TAG jointly optimizes binary cross-entropy with \textitSoft Boundary Ranking objectives to calibrate presence confidence and distinguish present concepts from confusable negatives. Building upon this, we devise \textitConfidence-Weighted Distillation to couple rich semantic context of a multimodal large language model with fine-grained spatial priors of SAM3, thereby achieving effective calibration in the absence of manual annotations. Extensive experiments demonstrate that \TAG unlocks SAM3’s segmentation potential with only 3.1M trainable parameters, achieving impressive performance across diverse scenarios, and notably, even surpassing leading mask-supervised methods on several metrics. \hrefthis https URLOur code will be publicly available at GitHub \faGithub.
[CV-380] Can Synthetic UAV Benchmarks Be Trusted? Real-Synthetic Validation for UAV Model Selection
链接: https://arxiv.org/abs/2610.14647
作者: Yuyi Zhi,Min Du,Bill Shi,Guodong Yuan,Junheng Liang,Peifan Liao,Xin Guan
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: 8 pages, 5 figures
Abstract:Synthetic UAV imagery can expand pre-deployment test evidence when task-specific Real data are scarce, but visual plausibility or semantic label alignment alone does not establish validity for model selection. We study when Synthetic evaluation preserves the model-selection conclusions obtained from Real UAV data through two complementary cases under fixed evaluation protocols. In a requirement-specific road-debris benchmark, four language-conditioned perception systems obtain lower absolute F1 on Synthetic data, yet the point-estimate ordering and top-ranked model are preserved; bootstrap analysis yields no supported pairwise reversal, and size-matched resampling preserves the top-model decision. In contrast, a closed-set stress test on a separately materialized business corpus, using four VisDrone-compatible detectors over eight empirically supported classes, changes the point-estimate top-ranked detector and reverses three of six pairwise point-estimate preference directions. A complementary audit of five UAV business scenarios further reveals substantial Real-Synthetic mismatch in class composition, object density, and target scale, including systematic under-representation of small targets. These results show that Synthetic UAV benchmark trust is conditional: semantic coverage is insufficient, and validity must be established empirically for the task, operating regime, and model panel of interest.
[CV-381] Automatic assembly modeling through attention-based assembly feature mate prediction and assembly feature pattern recognition
链接: https://arxiv.org/abs/2610.14633
作者: Ziheng Dai,Wanbin Pan,Zhihong Tang,Shuming Gao
类目: Computer Vision and Pattern Recognition (cs.CV); Computational Engineering, Finance, and Science (cs.CE); Graphics (cs.GR)
备注: 28 pages, 12 figures, 5 tables
Abstract:Assembly modeling plays an important role in constructing product models, yet it is still mainly performed via interactive operations that are experience-dependent and labor-intensive. Data-driven strategies hold promise for automating this process; however, existing approaches either predict poses that cannot directly generate editable B-Rep assembly constraints or require prescribed part pairs or predefined assembly faces. To reduce the large candidate space of constraint prediction without predefined assembly faces, we propose a data-driven framework based on assembly feature mate prediction. The framework performs automatic assembly modeling from unassembled B-Rep parts, using assembly features to group the geometric entities of local mating interfaces and restrict candidate formation. Given the lack of suitable assembly feature annotations in existing datasets, we construct a large-scale dataset through an automatic annotation pipeline. Using the annotated dataset, we train an attention-based network to predict assembly feature mates, with Engineering Compatibility Fusion (ECF) incorporating explicit engineering compatibility into learned matching scores. To improve the accuracy of automatic assembly modeling, we use assembly feature patterns (AFPs) to guide the matching process. Finally, a post-processing search selects assembly feature mate and AFP mate candidates, whose associated constraints are generated and solved to complete assembly modeling. Experimental results show that the framework achieves assembly constraint accuracies of 74.79% on the six-assembly comparison set and 64.09% on the extended test set of 16 assemblies, which includes the comparison set. With ground-truth face labels supplied as additional input, the corresponding accuracies increase to 81.78% and 80.93%, respectively. Comments: 28 pages, 12 figures, 5 tables Subjects: Computer Vision and Pattern Recognition (cs.CV); Computational Engineering, Finance, and Science (cs.CE); Graphics (cs.GR) ACMclasses: I.3.5; J.6; I.2.6 Cite as: arXiv:2610.14633 [cs.CV] (or arXiv:2610.14633v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2610.14633 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-382] DuSRFlow: Dual-Lens Super-Resolution via Parallax-Aware Flow Estimation and Complementary Reference Warping
链接: https://arxiv.org/abs/2610.14630
作者: Fangpu Zhang,Huanjing Yue,Zifan Cui,Shaoxiong Tu,Qirui Yang,Jingyu Yang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by IEEE Transactions on Multimedia (TMM)
Abstract:Dual-lens super-resolution (DuSR) aims to enhance a wide-angle low-resolution (LR) image by exploiting a telephoto reference (Ref) image. A central challenge in DuSR is reliable reference utilization under large parallax, occlusion, and cross-camera imaging discrepancies. Existing methods often employ off-the-shelf monocular optical flow models to provide geometric priors. However, their assumptions of photometric consistency and smooth motion are violated in dual-lens scenarios, leading to unreliable alignment and degraded reference-based reconstruction. To address these issues, we propose DuSRFlow, a dual-lens super-resolution framework that improves reference utilization from both correspondence learning and reference aggregation. First, we introduce DuFlowNet, a dedicated dual-lens flow estimator for robust cross-view alignment. To train DuFlowNet without ground-truth flow annotations, we develop a real–synthetic data fusion strategy that composites synthetic foreground objects onto real dual-lens image pairs, generating controllable large-parallax samples while preserving realistic cross-camera imaging characteristics. Second, we observe that reference reliability is highly spatially uneven in DuSR, where unreliable local reference responses may suppress the utilization of reliable cues. We therefore propose a Complementary Reference Warping strategy that coordinates patch-based warping and kernel-free-guided deformable warping according to local reference reliability, producing a more reliable reference representation before LR–Ref fusion. In addition, we construct DuSR-RealV2, a challenging real-world benchmark with complex depth variations, large parallax, and occlusion. Extensive experiments on multiple DuSR benchmarks demonstrate that DuSRFlow achieves state-of-the-art performance in reconstruction quality, perceptual quality, and cross-dataset generalization.
[CV-383] EarthAlign: Learning Language-Aligned Representations of Earth System Data
链接: https://arxiv.org/abs/2610.14624
作者: Fadillah Adamsyah Maani,Numan Saeed,Jiaqi Ma,Dat Tien Nguyen,Youssef Wehbe,Duncan Watson-Parris,Salman Khan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Interpreting Earth system conditions is central to understanding environmental risks and informing decisions across weather- and climate-sensitive sectors. Existing multimodal weather systems primarily target task-specific prediction, generation, or question answering, while general-purpose alignment between Earth system data and language remains underexplored due to the limited availability of diverse language supervision grounded in meteorological conditions. To fill this gap, we introduce EarthAlign, a large-scale dataset with broad geographic coverage spanning 2000-2024 that combines 10 expert-curated natural event and disaster data sources with climatology-derived normal-weather samples. Each sample includes temporal and geospatial information and four grounded caption styles aligned with 63-variable ERA5 meteorological data. Building on EarthAlign, we introduce MeteoLIP, a dual-encoder framework for learning aligned meteorology-language representations. MeteoLIP incorporates MetTok, which uses a geolocation- and time-conditioned query to aggregate variable-specific patch embeddings into a compact sequence of meteorology tokens. To comprehensively assess Earth system multimodal representations, we introduce three downstream tasks: cross-modal retrieval, weather-condition classification, and physical variable estimation. Our experiments show that MeteoLIP improves upon the corresponding CLIP and SigLIP baselines across these tasks. Further analyses highlight the importance of meteorological input, temporal and geospatial conditioning, and multi-style supervision for meteorology-language alignment. Thus, EarthAlign and MeteoLIP provide a foundation for learning and evaluating language-aligned representations of Earth system data. Our code, dataset, and weights are publicly available at this https URL.
[CV-384] EviScene: Beyond “Is It AI?” for Agent ic Image Factuality Assessment
链接: https://arxiv.org/abs/2610.14609
作者: ZiAn Wang,Ruiqi Liu,Zhengbo Zhang,Yi Han,Jialiang Shen,Ruoxin Chen,Manni Cui,Ximo Zhu,Jin Jiang,Mingzhe Liu,Linhao Wang,Wenhao Wang,Shu Wu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Determining whether an image is AI-generated does not establish whether the situation it depicts is factually supported. We formulate Image Factuality Assessment: given an image and a generic verification instruction, a system returns a supported or refuted verdict with an evidence-grounded explanation. Without a supplied textual claim, the system must progressively refine factual questions during investigation and align external evidence with the specific situation depicted. We introduce EviLens, an expert-reviewed benchmark of 10,174 supported and refuted images from controlled construction and web collection, covering twelve factual error types and evaluating both verdict accuracy and evidence sufficiency. We develop EviScene, a multimodal ReAct agent that interleaves visual inspection, external retrieval, and evidence analysis, using new findings to refine its questions and revisit relevant image details. We train a compact policy with supervised fine-tuning on reviewed teacher trajectories, followed by privileged on-policy self-distillation applied to verified decisions from the policy’s own investigations. On the 1,527-image test set, the trained Qwen3.5-9B agent achieves 79.05% balanced accuracy and a Strict Evidence Sufficiency Rate (SESR) of 52.26%, improving over the same-backbone base agent by 6.71 and 9.37 percentage points, respectively. Its SESR is the highest among the evaluated this http URL, models, and data will be released at this https URL and this https URL.
[CV-385] Bidirectional Adaptive Transformers for Multimodal Anomaly Detection
链接: https://arxiv.org/abs/2610.14598
作者: Yuxin Jiang,Yunkang Cao,Weiming Shen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The effective integration of features derived from RGB images and 3D point clouds presents a substantial challenge in the field of multimodal anomaly detection (MAD). Existing MAD methods typically rely on static fusion paradigms built upon frozen pre-trained encoders. While these frameworks facilitate spatial alignment, they lack the mechanism to establish latent semantic correspondence, thereby inducing cross-modal interference that degrades anomaly discriminability. To mitigate these limitations, this study proposes a Bidirectional Adaptive Transformers (BAT) framework, which constructs comprehensive and semantically aligned multimodal representations by leveraging complementary attributes across modalities. The BAT framework introduces a Bidirectional Feature Alignment (BFA) module, which employs a mutual predictive mechanism to bridge the gap between 2D textural patterns and 3D geometric structures. This bidirectional interaction explicitly enhances the textural granularity of point cloud descriptors while augmenting the spatial contextual awareness of RGB features. To ensure that the latent representations possess sufficient informational density for alignment, BAT incorporates a parameter-efficient adaptation architecture. By utilizing lightweight image and point cloud adapters, the framework facilitates specialized fine-tuning of pre-trained models, mitigating the information loss inherent in frozen encoders. Extensive experiments conducted on MVTec 3D-AD and Eyecandies confirm that BAT achieves state-of-the-art anomaly detection performance, significantly outperforming existing MAD methods. Subjects: Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2610.14598 [cs.CV] (or arXiv:2610.14598v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2610.14598 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Yuxin Jiang [view email] [v1] Sat, 10 Oct 2026 05:35:15 UTC (7,701 KB) Full-text links: Access Paper: View a PDF of the paper titled Bidirectional Adaptive Transformers for Multimodal Anomaly Detection, by Yuxin Jiang and 1 other authorsView PDFHTML (experimental)TeX Source view license Additional Features Audio Summary Current browse context: cs.CV prev | next new | recent | 2026-10 Change to browse by: cs References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[CV-386] oF-Wheel Odometry: Navigating Degeneration of Localization With Limited Sensing Capabilities
链接: https://arxiv.org/abs/2610.14590
作者: Xu Yang,Wenhao Li,Zhengyu Wei,Zening Li,Longxiang Huang,Bo Wang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: 9 pages, 4 figures, 5 tables. Accepted by IEEE Robotics and Automation Letters (RA-L)
Abstract:Wheel odometry provides an efficient motion prior for ground robots but is prone to transient violations such as wheel slip or terrain-induced tilt. In filtering-based odometry, these violations frequently trigger divergence when they coincide with constrained geometric observability (e.g., featureless corridors). We address this challenge by proposing ToF-Wheel Odometry (TWO), a tightly-coupled filtering framework that integrates wheel kinematics with geometric constraints from Time-of-Flight (ToF) cameras. The primary contribution of this work is a degeneracy-aware, directional covariance-shaping strategy that replaces heuristic online covariance tuning with geometry-conditioned inflation. The core idea is to infer process noise from geometric corrections with a strict reliability check. We decompose the posterior-to-prior correction into a wheel-explainable subspace and an orthogonal remainder. The process covariance is anisotropically adjusted according to the observability of the ToF update: if the geometry is degenerate in non-kinematic directions, the remainder is suppressed as sensor noise; if the geometry is distinct, it is admitted as a valid kinematic violation. Additionally, we introduce an intra-frame feedback loop that retrospectively repairs the process covariance within the current frame to reduce structural lag during bursty disturbances. Extensive experiments on challenging scenarios demonstrate that TWO improves trajectory accuracy and robustness compared to static and adaptive Gaussian filtering, robustly navigating the trade-off between kinematic reliability and limited sensing capabilities.
[CV-387] Attribute-Centric Vision-Language Alignment for Fine-Grained Gait Understanding
链接: https://arxiv.org/abs/2610.14589
作者: Saihui Hou,Chenye Wang,Wenpeng Lang,Qiannan Guo,Zhengxiang Lan,Chao Fan,Shiqi Yu,Yongzhen Huang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Walking sequences reveal human-interpretable attributes related to body structure, posture, and motion. Gait attribute recognition provides a concrete route to fine-grained gait understanding by predicting these attributes. However, existing benchmarks provide limited fine-grained annotations, while classifier-based gait attribute recognition methods typically represent candidate classes through learned output weights rather than directly matching visual evidence with textual class representations. To address the annotation gap, we present RA-GAR, a large-scale, richly annotated multimodal benchmark for gait attribute recognition, containing 1,001 subjects, 242,729 walking sequences, and annotations for 15 static and motion gait attributes. We further propose Gait-VL, an attribute-centric vision-language alignment framework that recognizes gait attributes by matching attribute-specific multi-frame visual evidence with textual class prototypes. On the visual side, attribute-specific learnable queries aggregate RGB features across spatial locations and frames into per-attribute visual embeddings. On the language side, diverse descriptions expanded by a large language model are combined with context about relevant body regions to construct textual class prototypes. Beyond attribute recognition, we evaluate the transferability of the attribute-pretrained visual encoder by using it to replace generic upstream large vision models in identityoriented gait recognition pipelines; attribute pre-training uses no identity classification or metric-learning objectives. Experiments demonstrate state-of-the-art attribute recognition performance on RA-GAR and consistent improvements in downstream identity recognition across multiple pipelines and benchmarks, supporting the effectiveness and transferability of attribute-centric gait representation learning.
[CV-388] Flow-Support-Weighted Proposals for Test-Time Verified VLA Policies AAAI2027
链接: https://arxiv.org/abs/2610.14588
作者: Chen Zhao,Zipeng Ye,Haitao Shen,Guanlin Li,Yuchen Xie,Jing Zhang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: 35 pages, 5 figures, under review in AAAI 2027
Abstract:Test-time sampling promises to turn one action prediction into a set of distinct executable candidates that a verifier can choose among. Downstream fine-tuning works against that promise. A vision-language-action model trained to solve one task distribution concentrates its flow-matching decoder around the dominant demonstrated action, so repeated samples differ only marginally and the verifier receives many copies of one decision. We call this proposal contraction and identify it as a key constraint on test-time scaling: repeated samples can remain highly redundant as the candidate budget grows. Action Dynamic Sampling (ADS) alleviates the constraint on the proposal side, during training. At each training step, ADS integrates K additional trajectories from the current policy under fresh base noise, scores each by the flow support along its path with a Hutchinson trace estimator, and corrects the policy’s local action distribution toward this weighted set through a detached finite-set KL term. Training cost rises; deployment cost is unchanged. Across simulation and four physical-robot tasks, ADS raises proposal spread without the success penalty that Gaussian perturbation incurs, lifts oracle best-of-8 success on RoboCasa from 57.6% to 63.0%, improves fixed verifiers by 2.4 to 4.9 points in simulation, and raises TokenPB by up to 13.0 points on real-world manipulation.
[CV-389] Remembering What Matters for Long-Horizon Robotic World Models
链接: https://arxiv.org/abs/2610.14583
作者: Hao Wu,Tianyu Shao,Shengju Qian,Yizhou Zhao,Fan Zhang,Penghao Zhao,Weiyan Wang,Yufei Huang,Fan Xu,Kun Wang,Xin Wang,Ming-Hsuan Yang,Yuxuan Liang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: 35 pages
Abstract:Autoregressive robot video world models based on discrete visual tokens remain unreliable over long rollout horizons. Retaining the entire generated history does not necessarily improve prediction quality. We investigate a possible memory-role conflict in Full Context decoding: persistent scene information, recent visual-action states, and stale generated history compete for attention despite their different predictive roles. This perspective motivates ReCAP (Receding Context with Anchor Preservation), a training-free context-scheduling method for long-horizon robot video world models. ReCAP keeps the pretrained tokenizer, world model, and visual decoder frozen. It preserves the initial scene anchor and a contiguous window of recent complete visual-action blocks while evicting older generated history. The schedule preserves within-block structure and temporal order, requires no additional modules or parameter updates, and bounds active-context growth independently of rollout length. We evaluate ReCAP on RT-1, BridgeV2, LIBERO-90, and CALVIN using frozen autoregressive world models. Across evaluated environments and horizons, ReCAP improves visual fidelity, action-conditioned consistency, and robot structure preservation relative to Full Context. On BridgeV2, it reduces LPIPS by 63.3%. In the system benchmark at H=64, it reduces the logical KV footprint by 73.3% and p50 latency by 6.51%. Attention analyses show increased attention to retained recent states, while history interventions associate larger amounts of stale context with higher prediction error. These observations are consistent with the memory-role-conflict hypothesis, although they do not isolate it from context-length and position-handling effects. Our results support inference-time context scheduling as a practical approach to improving long-horizon robot video prediction and computational efficiency without retraining.
[CV-390] FrameFold: Response-Guided KV Cache Consolidation for Autoregressive Long Video Generation
链接: https://arxiv.org/abs/2610.14568
作者: Muyu Liu,Mingming Liu,Hong Gu,Yuyao Zhang,Qin Lu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 26 pages, 14 figures
Abstract:In autoregressive video generation, retaining all historical key-value (KV) states increases memory usage and per-step attention costs as videos grow longer. A fixed cache budget bounds these costs but raises the challenge of preserving historical content that supports consistent, high-quality generation. Existing methods control cache size through history selection or consolidation, yet deciding what to retain and how to limit the loss of useful content remains a shared challenge. Our analysis shows that spatial queries depend on different historical content and that higher current attention coverage does not necessarily yield better long-video quality. We introduce FrameFold, a training-free KV cache consolidation method that compares the content retrieved by the same recent historical queries before and after merging to guide both representative construction and merge selection. Exploiting temporal redundancy between adjacent historical entries, FrameFold merges their KV states into shared representatives while retaining complete spatial grids. It fits value-merging coefficients in closed form by matching conditional attention responses, then selects the merge with the lowest remaining response distortion. Under matched KV cache budgets, experiments on Self Forcing and Causal Forcing at 30 and 60 seconds and on LongLive at 120 seconds demonstrate improved consistency and visual quality. Continuous eight-minute Self Forcing rollouts further show less quality degradation than the baselines.
[CV-391] Adversarial Gradient Forcing for Robust Few-Step Autoregressive Video Generation
链接: https://arxiv.org/abs/2610.14553
作者: Chenxi Xia,Yixuan Xu,Shan Liu,Jiapeng Wang,Maohua Li,Hanlin Tang,Kan Liu,Tao Lan,Weiyi Lu,Xinggao Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL ; Code: this https URL
Abstract:Autoregressive video generation models generate videos sequentially, with each new block conditioned on previously generated blocks. By training on the model’s own rollouts, Self Forcing and its follow-up methods substantially mitigate exposure bias caused by the mismatch between training- and inference-time history distributions. However, these models still suffer from error accumulation during long-horizon generation. We identify a key underlying limitation: the model remains highly sensitive to errors in its generated history, allowing small deviations to be amplified through subsequent generations. We refer to this underexplored problem as the local history robustness gap. To address this gap, we introduce Adversarial Gradient Forcing (AGF), a framework designed to improve the robustness of few-step autoregressive video models to perturbations and accumulated errors in their own generated histories. AGF uses distribution matching gradients from future block predictions to select adversarial perturbations of self-generated histories. It then jointly trains the generator on the original and perturbed histories to improve the robustness of future generation. Extensive experiments show that AGF achieves the highest VBench Total score on 60-second video generation and the best visual quality on four-minute video generation. Notably, AGF achieves such long-horizon robustness while requiring only 5-second rollouts for training.
[CV-392] InfraWorld: Motion-Aware World Modeling for Roadside 4D Occupancy Forecasting
链接: https://arxiv.org/abs/2610.14544
作者: Songkai Wang,Xiaokai Bai,Lei Yang,Hui-liang Shen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Roadside 4D occupancy forecasting predicts future semantic occupancy in a fixed infrastructure coordinate system. Roadside scenes are globally structured yet locally evolving, requiring comprehensive scene-level context while precisely modeling sparse localized motion. However, existing forecasting paradigms struggle to tackle both requirements. To address this limitation, we propose InfraWorld, a world model formulating roadside occupancy forecasting as selective state evolution. InfraWorld maintains a dense BEV state for the complete scene and persistent motion queries for moving regions, assigning complementary states to global context and local dynamics. Specifically, we propose Dynamic Query Initiation (DQI) to initialize queries in regions supported by motion in the observed history. During rollout, the dense state is updated through Long–Short Temporal Fusion (LST), which combines short-term adjacent-frame continuity with long-term motion trends to preserve stable context while capturing scene evolution. Motion-Aware Query Injection (MAQI) then refines the queries and injects confidence-weighted query information into corresponding BEV tokens, enabling targeted dynamic updates without perturbing global context. Together, InfraWorld concentrates predictive capacity on evolving traffic regions. Extensive experiments on the InfraOcc and Occ3D–nus datasets demonstrate that InfraWorld achieves state-of-the-art performance for roadside 4D occupancy forecasting while retaining its effectiveness under ego-centric settings. Code will be released.
[CV-393] ScienceBuddy-PV: A Verifier-Governed Recursive Improvement Protocol for Panvascular Intervention Research
链接: https://arxiv.org/abs/2610.14523
作者: Lingsen You,Yujun Guo,Xinyu Zhong,Zijie Feng,Wentong Wang,Li Shen,Junbo Ge
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: 17 pages, 17 figures, 19 references. Public-data component audit; recursive improvement is specified as a protocol and not experimentally demonstrated
Abstract:ScienceBuddy-PV adapts the ScienceBuddy recursive-in-recursive design to panvascular intervention research through a structured evidence ledger, device-vessel ontology, abstention contract and retained failure bank. Recursive harness optimization and outer-loop model learning, including reinforcement learning, are specified but not executed. We audit a fixed component, rather than the complete agent system, using separate ARCADE syntax and stenosis tasks (1,000 training, 200 validation and 300 test images per task). A shared task-masked convolutional model was trained once per representation setting; its parameters and the heuristic support gate were then held fixed, with the stenosis threshold selected on validation data. On the locked test sets, syntax Dice was 0.475 (95% CI 0.459-0.490) and stenosis Dice was 0.231 (95% CI 0.214-0.249). Stenosis pixel AUROC was 0.957 and AUPRC was 0.180 at prevalence 0.00975. The gate accepted 76.7% of stenosis-frame outputs versus 100% for a score-only comparator. In a constructed mask audit, it accepted 44.3% of reference candidates and rejected 55.7%, exposing the cost of abstention. Acceptance does not establish correctness or clinical safety. This public-data component audit makes support and abstention measurable and provides a reproducible protocol for future collaborative agent studies. It does not demonstrate recursive improvement, patient-level validity or clinical device recommendations.
[CV-394] We Have Been Interpreting Diffusion Transformer Backbone Behaviors Wrong
链接: https://arxiv.org/abs/2610.14517
作者: Benyuan Meng,Longtao Huang,Yang Liu,Liang Lin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:During a network forward call, diffusion backbones naturally produce many intermediate activations. If we evaluate each activation against an identical downstream benchmarking protocol, we can plot an inverted-U-shaped curve as the depth profile: the activation’s performance goes up first and then down as we go into deeper layers. This is common knowledge in the diffusion model community, which inspires many applications including the widely-adopted REPA. However, the common interpretation of this depth profile may be inaccurate and incomplete: (i) The cause of this curve has long been treated as signal granularity changing from coarse to fine, but there has been little evidence in DiT backbones. (ii) The community has been mainly studying this phenomenon in a static way, without investigating how it changes across different training setups. In this work, we show a different mechanism picture for the profile, where the prediction target (e.g., \epsilon , u , x_0 ) sits at the center. The semantically meaningful signal is only an intermediate by-product for predicting the target, while the prediction target is the real goal for the backbone. Furthermore, the choice of prediction targets as part of the training setup can predictably affect the curve’s shape. With this updated interpretation, many applications of this depth profile can be better grounded and predicted. As an example, we show how to predict the near-optimal placement of REPA based on the prediction target of the training setup. The code is at this https URL.
[CV-395] Compositional Concepts Erasing: Disentangling Object-Style via Domain Manifold Unlearning
链接: https://arxiv.org/abs/2610.14514
作者: Yang Xiao,Kaiyuan Deng,Gen Li,Zinan Ling,Zhiwen Fan,Xiaolong Ma,Bo Hui
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 14 pages
Abstract:Diffusion models pose significant risks regarding the generation of illicit content and copyright violations. This has catalyzed the development of machine unlearning to erase harmful concepts or copyrighted styles from a model’s generative repertoire without compromising its general performance. However, previous unlearning approaches on diffusion models have largely focused on either object-level or style-level removal and failed to address the challenge of object-style combination unlearning. At the same time, multiple unlearning requests often lead to catastrophic forgetting. To address these challenges, we propose Domain Manifold Unlearning (DMU), which uses an EMA-updated feature proxy extracted from intermediate diffusion features to represent the retained target-object direction. As a lightweight estimating representation of an object, DMU enables selective unlearning of domain-specific components in a specific object while preserving other semantic representations, thereby significantly enhancing robustness in multiple combination unlearning scenarios. Extensive experiments show that DMU significantly mitigates catastrophic forgetting, outperforming state-of-the-art methods of batch unlearning and sequential unlearning.
[CV-396] artanMatch: Towards Universal Dense Correspondence Across Modalities ATC
链接: https://arxiv.org/abs/2610.14513
作者: Hyeokjoon Kwon,Jiting Cai,Ruogu Li,Kritan Bhandari,Geethika Hemkumar,Parv Maheshwari,Yuheng Qiu,Yuchen Zhang,Sebastian Scherer,Wenshan Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Robotics (cs.RO)
备注: Project page: this https URL
Abstract:Robotic platforms combine multi-modal sensors to operate across diverse conditions, but most dense correspondence methods are limited to RGB-RGB or a small set of modality pairs. Training a model for each modality pair scales poorly with the number of sensors and forgoes potential positive transfer across modality pairs. We introduce TartanMatch, a unified model for dense correspondence across all 25 ordered pairs among RGB, event, thermal, depth, and projected LiDAR. TartanMatch adapts an RGB-pretrained correspondence model to all modality pairs within one shared network using lightweight modality adaptation and a mixture of real and synthetic data. On zero-shot real-world benchmarks, TartanMatch reduces Endpoint Error (EPE) over the strongest baselines by 61.9% for cross-modal matching and 49.6% for same-modal matching. We further show that joint training across all modality pairs improves average matching accuracy over pair-specific training, especially on challenging cross-modal pairs. These results demonstrate that a single model achieves broad multi-modal coverage and state-of-the-art matching accuracy.
[CV-397] Spatiotemporal Recognition of Abdominal Wall Layers During Optical Trocar Entry
链接: https://arxiv.org/abs/2610.14512
作者: Shreeram Athreya,Andrew Melehy,Carter Prentice,Michael Robinson,Corey Arnold,James Wu,William Speier
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 16 pages, 3 figures
Abstract:Objective. To evaluate whether spatial context and temporal modeling improve tissue recognition during optical trocar entry, where multiple abdominal-wall tissues may be visible simultaneously. Methods. We retrospectively analyzed 71 recordings with 1,516 annotated keyframes and 29,151 clinician-annotated sectors. Each frame was divided into 28 polar-grid sectors assigned to nine tissue classes. Color, neighboring-sector, and positional features were used by a gradient-boosted emission classifier. The full pipeline added learned temporal smoothing, spatial regularization, an anatomically ordered hidden Markov model, and posterior blending. Five-fold cross-validation was grouped by recording, with cross-fitting to prevent within-recording leakage. Primary evaluation used clinician-derived targets; ablations assessed classification, probability quality, spatial consistency, and peritoneum-or-deeper onset timing. Results. The complete pipeline achieved accuracy 0.646, macro-F1 0.507, and macro-AUROC 0.882, versus 0.600, 0.440, and 0.861 for emissions alone. Macro-F1 improved by 0.066 across all five folds. Label changes decreased from 66.2 to 19.5 per sector-minute, and median onset error decreased from 15.58 to 2.45 seconds. Frames with a tissue-depth-code span of at least three decreased from 53% to 26%, versus 22% for clinician annotations. Performance remained limited for preperitoneal fat and bowel, and the full model retained a 44% premature-onset proportion. Conclusions. Combining appearance, spatial context, and temporal constraints improved retrospective tissue recognition and onset localization. Tissue-specific limitations, residual timing errors, and reliance on future observations warrant external validation and prospective evaluation of causal models before clinical use.
[CV-398] DSR-Net: Learning Dynamic Skeletal Relations for 3D Human Pose Estimation
链接: https://arxiv.org/abs/2610.14499
作者: Fan Li,Jian Xiang,LinLin Xue,Yuwei Gao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 16 pages, 10 figures
Abstract:3D human pose estimation requires accurate modeling of motion-dependent skeletal structures and inter-joint dependencies. Existing methods typically rely on fixed skeletal topologies in Graph Convolutional Networks (GCNs) or implicitly capture joint relations through attention mechanisms, limiting their adaptability to diverse motion patterns. To address this issue,we proposeDSR-Net, a dynamic structural representation learning framework for 3D human pose estimation. Specifically, a Dynamic Skeletal Relation (DSR) module is introduced to learn adaptive inter-joint dependencies beyond predefined skeletal connections. Based on the learned relations, a hierarchical relation-aware fusion strategy is developed to facilitate effective interaction between structural and temporal representations. In addition, a lightweight Residual Skeletal Recalibration Module (RSRM) is designed to enhance the stability of structure-aware features with minimal computational overhead. Extensive experiments on Human3.6M and MPI-INF-3DHP demonstrate that DSR-Net achieves MPJPEs of 37.9 mm and 14.9 mm, respectively, consistently improving over strong baselines. The results validate the effectiveness of dynamic skeletal relation modeling for robust 3D human pose estimation.
[CV-399] Multi-modal Representation Fusion for Long-term Microsurgical Anastomosis Trajectory Anticipation
链接: https://arxiv.org/abs/2610.14492
作者: Yuxuan Liu,Yating Luo,Yunfei Luan,Xinyao Zhou,Yuyang Zhuge,Yao Guo,Guang-Zhong Yang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: 16 pages, 13 figures
Abstract:Microsurgical anastomosis is one of the most challenging and sophisticated procedures for surgical robotics, requiring accurate manipulation of surgical instruments within a limited space to achieve precise operations. Accurate surgical trajectory anticipation is essential for surgical scene understanding, intraoperative planning, robotic manipulation, and reliable behaviour analysis. However, existing approaches mainly focus on features extracted from surgical videos and historical trajectories, while overlooking rich semantic and motion information that can provide more discriminative multi-modal representations for complex surgical instrument-tissue interactions. In this work, we propose a cross-modal feature fusion framework for long-term 2D and 3D microsurgical anastomosis trajectory anticipation. The proposed framework consists of a multi-modal representation extractor and a cross-modal feature fusion network. Semantic recognition of surgical objects and consecutive motion estimation are incorporated as multi-modal visual cues, providing complementary spatial and temporal guidance for instrument-tissue interaction characterization. A coarse-to-fine progressive fusion strategy with a long-short-term memory mechanism is introduced to capture temporal dependencies step by step. The training and evaluation are conducted on our established robotic platform and a comprehensive dataset containing both phantom and ex-vivo surgical scenarios is constructed. Extensive quantitative and qualitative experiments demonstrate that the proposed framework consistently outperforms existing methods and exhibits improved robustness across different surgical scenes, highlighting its potential for both intelligent robot-assisted and autonomous surgical systems.
[CV-400] FreeCHROMA: Generative Chroma Keying from Unaligned Background References
链接: https://arxiv.org/abs/2610.14445
作者: Jiayi Wu,Tianfu Wang,Rui Zhao,Florian Schiffers,Zhimin Chen,Cornelia Fermuller,Yezhou Yang,Christopher Metzler,Yiannis Aloimonos
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:In film production, chroma keying composites an approved foreground performance into a new environment, but conventional workflows require coordinated background plates and dedicated harmonization. Recent generative methods often require pre-aligned backgrounds, focus on background-conditioned foreground relighting, or represent the environment only through an edited first frame. These restrictions limit background selection, scene composition, and camera motion while adding preparation or editing effort. We present FreeCHROMA, which adapts a pretrained video diffusion model in a parameter-efficient manner for generative video compositing from one to three unaligned background references. The model integrates the foreground into the referenced environment, preserving foreground identity, motion, and fine visual details while generating a spatiotemporally coherent background. Foreground-background harmonization occurs within the same generation process, without a separate relighting stage. Optional text instructions help disambiguate complex spatial relationships and camera behavior. To train the model, we construct pseudo-paired data from readily available real monocular videos through foreground-background separation and augmentation with existing foundation models. The original videos serve as training targets, while automated processing and quality verification enable scalable data construction. Across reference fidelity, foreground preservation, scene integration, harmonization, and overall video quality, FreeCHROMA consistently outperforms recent compositing and conditional video-generation baselines, with particularly clear gains under camera motion. FreeCHROMA relaxes the constraints of existing tools to enable end-to-end generative chroma keying from unaligned inputs, with exploratory experiments demonstrating its potential in robotics.
[CV-401] SceneForesee: Learning Scene Evolution for Spatiotemporal Reasoning
链接: https://arxiv.org/abs/2610.14439
作者: Chaoyue Li,Shoudong Han,Haoyang Wu,Boxue Yang,Sijia Chen,Yufan Chen,Ruping Zhang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Large multimodal models (LMMs) need to understand how entities move and interact as a scene develops. This requires following the same entities over time and relating their observed motion to subsequent changes. We formulate Scene Evolution Understanding (SEU), a task for predicting and interpreting this change. Given observed video, a model estimates entity histories and predicts one scene continuation through future trajectories and Events, accompanied by Visual Evidence from observed motion and a scene description. We construct SceneForesee, a benchmark for evaluating entity continuity and scene evolution reasoning through paired observed and future scenes with shared entity identities. It contains 144.57K temporal pairs, 1.05M entity correspondences, and 272.49K Events paired with observed evidence. We propose a spatiotemporal reasoning framework, EvoSee, that learns future geometry and Event semantics through shared candidate states and joint assignment. It selects one continuation per entity from observation and uses the selected changes for entity refinement, retention, and scene representation. On SceneForesee, EvoSee outperforms Qwen3.5-9B by 10.45 points in Combined Point-F1 and 17.24 points in Event Global F1. On the 4D reasoning benchmark Dyn-Bench, it improves question answering accuracy and grounding JF by 4.8 and 3.58 points over their respective base systems. These results support scene evolution as both a prediction target and a dynamic representation that brings predicted entity changes into spatiotemporal this http URL project is available at this https URL.
[CV-402] Design Choices for Multimodal Transformer Learning with Missing and Longitudinal Healthcare Data
链接: https://arxiv.org/abs/2610.14436
作者: Rita Cordeiro Mendes,Maria Manuel Pereira,Carlos Santiago,Catarina Barata
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal learning offers the potential to improve predictions by integrating complementary information from heterogeneous data sources. However, multimodal datasets are frequently incomplete, raising important questions about how models should handle missing information. Although recent approaches increasingly rely on Transformer-based architectures for multimodal fusion, incomplete multimodal data introduces challenges such as missing modalities and intra-modal missingness, which motivate key architectural design choices, such as tokenization granularity. In addition, many datasets are longitudinal, with observations collected across multiple time points. The sequential nature of the Transformer architectures provides a natural framework for modeling such data, but introduces additional considerations, such as how temporal information should be represented. In this study, we systematically investigate these design choices using three different cancer datasets. We compare modality-level and feature-level tokenization and evaluate attention masking and imputation as strategies for handling missing information, both naturally occurring in the datasets and synthetically induced. We further investigate approaches for incorporating temporal information in our training process and how to model it. Experiments are conducted with unimodal and multimodal configurations, as well as different levels of missingness for binary survival classification and multiple time encoding strategies for discrete-time survival prediction. Our results show that while multimodal outperforms unimodal results, missing data heavily impacts predictions and thus effective design choices such as tokenization granularity, and intra-missing handling is essential for reliable prediction.
[CV-403] VisualCoding: Learning Executable Skills for Reliable Visual Reasoning
链接: https://arxiv.org/abs/2610.14403
作者: Liu Jiaqi,Jin Ruinan,Li Xiaoxiao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Learning reusable skills can improve visual reasoning agents, yet applying a skill still requires the agent to implement or invoke a procedure and produce the final answer. Errors and hallucinations at these steps can prevent acquired experience from translating into reliable single-run performance. We propose VisualCoding, which combines execution-guided acquisition of reusable visual programs with direct adoption of their answers on applicable inputs. Execution-Guided Program Acquisition composes visual primitives and additional code into family-level solvers, then iteratively tests and revises them on labeled training instances. Scope-Aware Direct Execution runs a retained solver and returns its valid in-scope answer. We evaluate VisualCoding on four models across four benchmarks spanning 69 task families. VisualCoding exceeds the strongest evaluated baseline in mean accuracy by 4.45 percentage points with Gemini-3.1-Pro and by 14.78 points when reusing the frozen Pro-acquired artifacts with Flash-Lite. Ablations support execution-guided acquisition and show that direct execution outperforms textual or callable reuse of the same programs. Code is available at this https URL.
[CV-404] Learning Transferable World-Action Models from Task-Paired Human Videos ATC
链接: https://arxiv.org/abs/2610.14351
作者: Jeongeun Park,Taekyung Kim,Dongyoon Han,Sangdoo Yun
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: Project Page: this https URL
Abstract:World-action models inherit broad visual and physical priors from video pre-training. However, standard within-video next-frame prediction allows them to rely on appearance shared between the context and future, rather than learning interactions that transfer across scenes and embodiments. We introduce WATCH, a cross-video pre-training that learns from task-paired human demonstrations collected across diverse environments. Conditioned on the video and actions of one demonstration, the model jointly predicts another demonstration’s actions and future frames from its current observation. Predicting one demonstration from another encourages the model to learn task-relevant interactions that transfer across scenes and demonstrators. Our pre-training shows that these capabilities can be learned from readily scalable human-human pairs alone, without requiring matched human-robot demonstrations. Analyses before robot post-training show improved motion alignment across embodiments and better action and video prediction on unseen robots, with representations closer to those learned from robot pairs. After robot post-training, WATCH improves MimicDroid in-context success from 40.9% to 68.3% with Cosmos Policy and from 59.8% to 77.7% with Cosmos3, with the largest gains on the unseen levels. Gains persist under context-free post-training across robot embodiments. WATCH outperforms within-video pre-training and co-training baselines, supporting cross-video prediction as a scalable approach to learning transferable world-action representations.
[CV-405] raceSampler: Making Sampling Errors Predictable and Correctable for Accelerated Generative Sampling
链接: https://arxiv.org/abs/2610.14332
作者: George Matthew(1),Guoqiang Zhang(1),Kenta Niwa(2) ((1) University of Exeter, United Kingdom, (2) NTT Communication Science Laboratories, Japan)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Accurate sampling from continuous-time generative models can require many model evaluations. Low-step sampling reduces cost, but coarse updates introduce endpoint errors that can accumulate and degrade generation quality. We introduce TraceSampler, a post-training method that corrects these errors without modifying the pretrained model. Rather than optimising low-step updates directly, TraceSampler models endpoint error as a sample-specific prediction problem. A higher-step reference integration over the same interval provides the supervised target, while a compact representation makes high-dimensional error predictable from the sampling trace. This representation combines shared residual structure, trace-derived directions, and a low-rank remainder, so shallow per-step MLP heads predict sample-dependent coefficients. These coefficients drive a two-stage learned residual correction: the first stage predicts the full endpoint residual, while an intermediate backbone evaluation provides new trace information to predict the remaining residual; this evaluation is reused by the following step. TraceSampler is evaluated with Rectified Flow on CIFAR-10, SiT-XL/2 on ImageNet-256, and SD3.5 Medium for text-to-image generation. At 10 steps, it reduces trajectory MSE relative to matched-step Euler by 90.5% on CIFAR-10 and 85.6% on ImageNet-256, while achieving FID scores of 2.692 and 2.189, close to 120-step Heun references of 2.601 and 2.056 with 24 \times fewer backbone evaluations. On SD3.5 Medium, TraceSampler improves image quality and prompt following; at 10 steps, ImageReward rises from 0.752 to 0.906 while trajectory MSE falls by 70.7%. These gains preserve Euler’s backbone-evaluation budget, add less than 0.4% inference FLOPs, and incur negligible wall-clock overhead, showing that coarse-step sampling error can be made predictable from the sampling trace and efficiently corrected.
[CV-406] Real-Time Traffic Density Estimation: A Hybrid Computer Vision and Predictive Machine Learning Pipeline
链接: https://arxiv.org/abs/2610.14326
作者: Prince Kwabena Appiah Boadu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 32 pages, 61 figures, 6 tables, including appendix
Abstract:Traffic density estimation from camera feeds requires connecting visual perception with short-horizon forecasting. We present a two-stage pipeline that combines vehicle detection and tracking with a spatial logic layer, followed by a temporal estimator. The spatial layer filters bounding boxes by their overlap with a precomputed road mask and extracts per-frame traffic features. LSTM, GRU, and Temporal Convolutional Network (TCN) estimators use these features to predict five future steps from a ten-frame input window. We benchmark ten detector variants from six families on a dataset of 5,716 annotated frames using a single NVIDIA GPU. YOLOv5l achieves the highest mAP@50:95 of 0.524 at 56 frames per second, while YOLOv8n reaches 258 frames per second at 0.487 mAP@50:95; these rates measure detector inference without tracking or spatial filtering. Temporal evaluation on 29,243 held-out windows yields aggregate R-squared values of 0.858 for GRU, 0.856 for LSTM, and 0.840 for TCN. These metrics measure forecasts of detector-derived telemetry and are dominated by direction variance, rather than establishing absolute traffic-density accuracy. No road mask was available for the evaluation stream, limiting validation of the spatial filtering component. A live run on a third camera stream examines forecasting behavior within the complete pipeline. The results characterize detector speed-accuracy trade-offs and the limitations of short-window traffic telemetry forecasting.
[CV-407] Manifold Navigation Learning Using Recurrent Networks: Efficient Training Techniques
链接: https://arxiv.org/abs/2610.14325
作者: Sarah Harkins Dayton,Sathya Rengaswami,Vasileios Maroulas,David L. Boothe,Ioannis D. Schizas
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Position tracking is a well-studied problem in Euclidean navigation, yet comparatively little attention has been given to navigation on general manifolds, where motion is governed by the underlying geometry of the space. In this work, we formalize the problem of manifold navigation as a velocity-driven tracking task and propose a coordinate-chart-based learning framework in which learning models operate entirely in a Euclidean representation. Predicted trajectories are mapped to the manifold via an immersion function prior to error computation, allowing geometric consistency to be evaluated without explicitly enforcing manifold constraints during learning. Within this framework, we focus on Gated Recurrent Units (GRUs) for sequential position prediction on curved manifolds after comparisons with alternative recurrent architectures. We evaluate our GRU models, trained with pertinent custom loss functions, against both a standard MSE loss function and a naive vector-addition baseline. We evaluate our approach on both analytically (and numerically) defined manifolds, as well as a surface reconstructed from a LiDAR point cloud. Our results show that recurrent models can learn to approximate manifold-constrained dynamics from velocity sequences and consistently outperform Euclidean baselines. Our results demonstrate that certain custom loss functions significantly improve long-horizon prediction accuracy compared to conventional point-wise loss functions.
[CV-408] EV-JEPA: Elastic V-JEPA for Any Device and Cross-Device Inference
链接: https://arxiv.org/abs/2610.14324
作者: Anxhelo Diko,Jieyu Lin,Stefano Ambrogio,Ziyun Li,Mahmoud Assran,Théo Moutakanni,Barbara De Salvo,Chiao Liu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Foundation models like V-JEPA are increasingly deployed across a heterogeneous edge-cloud continuum spanning smart glasses, phones, robots, and servers. Yet, they are typically trained as rigid, fixed-capacity networks bound to one device, supporting neither reconfigurable compute budgets nor collaborative execution. Existing approaches address only one axis at a time: slimmable and NAS-based methods provide width elasticity but assume single-device inference, whereas split-computing methods enable cross-device collaboration only for fixed partitions and static capacities. We introduce Elastic V-JEPA (EV-JEPA), a single model trained once and deployable everywhere along the continuum. Its width is reconfigurable, and it is natively partitionable across heterogeneous devices. At its core is Transition-Aware Triplet Training (T3). At each step we pick a narrow and a wide width, and train three paths: each width alone, plus a hybrid path that runs the early layers narrow and the rest wide. The hybrid switch point is where one device hands off to another, so the same weights learn both single-device and cross-device execution. Across various benchmarks, EV-JEPA falls within 1 accuracy point of independently trained baselines while improving over previous elastic training approaches by up to 11 accuracy points. Narrow-to-wide cross-device hybrid execution recovers to within 1.1 points of the highest-capacity standalone model and supports many-to-one, any-to-any, and variable-split configurations. Finally, we show that the elastic principle extends beyond the backbone to an elastic token compression mechanism that exposes controllable trade-offs among on-device computation, latency, and accuracy.
[CV-409] MvFFN: Multi-view Floor plan Feed-forward Network for Unposed Wide-Baseline Panorama Layout Reconstruction NEURIPS2026
链接: https://arxiv.org/abs/2610.14315
作者: Yuguang Li,Yichuan Deng,Zixuan Liu,Ivaylo Boyadzhiev,Xiuyuan Wu,Xinyu Zou,Linda Shapiro,Alex Colburn
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted at NeurIPS 2026
Abstract:Reconstructing accurate camera poses and floor-plan layouts from sparse, unposed, wide-baseline RGB panoramas remains a challenging open problem. Direct multi-room floor-plan prediction is especially difficult in this setting: pose, geometry, semantics, and global layout structure are tightly coupled, yet a complete floor-plan is hard to learn as a single global output. Our key insight is to keep the model’s predictions strictly local. Multi-view Ordered Wall Instance Segmentation (MOIS) is designed around this concept. Each panorama column predicts only the wall it observes plus the next two adjacent walls in clockwise order. The global wall partition, room layouts, and shared coordinates emerge from deterministic aggregation of overlapping local predictions. Building on MOIS, we propose Multi-view floor-plan Feed-Forward Network (MvFFN), which jointly predicts coarse camera poses, dense global geometry, and per-wall semantic outputs. A downstream Layout Chaining (LC) post processing pipeline then assembles these predictions into multi-room floor-plans with shared global coordinates, consistent wall identities, and connected room structure. The dense MOIS predictions recover cross-view wall identity and local within-room connectivity more accurately than composed modular baselines, and the assembled layouts substantially outperform prior methods on wall and junction level accuracy. We also introduce the ZInD-CrossView dataset, which augments ZInD with globally unique wall instances and cross-view segmentation labels for this task.
[CV-410] ConicFlow: constrained Inference-time optimal control for flow matching models
链接: https://arxiv.org/abs/2610.14307
作者: Shervin Khalafi,Kasra Fallah,Sung Cho,Dongsheng Ding,James Anderson,Alejandro Ribeiro
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Systems and Control (eess.SY)
备注:
Abstract:We propose ConicFlow, which frames constrained sampling from a flow matching model at inference time as an optimal control problem that balances deviation from the pretrained trajectory against reward, subject to hard terminal constraints. We discretize this problem and adapt receding-horizon control, optimizing only a short window of the trajectory at each step for computational tractability. We address the resulting sequence of receding-horizon subproblems with a receding-horizon primal-dual algorithm. Its dual multipliers capture the difficulty of satisfying each constraint and are refined throughout sampling, steering generation toward feasible final samples. We further propose a terminal-estimate variant that avoids accumulating local truncation errors by evaluating the reward and constraints at the predicted terminal point rather than at the end of the current horizon. Under standard regularity assumptions, we establish finite-horizon guarantees on suboptimality. Under convex-affine structure, we derive a sharper bound that scales linearly with omitted horizon length, and we show that the terminal-estimate variant is provably tighter when the flow dynamics change slowly. Experiments in both low- and high-dimensional settings with a recent large-scale flow matching model show that ConicFlow achieves high reward and strong constraint satisfaction while deviating less from the pretrained flow than recent baselines.
[CV-411] Controllable Diverse Sampling through Pullback Directions NEURIPS2026
链接: https://arxiv.org/abs/2610.14294
作者: Sylvain Dehayem Kenfouo,Aaron Zweig,Mingxuan Zhang,Elham Azizi
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at the DynaFront workshop at NeurIPS 2026
Abstract:Conditional diffusion models can generate high-quality images that align closely with a given prompt. However, samples generated from different random seeds may remain visually similar, limiting creative exploration. Existing approaches can increase variation, but often require additional optimization or sampling overhead, while stronger perturbations may introduce distortions that degrade image quality or prompt alignment. We propose a training-free method that leverages the local pullback geometry induced by the mapping from the conditioning-embedding space to the image space to identify directions of maximal output sensitivity. By steering different sampling trajectories along these directions, we encourage distinct generations while maintaining consistency with the conditioning signal. Experiments on MS-COCO and BrushNet show the efficacy of the method in improving diversity on prompt-conditioned image generation and inpainting.
[CV-412] GeoFix: Geometry Fixing for 4D Consistent Generation
链接: https://arxiv.org/abs/2610.14293
作者: Yeobin Hong,Hyojun Go,Dohun Lee,Jeongjae Lee,Sunghwan Hong,Federico Tombari,Zhaochong An,Jong Chul Ye,Konrad Schindler,Dominik Narnhofer
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Project Page: this https URL
Abstract:Video diffusion models increasingly serve as priors for 4D generation, but the generated videos often lack geometric consistency. Recent approaches address this through post-training by designing geometric rewards, yet this can degrade texture and fine detail, requiring auxiliary appearance rewards and careful reward balancing. To address this challenge, we propose GeoFix, a unified framework that improves geometric consistency and visual quality with a single geometric reward, applicable to both training and inference-time reward alignment. Our reward measures cross-view depth consistency in static scene regions, targeting structural inconsistencies without directly constraining appearance or penalizing valid object motion. Crucially, depth agreement increases with camera field-of-view overlap even in real videos and 4D datasets, allowing a smaller camera movement to be mistaken for better geometry. We account for this effect by calibrating the reward against the overlap-dependent trend. GeoFix achieves a 39% improvement in WorldScore 3D consistency and a 4% increase in VBench Quality Score.
[CV-413] GSR: Text-Grounded Diffusion for Faithful Image Super-Resolution
链接: https://arxiv.org/abs/2610.14281
作者: Zhiming Hu,Ran Zhang,Angela Ye,Raghav Goyal,Faraz Ali,Tristan Ty Aumentado-Armstrong,Samrudhdhi B. Rangrej,Konstantinos G. Derpanis,Iqbal Mohomed,Alex Levinshtein
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 22 pages, 9 figures, 12 tables
Abstract:Diffusion models have achieved remarkable success in image super-resolution (SR) by leveraging strong generative priors to enhance perceptual quality. However, for textual content, these priors often produce text hallucinations: non-existent or distorted text, which alter scene semantics and compromise model reliability. While recent methods focus on enhancing text generation, they do so indiscriminately, leaving hallucinations fundamentally unresolved. We argue the core problem is a lack of selective control. Therefore, we reformulate text SR as a selective generation problem: text is resolved only when reliable generation is expected, while other scene content is fully super-resolved. We introduce TGSR (Text-Grounded Super-Resolution), a test-time framework that progressively identifies and suppresses likely hallucinated text regions during generation. To enable this, we train dedicated detectors on a large-scale dataset curated from diverse diffusion-based SR outputs. TGSR consistently improves text fidelity across multiple backbones, increasing recognition precision by up to 17.68 points without degrading perceptual quality, underscoring the value of selective control in generative SR.
[CV-414] RoboFake: Are Deepfake Detectors Ready for Humanoid Face Robots?
链接: https://arxiv.org/abs/2610.14269
作者: Hao Chen,Junnan Xu
类目: Computer Vision and Pattern Recognition (cs.CV); Cryptography and Security (cs.CR); Robotics (cs.RO)
备注: 10 pages, 4 figures, 6 tables
Abstract:Humanoid face robots can now move their lips in sync with speech and show detailed facial expressions. Give such a robot the face of a real person and drive it with generated speech, and you get a face forgery that happens in front of the camera: the camera records a real object, and no pixel is edited. We call this a RoboFake. In this paper we ask what current deepfake detectors do when they see a robot face. We test six public detectors trained on FaceForensics++ on video of a physical lip-syncing face robot, on a synthesized video of the same robot, and on rendered robot faces. All thresholds are fixed on FaceForensics++, and every result is compared with real human faces that went through the same processing as the robot videos. At a threshold that flags 5% of real faces, three detectors flag almost every physical robot frame, while two flag fewer than 7%. A matched human control and masking show that the response comes from the face, not from the processing or the background. However, it does not track synthesis: with only the face visible, three of the six detectors give the physical robot a higher fake score than its synthesized video. The response is also fragile, and light JPEG compression is enough to push most robot frames below the threshold for some detectors. Our results suggest that current detectors flag today’s robots because their faces do not look human, not because they recognize a forgery. We release our evaluation protocol and code, and state three testable hypotheses about realistic, interactive RoboFakes.
[CV-415] hinking with Images without Coordinates
链接: https://arxiv.org/abs/2610.14265
作者: Meng Cao,Haoze Zhao,Yi Wang,Ian Reid,Xiaodan Liang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Recent advances in multi-modal large language models have shifted toward the “thinking with images” paradigm to improve reasoning faithfulness, yet existing methods rely heavily on explicit coordinate- based grounding. We identify two fundamental limitations in this approach, i.e., a semantic gap where sparse regressions miss high-dimensional latent structures, and a resolution gap due to the granularity mismatch between patch-level tokens and pixel-level coordinates. To address these, we propose TiCO (Thinking with images without COordinates). It reformulates visual reasoning as a semantic evidence discovery process within the latent space, and evidence is injected as token-level regions without pixel coordinates. Specifically, TiCO utilizes a dedicated probe token to estimate similarity distributions and performs importance sampling over spatially connected components to generate diverse, semantically dominant evidence rollouts. Extensive experiments across multiple high-resolution and general visual reasoning benchmarks demonstrate that our TiCO consistently outperforms state-of-the-art coordinate-based methods with significantly improved inference efficiency.
[CV-416] BigCell: Generating Gigapixel Whole-Slide Images NEURIPS2026
链接: https://arxiv.org/abs/2610.14261
作者: Srikar Yellapragada,Alexandros Graikos,Zilinghan Li,Kostas Triaridis,Tarak Nath Nandi,Ji Dong K Bai,Beatrice Knudsen,Tahsin Kurc,Prateek Prasanna,Rajarsi R. Gupta,Ravi K Madduri,Joel Saltz,Dimitris Samaras
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Accepted at NeurIPS 2026. Project page: this https URL
Abstract:Whole-slide images (WSIs) are gigapixel-scale digital scans of histopathology tissue and serve as the primary basis for cancer diagnosis. Existing histopathology generative models operate at the patch scale, producing fixed-size tiles of at most 1024 \times 1024 pixels which encompass only a small fraction of the WSI. In contrast, text reports for clinical diagnoses and slide-level diagnostic labels – the supervision that exists at scale – operate at the whole-slide level. We present BigCell, the first approach to generate coherent whole-slide histopathology images at native gigapixel resolution. BigCell trains a flow-matching transformer at 2048 \times 2048 pixels on over 120,000 publicly available WSIs spanning multiple organ types. Conditioning on embeddings from pathology foundation models and using an inference-time sliding-window pipeline, BigCell generates coherent images of up to 32,768 \times 32,768 pixels, outperforming prior tile generators on both tile-level and slide-level fidelity metrics. To make gigapixel synthesis practical, we introduce an adaptive scheduling strategy that prioritizes important slide regions, speeding up image generation by over 3\ttimes with minimal quality degradation. Furthermore, we propose the first text-to-image synthesis at 8k resolution by training a separate model that produces dense conditioning grids directly from pathology text reports. We demonstrate that augmenting with synthetic samples improves slide-level few-shot classification performance by up to 18 % over baselines trained on real data.
[CV-417] Image Diffusion Models Are Good 3D Human Motion Generators
链接: https://arxiv.org/abs/2610.14254
作者: Siyuan Yang,Chong Wang,Hao Cheng,Jun Liu,Bihan Wen,Alex Kot,Hedvig Kjellstrom
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 16 pages; under review. The first two authors contributed equally to this work
Abstract:Text-to-motion generation typically relies on dataset-specific skeletal representations, with models trained separately for benchmarks with different joint layouts. We investigate whether image diffusion architectures can model 3D human motion through a common image-compatible representation that accommodates heterogeneous skeletal layouts while preserving motion-specific kinematic structure. To this end, we organize root-centered joint trajectories into topology-informed joint–time images and model global root dynamics through a complementary latent stream. A motion autoencoder conditions local-motion decoding on the root latent, while a text-conditioned Scalable Interpolant Transformer (SiT) jointly generates both latent streams, with supervision on decoded 3D joint positions further constraining generation in motion space. In addition to dataset-specific models, we train a single motion autoencoder on HumanML3D and KIT-ML, followed by a single text-conditioned SiT shared across both datasets, preserving their respective skeletal definitions without skeleton retargeting to a common joint layout. Experiments on HumanML3D and KIT-ML show that our dataset-specific models achieve state-of-the-art performance, while shared training yields further gains on both benchmarks, demonstrating the benefits of shared modeling across heterogeneous skeletal layouts.
[CV-418] SliderEdit: Improved Slider Trajectories in Instruction-Based Image Editing
链接: https://arxiv.org/abs/2610.14253
作者: Arman Zarei,Sayan Nag,Bryan Wang,Franck Dernoncourt,Varun Manjunatha,Nedim Lipka,Soheil Feizi,Samyadeep Basu
类目: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
备注:
Abstract:Slider-based control aims to give users continuous, fine-grained control over the intensity of individual edits in instruction-based image editing. However, existing approaches fail to deliver reliable continuous controllability: traversing a slider can produce abrupt visual jumps and non-monotonic changes, while jointly controlled sliders can interfere with one another. To systematically study these limitations, we introduce SlideBench, a comprehensive benchmark for continuous image editing spanning diverse editing categories and equipped with a unified suite of metrics quantifying monotonic progression, perceptual continuity, and inter-slider disentanglement. Building on these observations, we propose SliderEdit++, which explicitly regularizes the intermediate editing trajectory rather than supervising only its endpoints. Our method combines velocity- and semantic-space trajectory regularization with spectral control to encourage smooth, predictable slider traversal, together with a disentanglement objective that reduces interference between independently controlled edits. On SlideBench, SliderEdit++ achieves state-of-the-art performance, improving the overall score by 18.5% relative to the strongest baseline. At the same time, using a few-step editing backbone makes our method approximately 3x to 20x faster than existing slider-based approaches, enabling substantially more responsive interactive editing.
[CV-419] CAPTOR: a vision-language foundation model for Chest-Abdomen-Pelvis Oncologic CT
链接: https://arxiv.org/abs/2610.14226
作者: Séléna Jean-Mactoux,Sophie Loizillon,Ghina Jardali,Amandine Crombé,Nathalie Lassau,Alexandre Bône
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Purpose : To develop and evaluate CAPTOR, a deep-learning vision-language foundation model for oncologic chest-abdomen-pelvis (CAP) CT interpretation. Materials Methods : CAPTOR (Chest-Abdomen-Pelvis foundaTion model for Oncologic Radiology) is a contrastive vision-language model, trained using a pretraining strategy for oncologic CAP CT (in single- or multiple-series acquisitions), on a multicenter retrospective dataset from North America, Europe and Asia, predominantly from tumor lesion examinations. CAPTOR was compared with state-of-the-art CT foundation models on three downstream tasks: disease detection, survival prediction using TumorImagingBench and image retrieval. Classification and retrieval were both evaluated on held-out internal and public test sets, stratified by anatomical coverage and indication. Bootstrap resampling was used to estimate confidence intervals, and pairwise comparisons were performed using permutation tests. Results : CAPTOR was trained and tested on 25,512 and 10,794 CT-report pairs respectively. TumorImagingBench datasets comprised between 134 and 421 studies each. CAPTOR outperformed all baseline models, on the classification task, on the CAP subset (AUC 0.767, 95%CI: 0.734-0.793, vs. 0.695, 95%CI: 0.645-0.736 for the next-best, p0.001) and oncologic subset (AUC 0.784, 95%CI: 0.759-0.805, vs. 0.734, 95%CI: 0.703-0.767 p0.001), ranked first on three of the four oncologic prognostic benchmarks, and achieved superior retrieval performance on CAP and oncologic subsets. Conclusion : A foundation model trained on oncology-enriched CAP CT data outperformed relevant state-of-the-art models in its target clinical context, underscoring the importance of clinically aligned pretraining. We release our preprocessing, training and inference code : this https URL and model weights : this https URL Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) Cite as: arXiv:2610.14226 [cs.CV] (or arXiv:2610.14226v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2610.14226 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Sophie Loizillon Dr [view email] [v1] Fri, 9 Oct 2026 18:39:01 UTC (4,108 KB) Full-text links: Access Paper: View a PDF of the paper titled CAPTOR: a vision-language foundation model for Chest-Abdomen-Pelvis Oncologic CT, by S’el’ena Jean-Mactoux and Sophie Loizillon and Ghina Jardali and Amandine Cromb’e and Nathalie Lassau and Alexandre B^oneView PDFHTML (experimental)TeX Source view license Additional Features Audio Summary Current browse context: cs.CV prev | next new | recent | 2026-10 Change to browse by: cs cs.AI References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[CV-420] Agreement versus Softmax under Spatial Disambiguation NEURIPS2026
链接: https://arxiv.org/abs/2610.14210
作者: Michał Zakrzewski,Mariia Shpir,Taras Rumezhak,Volodymyr Karpiv
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: 7 pages, 2 figures. Accepted at the NeurIPS 2026 Workshop on Embodied Spatial Reasoning (non-archival)
Abstract:ScanRefer Multiple queries require spatial disambiguation: several instances of the target class are present, and the referencer must pick one. When per-query logits were never logged, a common substitute is cross-epoch agreement. On the full ScanRefer validation set (9,508 queries; GeoClassifier with ground-truth proposals), object-level softmax is similarly miscalibrated on Unique and Multiple (ECE 0.068 vs. 0.102; the scene-bootstrap gap interval includes zero). Fixed-final agreement on the same checkpoints reports ECE 0.041 vs. 0.272 (gap +0.231). Modal-vote agreement is 0.045 vs. 0.318. Most Multiple errors are same-class wrong-instance (45.6%); geometric miss (H2) is 59.6%, and wrong-category (H1) is 16.7%. A non-monotone H1 pattern does not appear under the default fixed-final protocol, and we retract it. Logit-free thresholds rank spatially hard queries poorly. If only agreement was logged, calibrate abstention on Multiple and report wrong-instance rates with H1.
[CV-421] oken-Budgeted Candidate Retrieval for 3D Scene-Grounded VLMs: Category Names vs. Pairwise Geometry NEURIPS2026
链接: https://arxiv.org/abs/2610.14207
作者: Michał Zakrzewski,Mariia Shpir,Taras Rumezhak
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 10 pages, 2 figures. Accepted at the NeurIPS 2026 Workshop on Physical World AI (non-archival)
Abstract:Under a fixed context budget, which objects and relations should a 3D scene-grounded VLM see first? On ScanNet indoor referring benchmarks (Dai et al., 2017), matching the query to object category names outperforms pairwise geometry scoring at every reported K. The gaps are largest at high recall. Analytic pairwise cues beat per-object cue scoring and query-agnostic kNN graphs, including on paraphrases. MiniLM and BGE-small still score higher on those splits. We evaluate token-budgeted candidate retrieval with axis-aligned boxes, a token-ratio metric, training-free baselines, two small encoders that match q to each object’s category name, and QCRR (Query-Conditioned Relation Retrieval), a deterministic pairwise-geometry scorer. Across ScanRefer gt_val plus a custom 20% scan-id split of Nr3D and Sr3D (35,504 queries; not official ReferIt3D val), additive noun matching reaches R@10 0.864 on ScanRefer against QCRR’s 0.626; MiniLM and BGE-small reach 0.911 to 0.913 on ScanRefer and 0.941 to 0.947 on Nr3D. A corrected 3DGraphLLM harness (native Acc@0.25 0.67 on the full graph, n=300 ScanRefer queries) shows that context selection changes downstream accuracy under a matched object budget (Random 0.44, QCRR 0.56, MiniLM 0.67, oracle 0.70). Semantic retrieval beats hand-designed geometry on that harness. We will release the protocol, retriever implementations, and per-query outputs so new selectors can be swapped in without rewriting the analysis.
[CV-422] ClothCode: Generative Modeling for Garment State Estimation and Dynamics Prediction
链接: https://arxiv.org/abs/2610.14200
作者: Yingdong Ru,Lipeng Zhuang,Shiyu Fan,Zhuo He,Florent P. Audonnet,Gerardo Aragon Camarasa
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We introduce ClothCode, a unified generative framework for garment state estimation and action-conditioned dynamics prediction in a shared discrete configuration space. A material-aligned tokenizer encodes garment configuration as residual vector-quantized patch tokens, while task-specific Masked Generative Transformers predict these tokens through parallel confidence-guided refinement. Across three garment templates, ClothCode reduces mean displacement error by 26.6% for state estimation and 37.0% for dynamics prediction relative to the strongest baselines. Dynamics prediction requires only 16.7 ms per window, compared with 556.2 ms for diffusion. Integrated into the same model-predictive controller, ClothCode improves real-world folding and flattening success from 56.7% to 78.9%. Further results and videos are available at: this https URL
[CV-423] Semantic Edge-Cloud Communication for Real-Time Urban Traffic Digital Twins
链接: https://arxiv.org/abs/2610.14187
作者: Murat Arda Onsu,Arman Dogru,Poonam Lohan,Burak Kantarci,Aisha Syed
类目: Networking and Internet Architecture (cs.NI); Computer Vision and Pattern Recognition (cs.CV); Distributed, Parallel, and Cluster Computing (cs.DC); Signal Processing (eess.SP)
备注: 6 pages, 4 figures, accepted to IEEE Global Communications (Globecom) Conference, 2026
Abstract:Urban traffic systems are becoming increasingly complex, demanding real-time traffic perception, proactive collision prediction, and intelligent management to improve safety and reduce congestion. Although AI-driven traffic management has shown strong promise, its success relies on diverse, high-quality data spanning critical scenarios such as collisions, near-collisions, and traffic rule violations. Digital twins offer a powerful solution by enabling synchronized virtual replicas of physical traffic environments for real-time monitoring, collision prediction, traffic-flow optimization, and diverse realistic scenario generation. However, continuously synchronizing a high-fidelity urban traffic digital twin through edge surveillance cameras requires substantial edge-cloud communication, and transmitting raw video streams leads to severe bandwidth and latency constraints. To overcome this challenge, this paper proposes a semantic communication-based edge-cloud framework for real-time urban traffic digital twin applications. By leveraging edge intelligence to extract and transmit only task-relevant semantic traffic information instead of full video feeds, the proposed framework significantly reduces communication overhead while preserving the essential dynamics needed for accurate real-time digital twin reconstruction. Experimental results demonstrate more than 99% transmission savings and an average trajectory reconstruction error of only 0.12 m, confirming the proposed framework as an effective, scalable, and low-latency solution for AI-enabled urban traffic digital twin applications.
[CV-424] Disentangling Task-Relevant Latent Dynamics for Reward Modeling
链接: https://arxiv.org/abs/2610.14159
作者: Dayou Li,Jionghao Wang,Peihao Wang,Jiuzhou Lei,Hao Wang,Jichen Chai,Hezhen Hu,Minghui Zheng,Zhangyang Wang,Xin Li,Wenping Wang,Zhiwen Fan
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Reward design remains a major bottleneck for reinforcement learning in robotic manipulation. Handcrafted dense rewards can compensate for the limited guidance provided by sparse rewards, but require substantial task-specific reward engineering. Learning rewards from videos provides a scalable, data-driven alternative to manual reward design. However, existing video-based rewards are often computed in entangled visual spaces where task-relevant state changes are confounded with extraneous visual variation. As a result, the learned reward may respond to visual distractors instead of actual task execution. To address this limitation, we propose DiTaR (Disentangling Task-Relevant Latent Dynamics for Reward Modeling), a framework for learning dense rewards from heterogeneous videos. DiTaR learns a continuous task-relevant representation from multi-frame video latents through an asymmetric two-stage procedure that suppresses task-irrelevant visual variation. In the task-relevant latent space, we train a history- and language-conditioned rectified-flow model to capture task dynamics and derive a directional dense reward from temporal changes in dynamics compatibility. Together, these components enable reward modeling directly over task-relevant dynamics rather than generic visual variation. DiTaR substantially improves policy learning on MetaWorld-v3 and ManiSkill over video-based reward baselines, and transfers to real-world offline RL through reward relabeling. Reward analyses further show that DiTaR provides more reliable rewards.
[CV-425] Video-Arena: A Scalable Agent ic Video Benchmark for Spatiotemporal Capability
链接: https://arxiv.org/abs/2610.14158
作者: Kuan Zhang,Yukun Chen,Zhihao Yang,Yue Su,Xiangnan Wu,Zirong Chen,Run Luo,Tao Tan,Yinhe Zheng,Yiming Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 36 pages. Project page: this https URL
Abstract:Understanding and tracking a dynamic world are fundamental to visual intelligence. Existing video benchmarks provide limited insight into whether models can maintain a coherent understanding of evolving world state. Building such a benchmark from real-world videos is challenging because the underlying world state is partially observable, making annotations hard to obtain. To close this gap, we introduce Video-Arena, a benchmark that turns competitive games into scalable, verifiable tests of how well models understand evolving world states from videos. Specifically, inspired by how humans understand dynamic events, we evaluate spatiotemporal world state understanding and tracking through three task categories: perceiving critical events, interpreting state updates, and tracking the resulting states. Our pipeline can generate videos and aligned reference labels from game records at scale. To keep evaluation affordable and reproducible, we curate a golden set of 180 tasks across 9 games and run evaluations under two protocols: native visual understanding and tool-assisted agentic analysis. The best-performing systems achieve F1 scores of and under the two protocols, respectively. The result highlights that native models struggle to perceive events reliably, while tool-assisted systems improve event recovery but still struggle to interpret state updates and track the resulting states consistently over time. Our project page is in this https URL
[CV-426] Axiom-0: Hierarchical Self-Refinement for Vision-Language-Action Policies
链接: https://arxiv.org/abs/2610.14155
作者: Renbin Shu,Jing Yang,Xiangyuan You,Guolin Qiu,Yufei Xia,Yiming Zhu,Nanjun Dai,Fei Zhang,Feng Wu,Keke Tang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Reliable manipulation requires policies that generalize across tasks, layouts, and object interactions. Vision-language-action (VLA) models combine pretrained multimodal understanding with learned control, but their performance remains limited on unfamiliar task compositions. We present Axiom-0, a VLA policy that addresses both the quality of its representations and their use in action generation through hierarchical self-refinement. Intra-VLM Representation Refinement (IVRR) strengthens selected intermediate features through consistency supervision with learned feature projections during training. VLM-Action Interface Refinement (VAIR) recursively updates the resulting conditioning before action generation. Joint optimization of the supervised VLM layers, VAIR, and the action generator connects the two components. On RoboCasa365, Axiom-0 achieves 62.32% overall success in our evaluation, exceeding the highest published leaderboard score (Paimon-0, 58.1%) by 4.22 percentage points and the Xiaomi-Robotics-1 reference by 4.92 points. Under the evaluated profiling setup, mean action-chunk generation latency increases by 1.7% over Xiaomi-Robotics-1. These results demonstrate improved task completion with modest inference overhead.
[CV-427] rajCFG: Token-Wise Classifier-Free Guidance via Denoising Trajectory Feedback
链接: https://arxiv.org/abs/2610.14153
作者: Jingyu Liu,Hao Jiang,Yikang Ding,Huafeng Shi,Xu Yang,Xiaoqiang Liu,Guoxin Zhang,Pengfei Wan,Kun Gai
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 22 pages. Project page: this https URL
Abstract:Classifier-free guidance (CFG) typically applies a shared scale to all latent tokens, although their velocity predictions evolve differently during sampling. We introduce Trajectory-Adaptive Classifier-Free Guidance (TrajCFG), a training-free method for flow-matching models that adapts token-wise guidance using historical changes in guided velocity direction. TrajCFG maps this feedback to bounded scales, assigning stronger guidance to larger directional changes and weaker guidance to more consistent directions. It combines adjacent-step and interval-normalized longer-span feedback, reusing standard CFG predictions without additional model evaluations. Controlled diagnostics and ablations examine the roles of directional feedback, token-wise allocation, and history-span fusion. Experiments on text-to-video, image-to-video, text-to-audio-video, text-to-image, and image-to-3D generation show improvements over base samplers across multiple backbones. For text-to-video generation, TrajCFG improves VBench Total scores by 0.30-0.75 points over the strongest evaluated baseline on each of three backbones. These results support historical directional feedback as a broadly applicable signal for fine-grained guidance allocation. The project page is available at this https URL.
[CV-428] Scenes as an Attack Surface: Uncovering Scene-Grounded Safety Failures in MLLM s
链接: https://arxiv.org/abs/2610.14094
作者: Tongxi Wu,Yuyang Cheng,Jiacheng Xian,Hu Huang,Zhihao Wu,Menglin Yang,Jian Zhang,Yang Gao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Preprint
Abstract:Existing jailbreak attacks have exposed serious safety weaknesses in multimodal large language models (MLLMs). Most prior attacks rely on specially constructed visual inputs to carry or support harmful intent. In this paper, we uncover a different vulnerability: task-relevant scene semantics alone can shift an MLLM from refusal to compliance, even when the visual input introduces no additional attack instructions and the original task intent remains unchanged. We formalize this failure as a scene-grounded visual safety vulnerability and introduce SceneSafetyBench to investigate it across eight risk categories and nine white- and black-box MLLMs. Our analysis reveals that this failure is associated with how task-relevant visual evidence influences downstream safety representations, suggesting that visual grounding alone cannot fully explain the observed safety shift. Building on this finding, we propose Scene-Grounded Attacks (SGA), which preserve the original task while constructing validated scene-grounded queries and combining their independent responses. Across nine target MLLMs, SGA achieves an average attack success rate (ASR) of 86.17%, outperforming the strongest baseline by 8.55 percentage points and achieving the highest ASR on eight models. Our results identify scene semantics as an overlooked attack surface in MLLMs and highlight a growing tension between stronger visual grounding and robust multimodal safety alignment.
[CV-429] Human Pose Estimation by Probabilistic Mixtures of Gaussian and Uniform Distributions
链接: https://arxiv.org/abs/2610.14093
作者: Santiago Reina Alguacil,Ezequiel López Rubio
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 38 pages, 11 figures, 20 tables. Includes supplementary material. Code available at this https URL
Abstract:Despite advancements in sub-pixel decoding, modern Human Pose Estimation (HPE) architectures remain constrained by deterministic coordinate extraction from discrete heatmaps, which systematically fails under dense occlusions and Out-of-Distribution (OoD) ambiguities. To address these limitations, we propose a post-hoc continuous probabilistic framework that intercepts raw heatmaps and models them as a Gaussian Mixture Model (GMM) augmented with a spatially agnostic uniform distribution, requiring no architectural modifications or retraining. This continuous representation allows us to comprehensively explore the distinct contributions of density modeling to HPE. We show that the spatial expectation of our fitted GMM achieves precision parity with leading sub-pixel decoders without heuristic post-processing. Leveraging this continuous space, our test-time augmentation aggregates probability distributions rather than discrete activations, mitigating multi-scale heatmap poisoning. Furthermore, because our multimodal GMM parameterization isolates competing spatial hypotheses, it naturally interfaces with geometric regularizers such as Markov Random Fields. However, our analysis exposes the inherent vulnerabilities of 2D kinematic priors under perspective foreshortening and visual degradation. Crucially, this density formulation yields significant improvements in uncertainty quantification: our calibrated geometric covariance volume outperforms standard peak-intensity heuristics in OoD anomaly detection under severe occlusion. For in-distribution calibration, dynamically fusing our geometric metric via Adaptive Softmax reduces or matches baseline Area Under the Sparsification Error (AUSE) across benchmarks and degradation levels. Finally, we establish that this continuous parameterization is extensible to multi-view 3D architectures, bypassing the \mathcalO(N^3) voxel memory bottleneck.
[CV-430] Pristine Models
链接: https://arxiv.org/abs/2610.14091
作者: Anirudh Sundara Rajan,Krishna Kumar Singh,Richard Zhang,Yong Jae Lee
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Training on broad, unfiltered data enables high-quality image generation, but can also teach models concepts that should not be generated; restricting the training data ensures safety, but reduces quality because examples containing unsafe concepts also provide useful knowledge that is not itself unsafe. Existing approaches largely address this trade-off post hoc through concept unlearning, which can lead to catastrophic forgetting as the number of concepts to remove grows. We instead ask whether unsafe concept knowledge can be compartmentalized during pretraining so that access to it can be disabled at inference. We introduce Pristine Models, a novel text-to-image generation architecture with an attention mechanism that encourages the separation of safe and unsafe knowledge into independently parameterized branches. The branch-specific parameters are learned during training, while the unsafe branch can be detached at inference. On ImageNet, Pristine Models enable safer generation while retaining much of the utility of full-data pretraining, achieving higher image quality and better text alignment than training only on safe data or applying post-hoc concept removal. We further scale our approach to CC12M, demonstrating that it remains effective in a substantially larger and more diverse open-world pretraining setting.
[CV-431] SenseWeather: Multimodal Sensing for Robust Robotic Perception under Extreme Conditions
链接: https://arxiv.org/abs/2610.14082
作者: Fucai Ke,Sukai Huang,Junyi Hao,Kal Backman,Naim Rastgoo,Gholamreza Haffari,Hamid Rezatofighi
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Robotic perception must remain reliable when extreme environmental conditions severely degrade the primary RGB observation. Although robotic platforms commonly carry complementary sensors such as depth and infrared, much of the modern perception stack remains RGB-centric, creating a mismatch between multimodal onboard sensing and pretrained downstream models. We study whether complementary sensing can be used upstream to recover a task-compatible RGB observation that preserves existing perception capabilities without modifying or retraining the downstream stack. To this end, we introduce SenseWeather, a controlled multimodal benchmark for robotic perception under severe rain, snow, dust, fog, and darkness, with registered RGB, depth, and infrared observations, clean references, and grounded high-level perception tasks. We further develop Fusion-to-RGB, with two parameter-efficient sensor-fusion realizations based on token-space fusion for a diffusion transformer and residual latent conditioning for an inpainting UNet. Across frozen open-vocabulary grounding and vision-language models, RGB+D+IR fusion improves all paired downstream comparisons over the corresponding RGB-only restorers, including AP50 gains across all ten detector comparisons. Importantly, these perception gains do not consistently follow conventional image-fidelity metrics, and increasing weather severity does not induce monotonic downstream degradation. These results show that complementary sensing can preserve reusable robotic perception under extreme conditions, while highlighting task-level perception robustness as an objective distinct from visual restoration quality.
[CV-432] Safety-Oriented Trajectory Ranking from BEV Perception Outputs
链接: https://arxiv.org/abs/2610.14066
作者: Leon Pohl,Martin Frauendorfer,Corinna Schmitt,Mirko Maehlisch
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at the 2026 IEEE International Conference on Vehicular Electronics and Safety (ICVES 2026). 6 pages, 4 figures, 3 tables. Code: this https URL
Abstract:Bird’s-Eye-View (BEV) representations are increasingly used as compact inputs for autonomous driving perception and planning. However, lightweight and interpretable methods for evaluating candidate ego trajectories directly on BEV outputs remain useful for vehicular safety workflows, rapid prototyping, and human-in-the-loop validation. This paper presents a modular trajectory ranking pipeline that consumes BEVFusion object detections and map segmentations derived from nuScenes sensor data and ranks manually specified candidate trajectories according to safety-first criteria. The ranking prioritizes avoiding collisions and walkway violations, then smaller end distance to the target, and finally path efficiency. To study whether these rule-based rankings resemble human judgment, we created seven BEV traffic scenarios and compared the script output with preferences from a survey of 21 participants. Across 146 pairwise votes, 124 matched the trajectory ranked first by the evaluator, corresponding to an overall agreement of 84.9 %. The results indicate that simple BEV-centric safety heuristics can approximate human preferences in constrained trajectory choice tasks while remaining transparent enough for analysis, dataset curation, and autonomous driving studies.
[CV-433] RAIN: Robotic Task Generalization with Region-Aware Interaction Networks
链接: https://arxiv.org/abs/2610.14039
作者: Seunghun Lee,Kyuhwan Shim,Yeogyeong Song,Seung Hwan Kim,Sihaeng Lee,Byoung-Tak Zhang,Sunghoon Im,Gunhee Lee
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Existing vision-language-action (VLA) models benefit from large-scale vision-language pretraining, but tend to memorize training trajectories, limiting their ability to adapt learned interactions to new task and scene configurations. To address this problem, we propose \textbfRegion-Aware Interaction Networks (RAIN), which formulates a task as a sequence of goal-conditioned subtasks, each specified by an action type and a target-region mask. This formulation allows learned skills to be reused across different objects and scene layouts and composed into task sequences not seen during training. Its Target-adaptive Cross-View Encoder (TCE) conditions multi-view features on target masks and exchanges information across views, while PlanDiT predicts target-directed trajectories and actions. A Transition Head estimates subtask completion to enable automatic transitions. We also introduce a \textbfReference-based Retargeting Strategy (RRS), which uses recorded trajectories to synthesize smooth approaches to reference interactions with alternative target regions, without collecting additional data. Experiments on \textbfLIBERO-Analogy, our new benchmark for robotic task generalization, and real-world tabletop tasks demonstrate that RAIN generalizes more robustly than existing VLA baselines.
[CV-434] PLDiv: Measuring Dataset Diversity and Semantic Collapse via Persistence Landscapes
链接: https://arxiv.org/abs/2610.14031
作者: Yang Ba,Michelle V. Mancenido,Rong Pan
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (stat.ML)
备注:
Abstract:Modern data pipelines increasingly combine human-created and human-curated data with AI-generated data for model training and evaluation. As datasets grow, they often add semantically redundant samples rather than expanding semantic coverage, which deteriorates the quality of the whole dataset. Existing diversity metrics primarily summarize lexical variation, entropy, or pairwise similarity, but provide limited characterization of the shape of the underlying data distribution. In this paper, we propose PLDiv, a diversity metric derived from persistence landscapes that quantifies the structural organization of a dataset from its geometric representation. PLDiv summarizes how data neighborhoods evolve as nearby samples progressively connect, producing an interpretable scalar measure of structural diversity without requiring task-specific supervision. Whether measuring intrinsic diversity in text generations or image embeddings, or detecting semantic collapse in instruction datasets and reasoning trajectories, PLDiv performs competitively against evaluated baselines and aligns strongly with diversity labels during semantic collapse. Consequently, it effectively supports applications in synthetic data monitoring, dataset construction, data augmentation, and generative model evaluation.
[CV-435] Barycentric Anchored Weight Search for Robust Model Merging
链接: https://arxiv.org/abs/2610.14017
作者: Antoni Buades,Jose-Luis Lisani
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We propose Barycentric Anchored Search, a post-hoc weight-space procedure for improving fine-tuned vision models under distribution shift without additional gradient training or increased inference cost. Given a pre-trained initialization and two checkpoints fine-tuned on a source dataset, the method constructs barycentric candidate weights and performs a small derivative-free search along directions connecting each candidate to the pre-trained model. Candidate weights are selected using a small validation subset of one or more out-of-distribution (OOD) datasets; the selected model is then evaluated on the held-out portions of those datasets, on additional OOD benchmarks, and on the ImageNet validation set to quantify the robustness-accuracy trade-off. The method is inspired by simplex-based derivative-free optimization, including Nelder-Mead and Torczon multidirectional search, but is specialized to the geometry of neural network weight merging. Unlike Model Soup, which typically benefits from many fine-tuned checkpoints, and unlike Model Stock, which applies a fixed geometric merge, our approach uses only two fine-tuned checkpoints and iteratively adapts the merge using validation feedback from OOD data. Experiments with CLIP ViT-B/32 show that the proposed search substantially improves performance on held-out OOD data, including ImageNet-A, ImageNet-R, and ImageNet-Sketch, while keeping ImageNet accuracy competitive. The resulting weights also improve transfer to OOD datasets not used during candidate selection, outperforming the original fine-tuned checkpoints, naive model soups, and Model Stock in our evaluations. These results suggest that lightweight optimization in anchored weight space provides a practical alternative to fixed weight-averaging rules, geometric merging, or single-model selection.
[CV-436] SharedSAE: a unified sparse dictionary across u-net blocks and denoising time for diffusion models
链接: https://arxiv.org/abs/2610.13997
作者: Longhao Lei,Lan Yao
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Concept representations in diffusion models drift across network depth and denoising time, yet existing sparse autoencoders offer only per-block or per-timestep dictionaries, leaving no single feature dictionary for localization and intervention in the network. We present Shared Sparse Autoencoder (SharedSAE), which learns one sparse dictionary shared across four U-Net attention blocks and all 50 denoising steps of SDXL 1.0. A block-specific low-rank input adapter aligns the residual distribution of each block before the shared encoder, and a temporal conditioning module with a shared timestep branch and a block-timestep correction branch absorbs temporal non-stationarity. Despite spanning both axes, Shared SAE preserves a coherent dictionary structure. Built on the shared dictionary, localized feature suppression reduces the number of NudeNet-positive generations in the I2P detector-positive subset from 139 to 1. Tracing sparse features across the trajectory further reveals that concept representations are redistributed across U-Net blocks during generation, shifting from the earliest downsampling block toward intermediate and later blocks as denoising proceeds. SharedSAE thus provides a unified, interpretable, and intervention-ready representation space for diffusion models, enabling concept analysis and erasure at any covered block and denoising step.
[CV-437] Posterior Flow Matching
链接: https://arxiv.org/abs/2610.13994
作者: Chenrui Ma,Xi Xiao,Lin Zhao,Xinru Jiang,Tianyang Wang,Yanning Shen,Ferdinando Fioretto
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Flow matching learns how to transport noise toward data, but its mean prediction does not directly report the endpoint uncertainty remaining at an intermediate state. This paper introduces Posterior Flow Matching (PFM), a general generative modeling framework that learns conditional velocity means and componentwise variances with a Gaussian posterior model. The same model evaluation supplies both a transport prediction and endpoint uncertainty, making posterior diagnostics available throughout generation. These learned moments support a mean-driven ordinary differential equation (ODE) and a posterior-driven stochastic differential equation (SDE), whose local noise scales adapt to the predicted uncertainty. Across image, point-cloud, molecular, and action generation, PFM improves generation quality and provides endpoint-posterior readouts throughout generation.
[CV-438] How Vision Transformers Represent Position: A Mechanistic Comparison of Absolute and Rotary Positional Encodings
链接: https://arxiv.org/abs/2610.13992
作者: Mahmoud Mannes,Nikhil Maturi,Parva Mehta,Jiwon Jeong,Aravind Singaram Kannappan
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Positional embeddings encode the spatial position of each image patch in a Vision Transformer (ViT), allowing the model to distinguish different spatial arrangements of visual content. However, we know little about how the choice of embedding shapes a vision transformer’s internal representation. In this work, we show that Absolute Positional Embeddings (APE) and Rotary Positional Embeddings (RoPE) encode spatial position through different internal mechanisms: APE stores position in explicit features, while RoPE builds it inside attention and rebuilds it when it is withheld. Specifically, in APE models, sparse autoencoders recover causal row- and column-specific features, whereas interventions on RoPE reveal distinct rotary components that separately encode row and column information. Additionally, causal interventions selectively disabling RoPE’s row or column rotations show that decodable row and column information can be independently removed and, once rotations are restored, partly reconstructed, in part by axis-specialized attention heads. Both mechanisms are causally important for classification: permuting APE’s positional features across patches reduces top-1 accuracy by 24–31 points, and ablating the rotations of RoPE’s primary rebuilding head reduces it more than ablating those of any other head. Our findings show that the choice of positional encoding fundamentally changes how spatial information is organized and processed inside Vision Transformers. More broadly, our work demonstrates how mechanistic analysis can reveal how models construct, reconstruct, and maintain structured representations, providing a basis for targeted interventions in vision models.
[CV-439] Quasi-Binarized Autoencoders: An Architecture-Independent Information Bottleneck for Medical Image Anomaly Detection
链接: https://arxiv.org/abs/2610.09670
作者: Shouhei Hanaoka,Takahiro Nakao,Atsushi Takamatsu,Takeharu Yoshikawa,Osamu Abe
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Unsupervised anomaly detection, which learns only from normal images, is a central task in medical image analysis and remains an open problem. Reconstruction-based methods pass an image through an encoder-decoder network trained on normal data and detect anomalies from the residual between the image and its reconstruction. This works only if the information passed from the encoder to the decoder is limited; otherwise the network learns an identity mapping and reconstructs anomalies too. This limit is usually imposed through architectural choices, tuned per dataset, that cannot be stated in bits. We introduce the quasi-binarizing (QB) layer, which squashes each latent element into [0, 1] and adds Laplace noise of scale 1/epsilon. Each element is then epsilon-locally differentially private, and the mutual information between an image and its reconstruction is bounded by a quantity that depends only on epsilon and the number of QB elements, whatever the encoder and decoder. Placing a QB layer on every encoder-decoder path, including all skip connections, we build QBAE, a seven-level attention U-Net with 32,768 QB elements. On the seven datasets of the MedIAnomaly benchmark, QBAE with one architecture and one configuration reaches a mean image-level AUROC of 0.828, the highest among methods that do not adapt to each dataset, and the best reported results on BraTS2021 (AUROC 0.911, pixel-level AP 0.838). The noise is kept at test time, so that every reconstruction satisfies the bound. Without input corruption, the bottleneck alone prevents identity collapse (mean AUROC 0.805 vs. 0.590). Code is available at this https URL.
[CV-440] ArborSplat: Online Semantic Gaussian Splatting SLAM for Orchards
链接: https://arxiv.org/abs/2609.26315
作者: Alessandro Masini,Matteo Frosi,Mirko Usuelli,Matteo Matteucci
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 4 figures, 4 tables
Abstract:Orchard robots need maps that preserve small but semantically important structures such as trunks, trellises, and fruit. 3D Gaussian Splatting (3DGS) SLAM achieves high photometric fidelity. However, its optimization remains appearance-driven, and transferring image semantics to 3D points is unreliable for thin structures, whose pixels may receive depth from background surfaces. We present ArborSplat, an online semantic 3DGS SLAM system that tracks with LiDAR odometry and optimizes semantics directly on the Gaussian map, constrained by class-specific height bands above a ground plane fitted to each keyframe’s stereo point cloud, and fuses multi-view evidence into a semantic point cloud online while rejecting labels inconsistent with the local ground surface or with monocular depth. Class-constrained refinement reserves Gaussian capacity for underrepresented structures and, under reduced budgets, increases training-view accuracy on tree classes. We evaluate the approach on apple and pear orchards during dormancy, flowering, and harvesting. On full routes, it keeps ATE below 0.5 m on all 12 traversals. On shared 301-frame segments, it exceeds SGS-SLAM and GS3LAM by 0.23 to 0.50 training-view and 0.15 to 0.36 held-out mIoU while running 1.7 to 7.5 times faster, whereas SemGauss-SLAM runs out of GPU memory on all six.
[CV-441] Selective Knowledge Control for Continual GUI Agent Learning over Application Streams
链接: https://arxiv.org/abs/2609.06530
作者: Zirui Shang,Xin Shu,Yang Liu,Zhi Gao,Xinxiao Wu,Lifeng Fan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Continual learning is a crucial capability for Graphical User Interface (GUI) agents to adapt to evolving applications while retaining knowledge acquired from previous applications. Such application streams pose a challenging knowledge modeling problem: new applications often share underlying knowledge with past ones, yet also introduce specific knowledge that must not interfere with historical knowledge. In this paper, we propose activation-conditioned selective knowledge control, a lightweight method that achieves selective knowledge retention via neuron-level gradient manipulation. Our method maintains a compact historical knowledge state to protect highly activated MLP neurons that preserve previous knowledge. When a new application arrives, it performs real-time gradient surgery conditioned on forward activation. Concretely, the protected neurons are categorized into two types: unactivated neurons holding specific knowledge, whose gradients are truncated to prevent interference; and activated neurons holding shared knowledge, whose gradients are orthogonally projected to preserve stability while enabling adaptation. After each application stage, newly identified critical neurons are merged into the historical state for future learning. Empirical evaluations on multi-app sequential benchmark demonstrate that our method effectively mitigates catastrophic forgetting on prior applications while sustaining robust adaptation to new ones.
[CV-442] AERR-Nav: Adaptive Exploration-Recovery-Reminiscing Strategy for Zero-Shot Object Navigation
链接: https://arxiv.org/abs/2603.17712
作者: Jingzhi Huang,Junkai Huang,Haoyang Yang,Haoang Li,Yi Wang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Zero-Shot Object Navigation (ZSON) in unknown multi-floor environments presents a significant challenge. Recent methods, mostly based on semantic value greedy waypoint selection, spatial topology-enhanced memory, and Multimodal Large Language Model (MLLM) as a decision-making framework, have led to improvements. However, these architectures struggle to balance exploration and exploitation for ZSON when encountering unseen environments, especially in multi-floor settings, such as robots getting stuck at narrow intersections, endlessly wandering, or failing to find stair entrances. To overcome these challenges, we propose AERR-Nav, a Zero-Shot Object Navigation framework that dynamically adjusts its state based on the robot’s environment. Specifically, AERR-Nav has the following two key advantages: (1) An Adaptive Exploration-Recovery-Reminiscing Strategy, enables robots to dynamically transition between three states, facilitating specialized responses to diverse navigation scenarios. (2) An Adaptive Exploration State featuring Fast and Slow-Thinking modes helps robots better balance exploration, exploitation, and higher-level reasoning based on evolving environmental information. Extensive experiments on the HM3D and MP3D benchmarks demonstrate that our AERR-Nav achieves state-of-the-art performance among zero-shot methods. Comprehensive ablation studies further validate the efficacy of our proposed strategy and modules.
[CV-443] Deficit-Driven Stratified Acquisition for Building STRATA-S12: An Open SAR-Optical Corpus for Self-Supervised Learning
链接: https://arxiv.org/abs/2610.17680
作者: Christian Sanchez-Leon,Dongming Chen,Dongqi Wang
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 55 pages. Dataset on Hugging Face: this https URL - archived on Zenodo: STRATA-S12 corpus this https URL ; reduced subset (D_lite) this https URL
Abstract:Building a co-registered SAR-optical corpus for self-supervised Earth Observation is above all an acquisition problem, yet the main multi-modal datasets are collected opportunistically, with no per-stratum quota control and no mechanism against the geographic dropout that non-uniform cloud cover leaves behind. We present Deficit-Driven Stratified Acquisition (DDSA), an algorithm that treats building a co-registered Sentinel-1/Sentinel-2 corpus as a constrained maximum-coverage problem over a hemisphere-corrected (class x season x region) stratification, enforces strict temporal pairing and a multi-stage quality gate, and closes the loop during acquisition with a human-in-the-loop deficit-repair stage. Measured on its own product, the algorithm yields three results a single-pass acquisition cannot report. Deficit-driven ordering makes a partial run a covering one: it holds every one of the 192 strata populated from the first few percent of the budget, where an abundance-first order leaves most strata empty until the end. Whether a failed attempt can be recovered depends on the temporal persistence of its cause, not the kind of defect: geometric failures such as nodata are among the most repairable (eta = 41.4%) while cloud is among the least (eta = 18.8%), because a cloudy location is cloudy on most dates of its season. The repair loop supplies 39.5% of the released pairs at no measurable quality cost, with 89.8% surviving manual review. The quality guarantee has a price: 2,928,370 acquisition attempts produced 740,746 released pairs. We release the corpus, STRATA-S12, with STAC/PDS4-aligned FAIR sidecars, and a pre-registered linear-probe check confirms it is fit for self-supervised pre-training. A metadata-level coverage comparison against SSL4EO-S12 and SEN12MS situates the corpus over a shared frame; systematic cross-corpus and cross-method comparison is left to follow-up work.
[CV-444] Background2Background: a self-supervised background removal framework for high dimensional microscopy
链接: https://arxiv.org/abs/2610.17615
作者: Guangrui Ding,Haonan Lin,Rong Tang,Bethany Weinberg,Jianpeng Ao,Qing Xia,Guorong Hu,Dingcheng Sun,Jiaze Yin,Lei Tian,Ji-Xin Cheng
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Optics (physics.optics)
备注:
Abstract:High-dimensional microscopy captures features across space, time, and spectrum, but complex backgrounds obscure signals. We present Background2Background (B2B), a self-supervised background removal framework leveraging the backgrounds larger correlation relative to the signal. Local pixel shuffling disrupts signal correlations while preserving background structure, transforming signals into effective noise for Noise2Noise-based background removal. Across fluorescence, stimulated Raman scattering and mid-infrared photothermal imaging, B2B achieves over 40-fold background suppression and reveals hidden signatures.
[CV-445] Uncertainty-Aware Cortical Surface Reconstruction and Spherical Registration for Heterogeneous Brain MRI
链接: https://arxiv.org/abs/2610.16595
作者: Jesper Duemose Nielsen,Karthik Gopinath,Andrew Hoopes,Adrian Dalca,Axel Thielscher,Juan Eugenio Iglesias,Oula Puonti
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Surface-based cortical analysis is useful in many applications as it is a natural way to model the highly folded structure of the cerebral cortex, allowing quantities such as curvature and thickness to be easily calculated. Existing deep learning (DL) methods have focused on 1x1x1 mm3 T1-weighted (T1w) scans, thus replicating classical pipelines. Here we present a neural network that is trained exclusively on synthetic scans with random contrast and resolution. Consequently, it can be applied to almost any kind of magnetic resonance imaging (MRI) scan. In addition, our model provides vertex-wise uncertainty estimates of the predicted surface as well as a spherical registration to fsaverage space. Our model performs on par with other state-of-the-art DL methods on 1x1x1 mm3 T1w scans and outperforms the only other method capable of estimating cortical surfaces from low resolution, non-T1w scans. The predicted uncertainty, moreover, correlates with overall scan quality on low resolution scans and highlights areas of poor surface reconstruction on 1x1x1 mm3 T1w scans, suggesting that it can be used for quality control. Finally, the predicted spherical registrations prove comparable to other DL and classical registration methods. These qualities, when combined with the fast inference time, will provide the ability to analyze large, existing databases of clinical MRI scans, with the potential for finding disease-specific patterns that would otherwise be infeasible in a research setting.
[CV-446] Variational Filtering for Bayesian Last Layers with an Independent Binary Probit Surrogate
链接: https://arxiv.org/abs/2610.16578
作者: Dimitrije Markovic,Conor Heins,Alexander Tschantz,Christopher L. Buckley
类目: Machine Learning (stat.ML); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Modern deep classifiers trained by gradient descent need many passes over the data to become accurate, and even then are not always well calibrated; post-hoc temperature scaling corrects calibration after training but cannot influence training-time accuracy or convergence. We propose a variational filtering algorithm for the last layer of deep classifiers, whose parameters are treated as the latent state of a linear-Gaussian state-space model. Each mini-batch is assimilated by an exact prediction step followed by a closed-form variational update, made conjugate by an independent binary probit (IBProbit) surrogate for the categorical likelihood. The filter’s process noise stabilises the updates under feature drift and keeps the predictive distribution from collapsing to a point estimate, allowing the same procedure to be used for linear probing on a frozen backbone and for full-network fine-tuning. We evaluate on DINOv3 backbones against two baselines tuned under an identical protocol: AdamW with cross-entropy and post-hoc temperature scaling, and Variational Bayesian Last Layers (VBLL). As a linear probe on frozen features, IBProbit attains the best accuracy and negative log-likelihood on all five downstream tasks at the largest backbone, and on two of the five at the smallest. In out-of-distribution detection with CIFAR probes on the largest backbone, IBProbit reduces the false-positive rate at 95% true-positive rate by a factor of 1.4 to 3.2 relative to the better of the two baselines, on all three dataset pairs. When the whole network is fine-tuned on ImageNet-1k, IBProbit reaches the final performance of the cross-entropy baseline in half the wall-clock time.
[CV-447] Hybrid Context Forests and Bilateral Viterbi Matching for 3D Rib Centerline Tracing in CT
链接: https://arxiv.org/abs/2610.16047
作者: Quan Wang
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Tissues and Organs (q-bio.TO)
备注:
Abstract:Automatic extraction and anatomical numbering (R01-R12, L01-L12) of all 24 rib centerlines in thoracic CT is essential for unfolded rib visualization, fracture detection, and surgical navigation, but remains challenging due to adjacent-bone confluence, calcified costal cartilage bridges, kyphotic pose variation, and partial visibility of terminal ribs. We present ribline, an open-source three-stage framework combining: (1) an automatically detected lung-apex and vertebral-midline coordinate frame with a kyphosis-aware azimuthal angle for point ordering along each rib; (2) a hybrid voxel detector fusing an anatomically anchored Semantic Context Forest (SCF) with a lightweight 3D U-Net; and (3) a global stage that decomposes the probability skeleton into candidate curves, assigns rib indices via a bilateral Viterbi dynamic program coupled across both hemithoraces, and refines curves via template recovery and normal-plane centering (RibSnake). On the Siemens Corporate Research 2 mm thoracic CT benchmark (47 training and 12 evaluation scans), our pipeline achieves a labeled Chamfer distance of 2.05 mm, labeled clDice of 0.886 (3 mm) and 0.945 (4 mm), 97.2% labeling accuracy (278/286 ribs), and zero false-positive curves on the evaluation split, and 3.64 mm (median 2.39 mm) with 91.8% labeling accuracy across all 59 annotated scans (1,402 ribs) in 5-fold cross-validation. Code: this https URL Pretrained Models: this https URL Subjects: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Tissues and Organs (q-bio.TO) Cite as: arXiv:2610.16047 [eess.IV] (or arXiv:2610.16047v1 [eess.IV] for this version) https://doi.org/10.48550/arXiv.2610.16047 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-448] Automated carotid artery atherosclerotic inflammation segmentation in PET using MSR-Net
链接: https://arxiv.org/abs/2610.15245
作者: Ran Li,Abhinav K. Jha,Jie Zheng,Richard Laforest,Yongjian Liu,Mohamed Zayed,Tyler J. Fraum,Barry A. Siegel,Malak Itani,Joyce C. Mhlanga,Pamela K. Woodard
类目: Medical Physics (physics.med-ph); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Objective. To develop an automated method for segmentation and uptake quantification of carotid atherosclerotic inflammation on 64Cu-CANF-comb PET while accounting for partial-volume and tissue-fraction effects. Approach. Synthetic carotid PET images with known fractional-volume ground truth were generated by combining MR-derived high-resolution activity maps with GATE Monte Carlo simulation incorporating major PET physical processes and matched to the Biograph mMR acquisition and reconstruction. A two-stage network first localized the carotid vessel wall from T1-weighted MR and then estimated voxel-wise inflammation fraction from masked PET. Both stages used a multiscale residual (MSR) convolutional backbone. Performance was compared with U-Net, Transformer U-Net, DUCK-Net, and iterative Yang partial-volume correction. Main results. In a two-alternative forced-choice study, five nuclear-medicine physicians identified real images with 54% accuracy (54/100; Wilson 95% CI 0.44-0.63; exact binomial p = 0.48). On 184 simulated test slices from five held-out patients, the proposed method achieved a weighted DSC of 0.79, compared with 0.73 for U-Net, 0.69 for Transformer U-Net, and 0.75 for DUCK-Net, with corresponding HD values of 1.85, 2.29, 2.32, and 2.23 voxels. Patient-cluster bootstrap comparisons favored the proposed method for DSC and HD against all three baselines (Holm-corrected p = 0.031). Overall ENRMSE was 0.24 versus 0.59, 0.68, and 0.59 for the three DL baselines and 0.57 for iterative Yang. In seven patients, corrected SUV showed a positive association with MR-derived lipid-rich/necrotic core with hemorrhage. Significance. The proposed framework improved carotid PET segmentation and uptake quantification under simulated conditions. The patient analysis is preliminary and requires external validation with biological reference standards before clinical deployment.
[CV-449] Generative Quantum Machine Learning Methods for Cancer Detection using Augmented Datasets
链接: https://arxiv.org/abs/2610.15110
作者: Sophia Knutson,Nandika Goyal,Gennaro De Luca,Glen Uehara,Tanay Kamlesh Patel,Andreas Spanias
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Emerging Technologies (cs.ET); Quantum Physics (quant-ph)
备注:
Abstract:Cancer detection using AI methods and image databases faces several challenges, including limited and imbalanced image datasets, long algorithm training times, and high-complexity feature generation. In particular, many cancer image datasets contain very few malignant examples compared to benign ones, limiting the amount of minority-class data available for training. Although classical AI and especially deep learning methods have had some success in providing cancer diagnostics, Quantum Machine Learning (QML) methods may offer advantages through high-dimensional feature representations, generative data augmentation, and the exploitation of feature correlations through quantum properties such as entanglement. In this paper, we propose using a generative QML method on the compressed breast ultrasound image dataset (BUSI) to generate synthetic images and hence augmented datasets for classifier training. In particular, we train a convolutional neural network with real, synthetic, and a combination of real and synthetic images. We test the algorithm on real images and provide comparative results for all three training cases. Our results show that the classifier trained entirely on synthetic images achieved an AUC-ROC of 0.802 (80.2%), compared with 0.760 (76.0%) for the real-data baseline.
[CV-450] Beyond Gate Weights: Decision-Level Interpretation of Radiomics-Image Feature Fusion for RCC Classification
链接: https://arxiv.org/abs/2610.14152
作者: Yuan Liang,Sourav Bhattacharjee,Abraham Campbell
类目: Image and Video Processing (eess.IV); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: 12 pages, 1 figure. Accepted at the 2026 Irish Conference on Artificial Intelligence and Cognitive Science (AICS 2026). Proceedings forthcoming
Abstract:Gate coefficients describe how a fusion layer weights intermediate features, but they do not necessarily reflect the magnitude of branch contributions at the classifier output. We investigate this distinction in a fixed 3D MedVAE–radiomics model for clear cell versus non-clear cell renal cell carcinoma classification from contrast-enhanced CT. Because the model uses a linear classification head, the binary decision logit can be decomposed exactly into a gated image term, a direct radiomics term, and a bias. We define effective radiomics dominance as the direct radiomics term’s share of the summed absolute non-bias branch contributions and compare it with structural radiomics weighting. Internally, mean structural radiomics weighting was 0.823, whereas mean effective radiomics dominance was 0.148, with a modest case-wise association (\rho = 0.290). The gated image term retained substantially stronger discriminative ranking information than the direct radiomics term. The same structural-to-effective discrepancy was observed in exploratory external evaluation. Branch-targeted permutation, integrated-gradient, and Shapley-style analyses further identified radiomics descriptors influencing the direct radiomics term. Because radiomics also generates the gate, the gated image term remains radiomics-dependent, so additive branch contribution should not be interpreted as total input influence. These findings show that structural gate weighting and decision-level branch contribution represent distinct aspects of gated feature fusion and should be interpreted separately.
[CV-451] Beyond Discrimination: Evaluating Probability Calibration in Radiomics-Image Fusion for Renal Cell Carcinoma Classification
链接: https://arxiv.org/abs/2610.14151
作者: Yuan Liang,Morgan May,Abraham Campbell,Sourav Bhattacharjee,Simon Caton,Pierpaolo Dondio
类目: Image and Video Processing (eess.IV); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Quantitative Methods (q-bio.QM)
备注: 12 pages. 1 figure. Accepted at the 2026 Irish Conference on Artificial Intelligence and Cognitive Science (AICS 2026). Proceedings forthcoming
Abstract:Reliable probability estimates are important for the clinical interpretation of machine-learning classifiers, yet radiomics–image fusion models for renal cell carcinoma (RCC) are commonly evaluated primarily by discrimination. Using 396 patients from the KiTS23 dataset, we evaluated whether adding radiomics to four image-model families improved probabilistic performance in addition to discrimination. Seventeen individual models and four probability-average ensembles were assessed using discrimination, probability calibration, and proper scoring measures. Across 12 matched image-only versus fusion comparisons, AUC and AP improved in all 12, while Brier score and negative log likelihood (NLL) improved in 11 of 12 and expected calibration error (ECE) in eight of 12. The 3D MedVAE gated-fusion model achieved the strongest overall individual-model profile. Probability averaging across concatenation, cross-attention, and gated fusion produced favourable point estimates across all five primary metrics relative to image-only models in all four families, although paired-bootstrap support was stronger for Brier score and NLL than for ECE. These findings show that radiomics–image fusion can improve probabilistic prediction quality beyond discrimination, while calibration behaviour varies across image architectures and fusion strategies. We show that our proposed probability-average ensemble models improve discrimination and calibration simultaneously across all architectures in the study.
[CV-452] C2LUE: Clinically Aligned Patient-Level Ultrasound Diagnosis via Evidence-Aware Multi-Instance Learning
链接: https://arxiv.org/abs/2610.14143
作者: Qin-Yi Zhang,Linying Cao,De-Xing Huang,Haichuan Lin,Yi Yang,Manzhi Wang,Xiao-Hu Zhou,Chen Chen,Shuangyi Wang,Zeng-Guang Hou
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Patient-level ultrasound diagnosis demands not only high predictive accuracy but also clinically aligned mitigation of asymmetric diagnostic risks. While Multi-Instance Learning (MIL) effectively models routine multi-view exams, standard aggregators often dilute heterogeneous diagnostic cues and optimize solely for global accuracy. This frequently yields undesirable sensitivity-specificity trade-offs. To bridge this gap, we propose C ^2 LUE, a novel MIL framework coupling evidence aware aggregation with risk-aligned regularization. To preserve evidence heterogeneity and prevent unimodal attention collapse, C ^2 LUE employs a multi-query cross-attention mechanism built upon a LoRA-adapted DINOv3 backbone. Multiple latent queries explicitly extract and disentangle diverse morphological patterns from the ultrasound sequence. Furthermore, to improve clinical risk alignment, we introduce a dynamic trade-off regularization objective that penalizes sensitivity-specificity imbalance during training. Rather than leaving the clinical trade-off entirely to post-hoc threshold adjustment, this objective encourages the learned score distribution toward a more balanced operating region. Extensive evaluations on patient-level, biopsy-proven thyroid and breast ultrasound datasets demonstrate that C ^2 LUE consistently outperforms prior alternatives. By uniting strong discrimination with a balanced error profile, C ^2 LUE provides a promising solution for real-world ultrasound screening and diagnosis.
[CV-453] Frozen Site-Probe Accuracy Is Not a Standalone Harmonization Score: An Audit of Evaluation Practice for Multi-Site MRI
链接: https://arxiv.org/abs/2610.14140
作者: Muhammad Muqsit Islam,Gloria García Cuenco,José M. Martínez,Pierrick Coupé
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 4 pages, 2 figures, 1 table. Submitted to IEEE ISBI 2027. Code and results: this https URL . This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible
Abstract:Image-level MRI harmonization is often evaluated with a frozen site probe, a site classifier trained on raw scans and applied unchanged to harmonized ones: a drop in its accuracy is read as removal of site effects. We audit this practice on ABIDE I (1,112 T1-weighted scans, 17 sites) with ten output sets from seven harmonization methods and controls for chance, head shape, intensity, preprocessing and target alignment. We find that this verdict depends on how the probe is trained, even among probes that classify raw scans well: across random seeds, its balanced accuracy on the same CycleGAN outputs ranges from 0.13 to 0.59 (chance: 0.06). Low frozen-probe accuracy can also arise without harmonization: a Gaussian blur, which increases the intensity distance to the target site, lowers this accuracy to chance. Moreover, where the frozen probe fails, probes trained on the harmonized scans still recover the site, even from intensity histograms alone: on HACA3 outputs, the frozen probe scores 0.25, a probe trained on those outputs 0.75, and one using only their intensity histograms 0.71. Low frozen-probe accuracy therefore does not show that site information has been removed. We recommend reporting several probes trained to convergence, probes trained on harmonized scans, and target alignment.
[CV-454] Exploring the Latent Structure of AMD Severity Using Retinal Foundation Models
链接: https://arxiv.org/abs/2610.14138
作者: Niccolò Fineschi,Jovana Panic,Atsushi Watanabe,Hideaki Hayashi,Kazuya Nishimura,Yuki Maeno,Samanta Rosati,Gabriella Balestra,Kohji Nishida,Hajime Nagahara
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 7 pages, 5 figures. Accepted for publication at IEEE EMBS BHI 2026
Abstract:Retinal foundation models have demonstrated strong performance across a wide range of ophthalmic tasks, yet relatively little is known about how disease-related information is represented within their learned embeddings and whether this distribution influences downstream prediction performance. In this work, we use Age-related Macular Degeneration (AMD) as a case study to investigate how disease severity is represented within the embedding space of RET-CLIP, a retinal vision-language foundation model, using bilateral fundus images from the AREDS dataset, benchmarked against non-retinal encoders. We first characterize the distribution of AMD severity within the learned representation space and then evaluate how this structure influences classification performance across three clinically relevant task formulations of increasing granularity. Finally, we analyze the most influential latent dimensions combining feature attribution, statistical phenotype association, and qualitative image analyses. Our results show that RET-CLIP represents AMD severity as a continuous spectrum rather than a set of discrete categories, with healthy and advanced disease stages exhibiting clearer separation than intermediate ones. This organization was reflected in downstream performance, which decreased from a balanced accuracy of 0.844 and 0.814 for binary AMD and Late-AMD detection respectively to 0.446 for six-class grading. Furthermore, the most influential latent dimensions were statistically associated to distinct retinal phenotypes, including drusen accumulation, pigmentary alterations, and advanced retinal lesions. These findings suggest that the structure of retinal foundation-model representations influences the difficulty of clinically relevant prediction tasks and demonstrate the potential of representation-level analysis for understanding disease organization and improving interpretability.
[CV-455] mporal Diffusion Bridge Model for Vessel Extraction Using Physics-informed Synthetic Data
链接: https://arxiv.org/abs/2610.14105
作者: Jinkui Hao,Donald R. Cantrell,Ramez Abdalla,Sameer A. Ansari,Bo Zhou
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 13 pages, 6 figures. Under review at Medical Image Analysis
Abstract:Accurate artery extraction from X-ray coronary angiography (XCA) is essential for coronary artery disease diagnosis and intervention. However, low vessel visibility, respiratory and cardiac motion, and limited annotations challenge existing methods. Moreover, most deep learning approaches produce binary vessel masks without preserving clinically relevant intra-arterial contrast information. We propose a Temporal Diffusion Bridge Model (TDBM) that formulates vessel extraction as image translation and learns from physics-informed synthetic data without manual vessel annotations. A physics-based XCA synthesis engine generates diverse, perfectly aligned pairs of contrast-enhanced and contrast-free images. Using these pairs, TDBM learns cross-domain mappings to recover high-fidelity vessel information. Temporal conditioning incorporates adjacent frames to improve consistency, while a vessel-aware contrastive loss exploits background-invariant synthetic pairs to enhance vascular discrimination. We evaluate TDBM on two XCA vessel segmentation datasets and conduct a reader study using two real-patient XCA sequence datasets. Zero-shot segmentation is further assessed on retinal imaging and fluorescence microscopy datasets. Results demonstrate that TDBM consistently outperforms existing methods and generalizes effectively across imaging modalities. Code and trained weights will be available at: this https URL
[CV-456] RePrior: Measurement-Reconciled Prior Correction for Sparse-View CBCT Reconstruction
链接: https://arxiv.org/abs/2610.14102
作者: Mengrui Yin,Tianxingjian Ding,Wufan Zhao,Ziyang Yan,Yu Tian
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Sparse-view cone-beam computed tomography (CBCT) is severely under-constrained by limited X-ray projections. Learned reconstruction can provide complementary anatomical information, but using it as a prior in patient-specific Gaussian reconstruction introduces two challenges. First, the discrepancy between the learned prior and current reconstruction is spatially non-uniform: some local differences contain useful missing 3D information, whereas others reflect prior bias. Second, even a volumetrically beneficial correction can conflict with the observed X-ray measurements. Since X-ray measurements are line integrals, spatially separated corrections may affect overlapping rays, with their signed projection responses reinforcing or partially canceling in the measurement residual, so their feasibility cannot always be assessed independently. In this paper, we introduce RePrior, a multi-stage progressive prior-correction framework that addresses correction actionability, measurement feasibility, and representation expressivity. Selective Correction Proposal (SCP) first identifies actionable signed local corrections from prior–reconstruction discrepancies. Measurement-Constrained Reconciliation (MCR) then jointly enforces their consistency with patient-specific X-ray measurements, while Hybrid Residual Enhancement (HRE) recovers additional correction capacity through complementary voxel- and Gaussian-domain degrees of freedom around the reconciled anchor. RePrior requires neither test-time ground truth nor case-specific network fine-tuning. RePrior gains approximately 2.6~dB in 3D PSNR over R ^2 -Gaussian across LUNA16, PANORAMA, and PENGWIN. Beyond prior-assisted reconstruction, it delivers additional gains in volumetric quality and observed-view projection fidelity, with consistent gains in held-out-view fidelity in the backbone-matched evaluation.
人工智能
[AI-0] What Matters in Quality-Aware VLA Pretraining: A Benchmark and Empirical Study
链接: https://arxiv.org/abs/2610.18869
作者: Mingxuan Yan,Litian Gong,Dihong Huang,Zhixu Li,Tianyu Zhang,Wenqian Zhang,Hefeifei Jiang,Peihao Li,Jian Zhang,Mengfei Zhao,Yikai Tang,Hai Zhai,Zehao Wang,Ruijian Liang,Yanjia Huang,Lin Shao,Changliu Liu,Kaiyu Hang,Zhiwen Fan,Masayoshi Tomizuka,Jianfei Yang,Jiachen Li
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Project Website: this https URL
Abstract:How should demonstration quality be used during vision-language-action (VLA) pretraining to improve performance after target-task fine-tuning? We compare weighting, filtering, and conditioning across reward definitions and label granularities using 5,000 naturally suboptimal human demonstrations and 24 target tasks. Within each VLA, we match demonstrations and training budgets. Additional comparisons examine pretraining data size and the roles of quality tags during fine-tuning and inference. Across three VLAs, we find that approach-aware episode conditioning improves mean success over plain behavioral cloning; conditioning and filtering favor episode and chunk labels, respectively, on average, with model-dependent reversals; and constant-tag fine-tuning improves mean success after conditioned pretraining with either tagged or untagged inference. We introduce AXIS-Bench to support these controlled comparisons. Built on crowdsourced simulation demonstrations, it provides recorded object and contact states for recomputing quality labels on shared trajectories. Replayable trajectories support workspace-aligned rendering and user-defined real-world target tasks. In our real-world experiment, simulation pretraining raises success from 9.6% to 31.7%. Project page: this https URL.
[AI-1] AesCode: Aesthetic Code Generation with Decoupled Cross-Modal Rewards
链接: https://arxiv.org/abs/2610.18860
作者: Xin Gao,Yifan Yang,Xin Li,Pan Wang,Yan Li,Muzhao Tian,Mingxi Cheng,Bei Liu,Kai Qiu,Qi Dai,Dong Chen,Yue Dong,Leandros Tassiulas,Rex Ying,Ji Li,Chong Luo
类目: Artificial Intelligence (cs.AI)
备注: 44 pages,18 figures
Abstract:Generating information-rich visual artifacts such as slides, posters, and dashboards as code is appealing because the outputs are structured, editable, and verifiable. Yet code models cannot see how their choices of layout and color come together on the canvas, often limiting spatial planning and aesthetic quality. Image generators excel at visual design but can misrender requested content, making their outputs unreliable targets for direct reproduction. We therefore introduce AesCode, which generates code grounded in the content prompt while using an image generated from the same prompt as an aesthetic guidance. To do so, AesCode represents the target as a design graph that describes the whole canvas and makes each property attributable. From this graph, we derive complementary cross-modal rewards: deterministic verifiers score properties parsable from code and rendering, while a VLM judge assesses the non-parsable ones with a sample-specific Visual Graph Rubric (VGR) tied to graph elements and relations. We optimize these decoupled rewards through reinforcement learning with per-channel normalization to jointly improve spatial planning, aesthetic quality, and verifiable correctness. To evaluate this setting, we build a benchmark that scores verifiable correctness and visual quality separately, with visual scores highly consistent with human judgments. We find that reference images substantially improve the visual quality of code generated by existing models, and ablations show that AesCode benefits from references both as model input and as a source of visual rewards. Relative to same-scale Qwen3-VL baselines given the same references, AesCode-8B and AesCode-32B improve visual quality by 31.1 and 22.4 percentage points, reaching scores comparable to those of closed-source frontier models such as GPT-5.5 and Claude Opus 4.8.
[AI-2] wo Clocks: Why Larger Rollout Batches Learn Slower but Fail on Schedule in Reinforcement Learning with Verifiable Rewards
链接: https://arxiv.org/abs/2610.18853
作者: Yuante Li,Junwei Quan,Kai Qiu,Kate Zhang,Yicheng Tao
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Reinforcement learning with verifiable rewards (RLVR) relies on costly model-generated responses, motivating larger rollout batches and square-root learning-rate scaling. Reported to approximately preserve progress per sample with one optimizer step per rollout, this rule is less understood in practical multi-step training. We develop a two-clock framework that separates the samples needed to make progress from those consumed before collapse. Under a local quadratic model, the square-root rule preserves first-order drift per sample but not progress: larger batches take fewer, longer updates and lose more to curvature, yielding a learning-rate-dependent critical batch beyond which sample efficiency degrades. A drift-threshold hypothesis predicts a complementary failure clock whose sample count remains approximately invariant along the square-root rule. Controlled GRPO experiments on Qwen2.5-Math-1.5B across a 16\times batch range support both predictions. Ratio clipping delays failure, especially for larger batches, but the progress gap persists without it. Widening the clip trades stability for a transient speed gain. The progress gap reappears on another dataset, a general-purpose checkpoint, and a 7B model. Finally, we use the two-clock framework to derive a simple multi-step RLVR schedule: start below the critical batch and switch to a larger batch before failure. Our code is available at: this https URL.
[AI-3] MFG: Multi-turn Feedback Guided Molecular Optimization
链接: https://arxiv.org/abs/2610.18851
作者: Trieu Nguyen,Frazier N. Baker,Ruoxi Gao,Daniel Adu-Ampratwum,Xia Ning
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Quantitative Methods (q-bio.QM)
备注:
Abstract:Multi-objective molecular optimization is a critical process in drug discovery that aims to modify a given molecule to simultaneously improve multiple properties while maintaining a similar overall structure. This process often requires iterative modifications guided by computational or experimental feedback. Recent large language model (LLM)-based methods have begun to explore this iterative approach; however, these methods are not explicitly trained to leverage such feedback, while existing approaches primarily focus on single-turn molecular optimization. To address this gap, we introduce Multi-turn Feedback Guided (MFG) molecular optimization, a reinforcement learning framework that trains an LLM to progressively optimize molecules using feedback from previous turns. MFG introduces a feedback vector capturing both property improvements and source molecule similarity, alongside a feedback-aligned reward function that promotes progressive improvement across turns. We apply MFG over \textGeLLM^3\textO , a state-of-the-art LLM model for multi-objective molecular optimization, evaluate \textGeLLM^3\textO-MFG on ten multi-objective molecular optimization tasks. The results demonstrate that \textGeLLM^3\textO-MFG achieves the highest average similarity-constrained success rate ( 81.50% ), outperforming the base model \textGeLLM^3\textO ( 70.81% ).
[AI-4] LIBERO-RIG: A Hardware-Setup Robustness Benchmark for Vision-Language-Action Models
链接: https://arxiv.org/abs/2610.18843
作者: Andrea Yaoyun Cui,Hanyu Wang,Xinyi Wang,Bangzheng Li,Lin Zhang,Zhenggang Tang,Shyamal Buch,Dejia Xu,Matthew Tancik
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 30 pages (4 pages main text plus appendix), 15 figures, 18 tables
Abstract:Robot policies are expected to generalize across deployment environments, including variations in the robot itself: cameras are bumped and remounted, calibrations drift, controllers run at different rates, and end-effector pads wear down. We present LIBERO-RIG, a benchmark that perturbs only the robot’s hardware and setup while holding every LIBERO task fixed. The benchmark spans eleven perturbation axes, including camera mounting and optics, camera-to-arm calibration, actuation gain, control rate, proprioceptive noise, base pose, table height, and gripper pads. Each axis has five severity levels, yielding 210 conditions across the four LIBERO suites. We evaluate nine recent vision-language-action models using 500 episodes per condition and substantiate each finding with rigorous statistical tests. Pooled success falls from 0.932 to 0.767 at the middle severity and 0.509 at the highest severity. Performance degradation varies substantially across perturbation axes and action interfaces, and training on visual perturbations does not mitigate faults that leave camera images unchanged. We release the benchmark together with manifests specifying the fixed perturbation parameters for each episode.
[AI-5] Who Improves the Robot? Toward Physical Recursive Self-Improvement
链接: https://arxiv.org/abs/2610.18842
作者: Hai Zhou,Fan Yang,Zeyi Chen,Lejun Zhang,Chenyang Fu,Jianping Li,Jiayi Zhu,Xianyang Liu,Yingxuan Yang,Lili Meng,Lu Wang,Xinyang Han,Jiankai Sun,Weirui Ye,Guanchu Wang,Wanhua Li,Ming Zhou,Boyi Li,Ming Jin,Kun Shao,Muhao Chen,Guang Chen,Lei Ma,Joey Tianyi Zhou,Costas Spanos,Xia Hu,Mac Schwager,Weinan Zhang,Qian Feng,Shangding Gu
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Robotics (cs.RO)
备注:
Abstract:Robots are increasingly autonomous in task execution, yet their improvement remains largely human-driven. This study investigates Physical Recursive Self-Improvement (Physical RSI), in which physical experience guides lasting updates to both a robot’s capabilities and its improvement procedure, making subsequent improvement more effective across tasks. Physical RSI faces challenges from noisy observations, continuous dynamics, costly trials, and irreversible consequences. We present an L0-L5 framework that differentiates capability-level from meta-level updates and within-task from across-task gains, reserving L5 for independently verified recursive gain. We study existing methods around harness adaptation and model training, alongside safety and control mechanisms that govern execution and successive updates. Our analysis reveals that, despite substantial progress in self-improvement, the surveyed literature does not yet demonstrate Physical RSI as defined by our framework. We identify research priorities in reliable feedback, causal attribution, safety preservation, and benchmarking, outlining a path toward robots that autonomously improve how they learn while ensuring safety. The GitHub repository for Physical RSI is available at this https URL.
[AI-6] Fast Rates for Offline Q-Learning with KL Regularization
链接: https://arxiv.org/abs/2610.18839
作者: Yize Hao,Guangxin Jiang,Shuang Qiu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:This paper investigates offline Q-learning for finite-horizon Markov decision processes with Kullback-Leibler (KL) regularization. Recent studies have shown that KL regularization can yield fast rates in offline decision-making problems. However, it remains unclear whether Q-learning, which adopts incremental stochastic approximation updates, can achieve such fast rates in the offline setting. We first propose KL-regularized Q-Learning (KLQ), a simple variant of classical Q-learning. By exploiting KL regularization, we prove that KLQ achieves a sample complexity of \widetildeO(1/\epsilon) without explicit pessimism. This guarantee, however, has an exponential dependence on the KL regularization parameter and the horizon. To address this limitation, we further propose KL-regularized Pessimistic Q-Learning (KLPQ), which incorporates pessimism into KLQ. We prove that KLPQ retains the same (\widetilde O(1/\epsilon)) rate while removing the explicit exponential dependence, thereby further improving the sample complexity. To the best of our knowledge, these results establish the first fast-rate guarantees for offline Q-learning in tabular finite-horizon KL-regularized MDPs.
[AI-7] RoboSimStudio: An Interactive Toolkit for Rapid Real2Sim of Robot Manipulation Tasks
链接: https://arxiv.org/abs/2610.18835
作者: Kaiyuan Chen,Shuangyu Xie,Uksang Yoo,Xinyue Zhu,Cael Magner,Alvin Zhu,Daehwa Kim,Eric Heiden,Erwin Coumans,Miles Macklin,Spencer Huang,Ken Goldberg
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Graphics (cs.GR); Machine Learning (cs.LG)
备注: 13 pages, 7 figures, 1 table. Project website: this https URL
Abstract:It is difficult to accurately model the physics of the interaction between a robot and the objects it manipulates, including contact mechanics, coupled contact dynamics, deformations, stiffness and friction. We present RoboSimStudio (RSS), an interactive toolkit that makes choices of solver, contact model, and physical parameters explicit and editable, allowing both humans and coding agents to iteratively refine simulations based on observed physical behavior. We implement RSS with the open Newton physics simulation system and describe experiments with human-guided real2sim using (1) visual evaluation, (2) force-feedback and (3) agentic Real2Sim2Real. Code, video and additional information can be found on the project website.
[AI-8] Bytes to Threats: Byte-Sequence Transformer For Malware Detection
链接: https://arxiv.org/abs/2610.18787
作者: Abdelrahman Yassin Eljamal,Abdeltawab Hendawi
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 7 pages, 4 figures. Accepted and published at IEEE ICAD 2026
Abstract:Existing malware detection approaches heavily rely on handcrafted features derived from static or dynamic analysis of the executable files. These methods require specialized domain expertise, dedicated analysis tools, and time-consuming analysis, which limits their scalability and robustness in the face of the increasing complexity and diversity of modern malware. Consequently, there is a growing interest in approaches that operate directly on raw bytes, which contain rich semantic and structural information. However, capturing long-range dependencies within executable files remains challenging for conventional static and CNN-based methods. This paper presents MalwareFormer, a transformer-based, byte-level malware detection framework that operates directly on raw bytes, without requiring domain knowledge or feature extraction. MalwareFormer achieves an accuracy of 98.35%, a balanced accuracy of 98.32%, a macro-F1 of 98.32%, and a false positive rate of 1.94%, outperforming both baselines across all threshold-dependent classification metrics and demonstrating that transformers provide a powerful and scalable approach for byte-level malware detection.
[AI-9] Jeva: System One Models from First Principles
链接: https://arxiv.org/abs/2610.18773
作者: Qi Li,Xinchao Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:System one models enable direct, low-latency decision-making. Yet recent practice of predicting an answer from serialized text does not, by itself, respect the structure of a decision problem. We argue that current designs underemphasize four important structural requirements. (i) null-aware: Models should distinguish answer absence from uncertainty among candidates; (ii) permutation-invariant: Reordering alternatives should not change the decision; (iii) option-scalable: Growing candidate sets should not require redefining the output space; (iv) cacheable: Request-independent computation over recurring alternatives should be reusable across requests. We introduce Jeva, a decision-native architecture that addresses these requirements jointly. A shared encoder independently represents the request and each option, while a compact decision stack refines only the request state against immutable option memory. Symmetric aggregation yields permutation-equivariant probabilities; shared scoring accommodates variable-size candidate sets. A separate coverage estimate distinguishes answer absence from candidate competition, enabling explicit null outcomes. Request-independent option representations and layer projections support reuse across newly assembled sets. These mechanisms unify Choice, Noul, and Score outputs. Across evaluations, Jeva combines strong typed-decision performance with near-perfect null detection and no decision flips under any tested permutations. It scales to thousands of candidates and substantially reduces latency through cross-request reuse without changing decisions.
[AI-10] Beyond the Verdict: Agent ic Code Judging with Verifiable Feedback
链接: https://arxiv.org/abs/2610.18760
作者: Zheye Deng,Sha Li,Changlong Yu,Xin Liu,Qin Lu,Priyanka Nigam,Yangqiu Song
类目: Artificial Intelligence (cs.AI)
备注: 28 pages
Abstract:Test-time scaling for large language models (LLMs) can involve generating multiple candidate solutions for selection or revising an earlier attempt. To guide selection and revision, we need a verifier that can estimate candidate correctness and provide evidence of failure. In competitive programming, however, official judging systems provide authoritative verdicts under finite submission budgets, without revealing failure explanations or failing inputs. To provide informative feedback without consuming official submissions, we formulate a task in which a model judges and diagnoses a candidate program using only the problem statement and source code, with generated counterexamples serving as checkable evidence. To evaluate this task, we build USACO-JUDGE from the USACO 2026 contests, using accepted and unaccepted programs written by human experts and multiple LLMs. We then train ARBITER, a 35B mixture-of-experts (MoE) agentic verifier, with reinforcement learning in a generic execution sandbox. Training raises counterexample success on incorrect programs from 40.61% to 62.67% and mean binary verdict accuracy on USACO-JUDGE from 71.41% to 87.41%, approaching frontier verification performance. Beyond verification, ARBITER outperforms baselines in both code selection and refinement. In an IOI 2026 case study, ARBITER-guided refinement raises GPT-OSS-120B’s expected score above the gold-medal threshold under the official submission limit.
[AI-11] SolBehave: Evaluating the Behavioural Correctness of LLM -Generated Smart Contracts
链接: https://arxiv.org/abs/2610.18745
作者: Giovanni Pio Delvecchio,Stefano Fantazzini,Giacomo Frisoni,Gianluca Moro,Lorenzo Bacchiani,Mario Bravetti,Thomas Troels Hildebrandt
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models increasingly generate Solidity smart contracts that, once deployed, can never be patched, making a single latent fault permanent and exploitable. Yet current evaluations mostly ask only whether generated code compiles. We introduce SolBehave, a benchmark for evaluating the correctness of model-generated contracts at scale, moving beyond ad hoc unit tests. Across 250 real, reproducible Ethereum contracts, we apply three complementary tests of behavioural fidelity: differential fuzzing, historical-transaction replay, and, as our core contribution, synthetic-transaction replay guided by a Dynamic Condition Response (DCR) graph that enumerates the space of legitimate function-call sequences. Differential fuzzing compares each generated function with the original on automatically generated inputs; historical replay compares the whole contract on the transactions it actually received on-chain; DCR-guided replay compares it on sequences of calls that follow its intended usage, including orderings that never occurred on-chain. We evaluate seven state-of-the-art open-weight models after one round of automated repair. Compilation success ranges from 26.4% to 92.7%, but under differential fuzzing only 31.4% of all generated function bodies that compile, pooled over models and samples, are equivalent to the original. Divergence at the contract level is the norm rather than the exception: for six of the seven models, between 13% and 31% of the generated contracts that compile diverge from the original under historical replay, and between 13% and 44% under DCR-guided replay; only Gemma-4-31B stays below 3% on both. Compilation is therefore a necessary condition for correctness and far from a sufficient one.
[AI-12] A Rate Transfer Method: Convergence Rates in Stochastic Gradient Descent via Stochastic Calculus
链接: https://arxiv.org/abs/2610.18739
作者: Neil Walton
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Optimization and Control (math.OC); Probability (math.PR); Machine Learning (stat.ML)
备注: 33 pages
Abstract:We develop a rate-transfer method that converts finite-time Lyapunov estimates for an associated stochastic differential equation (SDE) into finite-time convergence bounds for stochastic gradient algorithms. Under generic moment and smoothness conditions, the convergence analysis of the stochastic gradient method reduces to a calculus argument verifying a Lyapunov generator inequality for the SDE. The method recovers established rates for classical stochastic gradient methods and provides a way to establish convergence rates for new optimization algorithms through calculus arguments applied to their associated SDEs.
[AI-13] Steertuning Policies from a Few Human Videos
链接: https://arxiv.org/abs/2610.18711
作者: Billy Yan,Isaac Tu,Kelly Lee,Blake Chang,Kenny Ma,Kanad Patel,Dhawal Kabra,Neer Patel,Kevin Wu,Lerrel Pinto,Irmak Guzey
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Current recipes for robot learning require massive amounts of robot demonstrations to obtain robust and adaptive policies. In this work, we introduce Steertuning, a system that enables robots to learn new tasks in new environments from 5 human demonstrations in under 5 Steertuning consists of two parts. (1) We pretrain a 400M-parameter diffusion policy on our egocentric human dataset, which consists of 100 hours of densely annotated random interactions. This pretraining makes Steertuning robust to background differences and enables spatial and object generalization when deployed on new tasks. (2) Given 5 demonstrations of a new task, we finetune our model for only 2 minutes (2000 training updates) and steer the policy’s actions during reverse sampling using the demonstration whose observation is closest to that of the deployment environment. Together with a carefully designed data processing pipeline for embodiment transfer, this enables Steertuning to achieve a 71% success rate across 12 tasks in 12 new environments, ranging from lab settings to kitchens in new homes. The entire process, from data collection for a new task to robot policy deployment, takes less than 5 minutes. Our full dataset, model weights, codebase, and robot videos will be made available at our website: this https URL. Subjects: Robotics (cs.RO); Artificial Intelligence (cs.AI) Cite as: arXiv:2610.18711 [cs.RO] (or arXiv:2610.18711v1 [cs.RO] for this version) https://doi.org/10.48550/arXiv.2610.18711 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-14] Unlocking Token Rewards via Training-Free Reward Attribution
链接: https://arxiv.org/abs/2610.18704
作者: Sitong Wu,Haoru Tan,Bin Xia,Xichen Zhang,Jingyao Li,Shaofeng Zhang,Xiaojuan Qi,Bei Yu,Jiaya Jia
类目: Artificial Intelligence (cs.AI)
备注: token-level reward, credit assignment, PRM, RL
Abstract:In this paper, we propose an extremely efficient, training-free method to extract token-level reward signals directly from an existing deep reward model. Our core idea is to attribute the overall process reward to individual tokens by estimating each token’s influence. This influence is defined as the change in the final macroscopic reward (e.g., the process reward) when a token is replaced with a semantically null token. Naively calculating this influence is computationally infeasible, requiring N forward passes through the PRM for an N-token sequence. We overcome this bottleneck by proposing a highly efficient gradient-based estimator. Specifically, we use a first-order Taylor approximation, which simplifies the influence calculation to the inner product of the difference between the token embedding and the null token embedding, and the gradient of the reward with respect to the token embedding. This requires only a single forward and backward pass. The resulting token-level rewards enable standard RL algorithms to perform precise credit assignment without requiring additional reward model training. Experiments on challenging reasoning benchmarks demonstrate that our method substantially improves policy optimization efficiency and enhances the generalization of LLM reasoning capabilities. Our P2T outperforms the outcome reward by +4.9% on MathVista for Qwen2.5-VL-7B-Instruct, and +11.5% on AIME24 for Qwen2.5-Math-7B. The code will be released via this https URL.
[AI-15] ProAS: A Progress-Aware Advantage Shaping Framework for Multi-Turn Agent ic Reinforcement Learning
链接: https://arxiv.org/abs/2610.18683
作者: Bowen Ding,Mingda Wan,Yuhan Chen,Jiyao Yuan,Qi Zhu,Futing Wang,Heyuan Deng,Fei Mi,Lifeng Shang,Tao Lin
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 23 pages, 2 figures
Abstract:Reinforcement learning (RL) has emerged as the primary paradigm for optimizing Large Language Model (LLM) agents, yet efficient critic-free algorithms (e.g., GRPO) suffer from a fundamental uniform credit assignment dilemma that broadcasts trajectory-level advantages equally to all actions. Our analysis reveals that this uniformity inherently reinforces structurally invalid actions in preferred trajectories and ignores the downstream cascading influence of early pivotal deviations in unpreferred ones. To resolve this without introducing critic overhead, we propose Progress-Aware Advantage Shaping (ProAS), a plug-and-play framework that leverages a dynamically evolving Agentic Reference Graph (ARG) to achieve fine-grained credit assignment via Masking Unreasonable Incentives (MUI) and Asymmetric Penalty Redistribution (APR). Extensive experiments across Qwen2.5-1.5B-Instruct and Qwen2.5-7B-Instruct demonstrate that ProAS universally elevates mainstream group-based baselines to new state-of-the-art performance, achieving peak success rates of 91.9% and 94.3% on ALFWorld, alongside 66.7% and 70.8% on WebShop, respectively. Furthermore, it accelerates training convergence by up to 2.44\times while maintaining strong robustness against imperfect or distilled ARGs. Codes: this https URL.
[AI-16] Model-Based Online Decision Making via Generative Trajectory Planning
链接: https://arxiv.org/abs/2610.18677
作者: Haldun Balim,Yilun Du,Na Li
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Robotics (cs.RO); Systems and Control (eess.SY)
备注:
Abstract:Online decision-making faces a tension between efficient inference and long-horizon reasoning. Value-based methods rely on short-horizon bootstrapping that limits the propagation of long-term information, while trajectory optimization methods plan explicitly but must solve a high-dimensional optimization problem at every decision step. We propose Generative Trajectory Planning (GTP), a model-based framework that addresses this tension by learning a generative model as a proposal distribution over action sequences. Instead of optimizing sequences from scratch, GTP samples candidate sequences from a diffusion prior and refines them using learned dynamics, reward, and value models, enabling efficient trajectory-level reasoning. Across continuous control and long-horizon manipulation tasks, GTP matches or exceeds state-of-the-art model-free, diffusion-policy, and model-based baselines, and remains stable in regimes where strong baselines degrade. Our results suggest that generative models are best understood not as policies, but as structured proposal distributions for planning, offering a new perspective on how to integrate generative modeling with model-based decision-making.
[AI-17] AMemOps: Evaluating Memory Updates in Long Audio Streams
链接: https://arxiv.org/abs/2610.18652
作者: Peiwei Chang,Li Zhang,Mark D. Plumbley,Philip J.B. Jackson,Wenwu Wang
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Audio memory must recognise recurring sources across acoustic changes while keeping their attributes up to date. Overall accuracy does not reveal how these demands interact. We introduce Audio Memory Operations (AMemOps), a benchmark for latest-attribute retrieval, preservation of unchanged attributes and absent-source rejection in long audio streams. Paired controls vary recordings, training identities and earlier events to examine their effects on updating. Training on other recordings of the test instruments increases a reference delta-rule memory’s update advantage over matched attention by 13.4 percentage points. At retrieval delays of 120,s, matched histories separate benefits of consistent repetition from costs of conflicting earlier values, while additive-memory controls identify benefits of erasure.
[AI-18] ny Recursive Message Passing
链接: https://arxiv.org/abs/2610.18649
作者: Tommaso Salvatori,Matteo Forasassi,Alexis Toumi,Giovanni de Felice
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 27 pages
Abstract:Message passing neural networks have shown remarkable performance in algorithmic and constraint reasoning, due to their ability to encode the structure of the task directly in the architecture. Recent work has shown that better results can be achieved using recursive transformers trained with deep supervision and truncated recursions, which make extremely deep loops stable to train, and equipped with test-time procedures such as discrete deduction lattices. In this work, we apply the same procedures to message passing networks, and show that the two ingredients are complementary, and lead to parameter efficient models: we obtain about 99% test accuracy on Sudoku-Extreme with a model of 16 k parameters that trains in about 10 minutes on an H100, or in 9 hours on a CPU. We further show that the cost of truncating gradients grows with the number of rounds information needs to cross the graph, making the choice of wiring a trade-off between robustness to truncation and transfer across problem sizes. We conclude by showing that our models transfer zero-shot to larger instances than those seen during training, outperforming popular neural baselines on N-queens and Boolean satisfiability with a fraction of their parameters.
[AI-19] DTScribe: Generating Models for Digital Twin Reporting using Local Large Language Models
链接: https://arxiv.org/abs/2610.18635
作者: Kérian Fiter,Angelica Portocarrero,Bentley Oakes
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:The development of a digital twin brings together many stakeholders and technical components. We have previously investigated the reporting of the conceptual architecture for these digital twins as an artifact for discussion. However, we find that the activity of modeling the digital twin for visualization and consensus-building can be quite tedious and time-consuming. In this article, we explore the automation of model generation using large language models, targeting the model required for digital twin reporting. Our approach is designed to ingest digital twin documents and output both the conceptual architecture visualization and a reporting page for that digital twin. We first extract from the input the key details aligning with our twenty-one-characteristic Digital Twin Description Framework described in previous work. Then, we generate a description model in the Ontological Modeling Language, representing the digital twin as a constellation of models, data, enablers, and services. We utilize refinement loops with large language models acting as judges and ontological reasoning to increase artifact quality as measured by our metrics. We investigate how architectural and parameter choices affect artifact quality, such as how refinement loops enable local large language models to produce sufficiently high-quality results comparable to larger models. We also detail insights and lessons learned useful for practitioners, such that they can integrate our techniques to accelerate their digital twin reporting or for other model generation tasks. These insights include the use of medium-sized local models (around 20 billion parameters), the avoidance of large language models as judges and clustering for characteristic extraction, and the recommendation of at least two reasoner loops for Ontological Modeling Language refinement. Subjects: Software Engineering (cs.SE); Artificial Intelligence (cs.AI) Cite as: arXiv:2610.18635 [cs.SE] (or arXiv:2610.18635v1 [cs.SE] for this version) https://doi.org/10.48550/arXiv.2610.18635 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-20] Geometric Calibration of Supervised Fine-Tuning
链接: https://arxiv.org/abs/2610.18634
作者: Jianzhu Bao,Jia-Chen Gu,Qingyuan Liu,Haozhen Zhang,Kai-Wei Chang,Nanyun Peng,Wenya Wang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Recent studies investigate the weight geometry through its spectral structure, which characterizes how the weights transform different input directions. Within this view, supervised fine-tuning (SFT) is shown to induce dense updates that perturb the spectral geometry of pre-SFT weights along their principal directions, potentially disrupting the model’s prior knowledge. These observations motivate using the geometry of pre-SFT weights as a reference for refining learned SFT updates. We study this perspective as Geometric Calibration and develop SPACE Subspace Projection for Adapted Checkpoint Enhancement), a lightweight post-hoc method that selectively reverts SFT update components within the principal subspace of the pre-SFT weights while preserving those in its orthogonal complement. SPACE is a data-free, training-free method that requires only the pre- and post-SFT checkpoints. We evaluate SPACE on Qwen2.5, Qwen3, Qwen3.5, Llama-3.1, and Olmo-3 models ranging from 2B to 14B, across mathematical, coding, and medical SFT, as well as several SFT variants. Overall, SPACE mitigates out-of-domain (OOD) capability regression while further improving in-domain (ID) performance over the uncalibrated SFT checkpoints. For math SFT, SPACE yields average ID/OOD gains of 2.85/2.26 points on Qwen2.5-7B, and 2.41/1.52 points on Qwen3-14B-Base, respectively. These results highlight how the geometric relationship between SFT updates and pre-SFT weights shapes the adapted model’s capabilities and, more broadly, motivate geometry-aware parameter updates for continual model improvement.
[AI-21] wo Routes to Interoceptive Prioritisation: Dissociating Likelihood Precision and Preference Reweighting in Active Inference ALT
链接: https://arxiv.org/abs/2610.18617
作者: St John Grimbly,Nicolas Kuske,Evert A. Boonstra,Charel van Hoof,Rowan Hodson,Benjamin Rosman,Ryan Smith,Mark Solms,Jonathan P. Shock
类目: Artificial Intelligence (cs.AI); Neurons and Cognition (q-bio.NC)
备注: Accepted at IWAI 2026, forthcoming in the Springer proceedings. 17 pages: the 13-page camera-ready paper, reproduced without alteration apart from one typographical fix, followed by 4 pages of supplementary material. Code: this https URL
Abstract:An embodied agent must prioritise among competing bodily needs. In active inference, it can respond to an urgent need by increasing the precision of information about that need or by increasing its preference for outcomes that relieve it. We compare these two interventions, separately and together, in a foraging gridworld with three lethal needs, and ask how each changes survival, perception and the body model the agent learns. Precision sharpens interoceptive posteriors and improves survival. Preference reweighting keeps bodily states further from their lethal boundaries and changes the states from which the agent learns, so its learned model covers fewer need levels, with lower quality across all levels but slightly higher quality on the levels it visits. An experiment on 72 further layouts does not establish the predicted preference benefit on difficult layouts, but both interventions improve average survival there and their interaction is positive. Survival alone therefore gives an incomplete account of how a model prioritises needs and learns about its body.
[AI-22] Can an Agent Learn to Predict Its Own Failure? Outcome-Grounded Confidence Calibration in LLM Agents
链接: https://arxiv.org/abs/2610.18614
作者: Ashutosh Hathidara
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Controlling an acting LLM agent means deciding, at each step of a long-horizon task, whether to trust its next action, abstain, or hand off to a human. That decision needs a per-step confidence signal predicting whether the current rollout will eventually succeed. The agent’s own self-reported verbalized confidence is the natural candidate, but existing reinforcement learning methods calibrate it against the correctness of a decision at the current step. We propose a method to calibrate it against the eventual outcome of a multi-step episode: under GRPO, the group of rollouts for a task supplies success labels, and a proper scoring rule scores each step’s confidence against the realized outcome. In offline RL on recorded trajectories of a Qwen2.5-7B agent in ALFWorld, TextWorld, and WebShop environments, the AUROC of step confidence against episode outcome rises from near chance to 0.68-0.85, and the top-10% of steps the agent is most confident about succeed ~4.3x more often than average. In online RL on the live ALFWorld agent, four calibration rewards improve early-step AUROC, expected calibration error goes from 0.42 to below 0.12, and selective execution improves over both the untrained agent and a format-only RL control, without degrading task success. We also introduce an evaluation protocol that separates episode depth, task difficulty, and attempt-level foresight, which shows where the gain lies: the trained agent recognizes from its first step which tasks it will fail, while telling a succeeding attempt from a failing one at the same task is harder, with some early-window signal emerging on TextWorld.
[AI-23] Nudge the Judge: Uncovering Program Decision Hijacking in Real-World Jev Applications
链接: https://arxiv.org/abs/2610.18612
作者: Yang Luo,Tiantian Ji,Zifeng Kang,Yucheng Wang,Yong Liu,Zhiyi Yin,Xinran Liu
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:System One models such as Jev change how software makes decisions by mapping open-ended semantic contexts into structured judgments that directly drive program execution. However, this paradigm introduces a new security risk, which we term Jev Program Decision Hijacking (JPDH): attacker-controlled contexts can redirect program decisions through schema-valid Jev judgments. Existing evaluations focus primarily on model-level attacks or predefined injection channels, leaving attack reachability and downstream security consequences in real-world Jev applications underexplored. In this paper, we present JevProbe, the first automated framework to evaluate JPDH by tracing adversarial influence from reachable context surfaces through program-consumed decisions to downstream security consequences in real-world Jev applications. Our key insight is that application logic maps diverse Jev judgments to observable decisions, enabling systematic, program-grounded attack evaluation. Across 228 popular real-world Jev applications and 2,273 test cases, 81.5% exhibit decision flips, of which 98.6% lead to potential downstream security impacts. Compared with the strongest adapted baseline, JevProbe improves macro-average Decision Flip Rate (DFR) by 50.4 percentage points, while existing defenses offer limited protection. We further construct JevProbeBench to support systematic evaluation of JPDH in real-world Jev applications. We have responsibly disclosed our findings to Jev application maintainers, pending their responses. The source code of JevProbe and the JevProbeBench dataset are publicly available on GitHub (this https URL). Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI) Cite as: arXiv:2610.18612 [cs.CR] (or arXiv:2610.18612v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2610.18612 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-24] AgentS tripes: An Evidence-Graded Maturity Model for AI Agents
链接: https://arxiv.org/abs/2610.18603
作者: Nutan Kumar Naik,Vijay Prasad Poudel,Aditya Kumar Saroj,Saurav Samantray,Abhishek Patel
类目: Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Systems and Control (eess.SY)
备注: 32 pages, 7 figures, 4 tables, 116 references; supplementary information (15 pages, 8 figures, 21 tables) included in the same PDF. Code and data available on request
Abstract:Benchmarks report what an Artificial Intelligence (AI) agent can do on a task suite, but not whether one released version has been engineered with the controls that make it reliable, safe and observable, or how strongly that is shown. This paper proposes AgentStripes, an evidence-graded maturity model for AI agents. It defines levels (0 to 7 stripes), process areas (12 areas holding 40 capabilities), three gates (basic, safety, proof) that hold an agent at a level however high its score, and evidence grades E0 to E5. Each applicable capability receives a rating from 0 to 3 and a separate evidence level; these are combined into a credit, aggregated into a severity-weighted score and converted into stripes that the gates limit. Instruments gather evidence, language-model judges interpret it under explicit criteria, and the arithmetic is deterministic. The model is compared with standards at topic level without a claim of conformance, and an unexecuted validation protocol is specified. An empirical evaluation (exploratory, non-confirmatory) of 30 public Python agent repositories, with static evidence only, is a demonstration and not a validation: static evidence caps every final stripe at 2, 26 repositories end at one stripe and 4 at two, and this distinction is unreliable, since re-judging changed the final stripe of 4 of 22 repositories (engine, dossier and panel changed), whereas none of 8 panel-only repeats, all at stripe 1, changed. The parameters are expert conventions without outcome based calibration, and the stripes have not been shown to predict reliability or safety.
[AI-25] Auditing Future-Memory Use in Point-in-Time Language Agents
链接: https://arxiv.org/abs/2610.18602
作者: Rongtian Ye
类目: Artificial Intelligence (cs.AI)
备注: 18 pages
Abstract:Point-in-time (PIT) agents can recall post-cutoff facts even when their external evidence respects a historical cutoff. We examine whether this access predicts evidence gathering in 48 historical company-comparison tasks. Independent closed-book revenue probes identify the better-remembered company for each model on tasks with unequal recall, fixing the predicted direction before agent runs. We then compare paired runs with real or anonymized company identities while keeping external evidence PIT-valid. Both Qwen and DeepSeek recalled future facts, but neither supported the prespecified shift toward the better-remembered company. Qwen’s four allocation estimates on 28 unequal-recall tasks were negative, with descriptive intervals spanning zero. A prospective DeepSeek replication showed an inverse pattern across all four metrics on 20 such tasks, with intervals below zero. These estimates remained negative after each company, pair, and future-fact omission, and all four post-hoc max-T-adjusted sign-flip p-values were below .05. In a prospectively specified follow-up, explicit future facts produced small estimated shifts away from the supplied-fact company, with intervals spanning zero, while final choices shifted toward it. The mechanism behind the natural inverse pattern remains unresolved. These findings support separate assessments of memory accessibility, search allocation, and final outputs alongside cutoff checks on external evidence.
[AI-26] PhysicsSliders: Learning Physical Controls for Audio Foley Generation
链接: https://arxiv.org/abs/2610.18597
作者: Peiwei Chang,Yongqiang Chen,Yuelan Cheng,Liting Gao,Yishun Lu,Li Zhang,Mark D. Plumbley,Philip J.B. Jackson,Wenwu Wang
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Physical properties provide intuitive controls for sounds made by interacting objects. The challenge is to make a generative model respond predictably to these controls, both individually and in combination. We present \method\ (PXS), trained on simulated contact sounds with 15 material, geometry and contact parameters. Two separately trained variants accept direct parameter inputs or numerical captions. PXS controls hardness, size, speed and damping, supports room descriptions and produces spectrally distinct outputs at fixed targets. When controlling damping and size together, specifying thickness improves frequency accuracy and joint target delivery. Measurements from real recordings define effective-damping targets, which simulator inversion maps to generation inputs without real-audio fine-tuning. Across 18 impact classes from five real-audio datasets, PXS-Physical follows these targets with mean grouped Spearman correlation \rho=0.976 and class-median error 0.094 dex (a factor of 1.24).
[AI-27] ACCRUE: Continual Off-Policy Reinforcement Learning for Robot Foundation Models
链接: https://arxiv.org/abs/2610.18596
作者: Changyeon Kim,Kimin Lee,Yashraj Narang,Yijie Guo
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Project website: this https URL
Abstract:Reinforcement learning (RL) post-training improves robot foundation models (RFMs) beyond their demonstrations, but real-world deployment demands more: learning new tasks efficiently while retaining previously acquired behaviors. Existing continual RL methods on top of VLAs mitigate forgetting but rely on on-policy updates with limited forward transfer, whereas off-policy methods improve single-task sample efficiency without addressing learner persistence across tasks. We propose ACCRUE, a continual off-policy RL framework that carries lightweight actor and critic networks across tasks, allowing each stage to reuse behavioral and value knowledge acquired through prior RL. The inherited critic adapts only on current-task transitions and guides current-task actor learning, while prior behavior is preserved by distilling the actor toward cached targets on a small number of retained episodes. Fixed visual features keep the inherited learner and its replay compatible with base-policy updates. Across contact-rich assembly sequences in simulation and on real-robot hardware, ACCRUE accelerates subsequent-task learning with near-zero forgetting and significantly reduces real-robot training time by 31.0% to 68.5%.
[AI-28] GeoMER: A Joint Euclidean-Hyperbolic Geometry Framework for Multimodal Emotion Recognition ICONIP2026
链接: https://arxiv.org/abs/2610.18587
作者: Ting Zhu,Huizhi Liang,Boguslaw Obara,Shufei Duan,Varun Ojha
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
备注: Accepted at ICONIP 2026 (33rd International Conference on Neural Information Processing)
Abstract:Multimodal Emotion Recognition (MER) aims to understand human emotions by integrating information from multiple modalities such as speech and text. Psychological studies suggest that emotions exhibit latent hierarchical structured relationships in their semantic representations. However, most MER approaches rely on Euclidean representations, which capture fine-grained cross-modal interactions but do not explicitly provide a geometric prior for organizing hierarchical emotional structures. To mitigate this gap, we propose GeoMER, a joint Euclidean-Hyperbolic modeling framework that introduces hyperbolic emotion-core routing to capture higher-level, abstract emotional representations for MER. The proposed framework jointly models token-level cross-modal semantic interactions and latent emotion representations in two complementary geometric spaces. Specifically, GeoMER consists of a Euclidean Multimodal Fusion Module (EuclFusion) for multimodal representation fusion and a Hyperbolic Geometric Core Routing (HypGeoRoute) module that performs geometry-aware core assignment and aggregation in a hyperbolic manifold to capture more abstract feature representations. The HypGeoRoute module complements the Euclidean token-level modeling. We experiment on the IEMOCAP multimodal MER benchmark and demonstrate that the proposed method achieves the best performance, with 83.54% weighted accuracy and 83.35% unweighted accuracy, outperforming Euclidean baselines, including MemoCMT.
[AI-29] Compare to Correct: Pairwise-Guided Logit Adjustment for LLM Test-Time Scaling
链接: https://arxiv.org/abs/2610.18583
作者: Cong Hua,Qianqian Xu,Zitai Wang,Zhiyong Yang,Shilong Bao,Linchao Zhang,Qingming Huang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Sampling-based test-time scaling enhances LLM reasoning by aggregating multiple reasoning paths for the same query. Existing methods use answer frequencies or path confidence to compute candidate scores, which do not explicitly characterize comparisons between competing answers. Inspired by the Condorcet principle in social choice theory, we introduce pairwise confidence comparisons between paths supporting different answers to capture answer-level comparative advantage. We combine this comparative evidence with voting support through a KL-regularized objective. Theoretically, we show that this comparative-advantage is equivalent to the adjustment to the traditional voting logits. In this way, our proposed Pairwise-Guided Logit Adjustment (PGLA) framework allows comparative evidence to compensate for weaker voting support and correct voting-based ranking errors. Experiments on 10 challenging benchmarks including mathematical reasoning, knowledge reasoning, and code generation with 7 models ranging from 1.7B to 32B parameters speak to the effectiveness of this plug-and-play term. Additional evaluations with external reward scores and online sampling further demonstrate its applicability.
[AI-30] Reading Right Acting Wrong: Diagnosing Decision Failures of System One Models in Agent Loops
链接: https://arxiv.org/abs/2610.18579
作者: Yubo Li,Ramayya Krishnan,Rema Padman
类目: Artificial Intelligence (cs.AI)
备注: Initial version released to encourage open discussion and share preliminary findings. Evaluations of additional open-source System One models are ongoing
Abstract:System One models choose an agent’s next action by scoring a supplied set of options in a single pass. This makes them fast and cheap, but when they choose wrongly there is nothing to inspect: did the model misread what the agent has seen, or read it correctly and still act wrongly? We answer behaviorally. On a frozen benchmark of agent decision states from simulated environments, customer-service policies, tool use, web navigation and coding, we pair every next-action question with factual probes about the same state, each answered independently. Across seventeen fast systems and a reasoning model, the answer is consistent: most wrong actions follow correct reading, and the more accurate a system, the more of its errors take this form. These errors are not noise; they gather on the same states across systems of every kind. Where the probed facts provably determine the action, models that read every fact still fail to combine them, and their own answers, composed by code, repair the failure. On real states, what is missing is often not a fact but the next sub-goal.
[AI-31] When More Objectives Make Identification Easier: The Role of Preferences and Feedback
链接: https://arxiv.org/abs/2610.18577
作者: Bo Xue,Ji Cheng,Yuanyu Wan,Shuang Qiu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Best-arm identification (BAI) aims to identify the optimal arm using as few samples as possible. While most existing work focuses on a single objective, many real-world applications involve multiple criteria, which motivates the study of BAI in multi-objective bandits. Intuitively, multi-objective identification seems harder, since the learner must identify optimal arms in a higher-dimensional reward space; however, the additional objectives also provide richer feedback that can reduce the required exploration. In this paper, we characterize when and why multiple objectives can make BAI easier. For Gaussian bandits, we establish a characteristic-time lower bound for fixed-confidence identification that holds uniformly across preference orders. Through explicit comparisons under Pareto, linear, and lexicographic preferences, we reveal three phenomena: (i) identifying a larger optimal set can require fewer samples than identifying a unique optimum; (ii) the advantage of vector feedback over coordinate feedback depends fundamentally on the preference order; and (iii) multi-objective identification can require fewer samples than even the easiest corresponding single-objective problem. Finally, we propose Order-Aware Track-and-Stop (OATS), an algorithm that is correct at any prescribed confidence level and asymptotically matches the lower bound.
[AI-32] LLM Agents for Automated Post-Training via Structured Recipe Optimization
链接: https://arxiv.org/abs/2610.18572
作者: Ruohong Liu,Bohan Tang,Lekang Jiang,Jack Umenberger,Yiwen Guo
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Automated post-training transforms pre-trained large language models (LLMs) into models capable of handling diverse domains by automatically constructing and optimizing a recipe of coupled decisions over data sources, data processing, cross-domain composition, and training configuration under a limited experimental budget. Existing automation either searches predefined recipe spaces that enable systematic optimization but remain fixed rather than adapting their available choices to the target tasks, or uses open-ended agents that discover task-specific choices but leave them implicit in trial-level edits, limiting their systematic reuse and recombination across trials. To fill this gap, we introduce optimization over a task-adapted recipe space, a novel paradigm that can adapt the recipe space itself to the target tasks while representing its choices explicitly for systematic reuse, recombination, and joint optimization. We instantiate this paradigm with Structured Recipe Optimization (SRO), comprising an LLM-based Recipe-Space Designer that constructs the task-adapted recipe space and a Multi-Domain Recipe Optimizer that uses structured trial evidence to guide LLM candidate construction and multi-objective Bayesian trial selection. Extensive experiments show that SRO achieves the highest average score in all evaluated target-model and agent-LLM combinations. In joint post-training across three capability domains and six benchmarks, SRO reaches 41.1 versus 15.4 for the strongest baseline on Qwen3-1.7B with GPT-5.4, and 53.8 versus 40.8 on Qwen3-4B with DeepSeek-V4-Flash. Our code is available at this https URL.
[AI-33] Order-Consistent Latent Geometry for Reliable Short-Horizon Planning in World Models ECCV2026
链接: https://arxiv.org/abs/2610.18570
作者: Junlin Wu,Wen Fan,Dandan Zhang
类目: Artificial Intelligence (cs.AI)
备注: Accepted for poster presentation at STEAI 2026 (ECCV 2026 Workshop)
Abstract:Latent planners often rank candidate action sequences by terminal Euclidean distance to a goal latent. Exact agreement with transition distance would make this score ideal, but is stronger than planning requires: along a transition-shortest rollout, the essential requirement is that latent distance does not reverse true transition progress. We show that this pathwise no-reversal requirement is precisely the classical self-approaching condition. To the best of our knowledge, we are the first to identify self-approachingness as a planner-facing compatibility condition for terminal Euclidean latent planning and to connect its violation to an exact finite-horizon candidate-order reversal. Because visual world models predict discrete latent sequences, we analyze their polygonal interpolation. We prove that an acute bend creates a nonempty region in which terminal Euclidean distance ranks a transition-worse candidate ahead of one that makes genuine progress, even under exact rollouts. We further interpret latent dimension as angular capacity through a classical spherical-code bound. A controlled six-branch experiment, including a zero-padded three-dimensional baseline, validates the predicted ordering failure and separates angular arrangement from ambient dimensionality.
[AI-34] Configuring RAG Systems: What Breaks Where Originates and How to Fix
链接: https://arxiv.org/abs/2610.18567
作者: Liangyu Chen,Youpeng Ma,Yulong Ye,Zezhen Xiang,Tao Chen
类目: Artificial Intelligence (cs.AI)
备注: 28 pages, 7 figures
Abstract:Retrieval-Augmented Generation (RAG) has emerged as a dominant paradigm for improving Large Language Model (LLM) by intertwining it with other components, such as a vector/graph database. Therefore, the resulting RAG systems are often highly configurable, under which a non-trivial amount of failures are related to the configurations. However, there remains a lack of systematic understanding regarding the configuration issues for RAG systems. To bridge this gap, in this paper, we conduct the first large-scale empirical study of 654 real-world configuration issues collected from 10 diverse RAG systems. Drawing on this, we provide a complete categorization of the symptoms, root causes, relevant stages, and fix patterns of configuration issues for RAG systems, as well as several new observations. Our findings not only reveal the characteristics of configuration issues, but also strongly link them to the unique pipeline stages of RAG, e.g., we found that most symptoms can be observed at the early or late stage, while the stage of the root cause might not be at the same stage as where the symptom is identified. We also provide nine actionable insights to the practitioners on testing, localizing, and fixing the configuration issues for RAG systems.
[AI-35] Rethinking Quality in Large Language Model Routing: Security-Conditioned Model-and-Budget Selection
链接: https://arxiv.org/abs/2610.18566
作者: Yichi Zhang,Zhenxiang Huang,Yuyang Zhang,Yuchen Yang
类目: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注:
Abstract:Large language model (LLM) routers reduce serving cost by choosing which model to call and how many output tokens to grant it for each query. Existing routers rank model-budget actions using a single quality metric. This does not work for mixed workloads. For ordinary queries, a good answer is correct and useful (utility); for security-sensitive queries, a good answer refuses harmful requests or resists malicious instructions (security). These two objectives prefer different models and budgets. We propose SCRouter, a router that learns when to prioritize security and how to allocate the token budget. For each model-budget pair, it estimates the probability that the response finishes within budget. It uses this probability to predict utility and security separately. A small probe determines how to weight these two predictions for each query. SCRouter then selects the model and budget by balancing the weighted utility and security against cost. We also build SCRouterBench, a benchmark of 38,869 utility and security queries spanning nine security dimensions, each answered under all 144 combinations of 12 models and 12 budgets, plus 2,753 out-of-domain queries. SCRouter improves the baselines by 13-29 points in-domain and 19-23 points out-of-domain. Subjects: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR) Cite as: arXiv:2610.18566 [cs.AI] (or arXiv:2610.18566v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2610.18566 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-36] Input-Conditioned Search for Efficient Test-Time Scaling of Recursive Reason ers
链接: https://arxiv.org/abs/2610.18547
作者: Haochen Shi,Tianwen Fu,Qianggang Ding,Yinhan Lu,Yizhan Li,Huan Zhang,Yajie Zhao,Xingdi Yuan,Hengyang Lu,Bang Liu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:People often tackle difficult problems by thinking more deeply or exploring more possibilities. Similarly, recursive reasoners, which reason by repeatedly applying a learned update rule to latent states, can also scale inference through more updates (depth) or additional trajectories generated by perturbing latent states (width). When people hit a dead end in their thinking, they may step back and approach the problem from a fresh angle. However, neither depth scaling nor perturbation-based width scaling of recursive reasoners can reliably achieve this: both may converge to incorrect attractors. To address this, we propose to explore diverse reasoning traces by applying different update maps to the same latent state, without any additional training. We realize this idea through \emphInput-Conditioned Search (ICS), which searches over input conditions to induce alternative update maps from the pretrained reasoner. Across six reasoning tasks, ICS achieves the highest mean accuracy, while requiring up to two orders of magnitude fewer model FLOPs than the evaluated scaling baselines. Controlled experiments confirm that ICS can redirect trajectories from the same state. Our theoretical analysis and rigorous numerical verification also confirm that ICS can locally eliminate incorrect attracting fixed points. Together, these findings establish ICS as an effective axis for efficient test-time scaling and motivate further research on shaping recursive reasoning during training and at test time.
[AI-37] APTree: Extended Behavior Trees for Adaptive and Scalable Automated Robotic Task Planning in Construction
链接: https://arxiv.org/abs/2610.18523
作者: Shermin Sherkat,Thomas Wortmann,Andreas Wortmann
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: Accepted for publication in The Journal of Systems and Software. 46 pages, 17 figures, 16 listings, 6 tables
Abstract:Task planning for robots in off-site construction prefabrication is primarily manual, leading to planning inefficiencies and reluctance to adopt robotic solutions. Existing approaches provide limited support for scalable, modular automated task planning and localized runtime replanning. This paper proposes APTree, a model-driven framework that integrates automated task planning with modular task execution using behavior trees as the execution model. APTree repeatedly applies and coordinates heterogeneous symbolic planners to automatically generate and compose modular task plans at runtime. To support this approach, we develop a domain-specific language for the formal specification and validation of task plans, together with a compatible execution engine. We evaluate APTree on the real-world livMatS Biomimetic Shell pavilion, where it models the construction robotic fabrication domain, detects injected specification and planning errors at design time and runtime, and automatically generates large-scale task plans by coordinating three symbolic planners across numerous planning modules. A machine-checked formal analysis further shows that composed execution and local replanning are safe under explicitly stated assumptions, including planner soundness and non-interference of generated plans, providing practitioners with explicit conditions for dependable plan composition. APTree thus enables construction practitioners to leverage automated task planning without requiring detailed knowledge of symbolic planning techniques or APTree’s internal implementation.
[AI-38] Lost in Implicit Routing: How LLM Agents Fail to Load Relevant Skills at Scale
链接: https://arxiv.org/abs/2610.18522
作者: Jinlin Liu,Wei Chen,Xiaojin Zhang
类目: Artificial Intelligence (cs.AI)
备注: 22 pages, including references and appendices. Code: this https URL
Abstract:Reusable skills let LLM agents load task-specific instructions on demand, but they make execution depend on implicit routing over skill metadata. We test whether this design scales by holding SkillsBench tasks and their golden skills fixed while adding irrelevant skills. We find skill routing collapse: agents increasingly load fewer or no skills, even though the relevant skills remain available. With 1,000 irrelevant skills, loaded-skill recall falls from 73.6% to 27.4% and the zero-load rate rises from 9.3% to 51.6%, whereas explicit selection from the same library retains 73.9% recall. Loading other skills, including plausible alternatives to golden skills, is too rare to explain this decline. Masking only the descriptions of irrelevant skills largely restores skill loading, yet with the metadata intact, explicitly asking agents to plan skill use still yields lower recall than preselection. These findings point to semantic interference from descriptions and instruction interference when skill choice competes with task execution. We also evaluate explicit preselection as a practical mitigation, including its coverage and latency trade-offs. Our analysis informs skill protocols and evaluation for agents with growing skill libraries.
[AI-39] A Global View of How LLM s Value Human Time
链接: https://arxiv.org/abs/2610.18515
作者: Yifan He,Sofie Goethals,Manon Reusens,David Martens
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) increasingly act as personal assistants; however, little is known about how they capture national-level preferences on a global scale. Given that “time” is a universal and representative concept, we examine LLMs’ time preferences by varying only users’ country and measuring the compensation LLMs deem appropriate for a 60-minute appointment delay. We then compare LLMs’ responses with GDP per capita on Purchasing Power Parity (GDP_PPP), Mean Income, and a literature-based index (Patience Score). Our results across 193 countries and 5 LLMs demonstrate that people’s time across the world is evaluated in noticeably different ways among LLMs, as well as varying levels of robustness under prompt variations. LLMs consistently recommend the highest compensation for Europe, but mostly the lowest for Africa. Based on GDP_PPP, LLMs underestimate the value of time for users living in Asia and South America. These findings indicate that geographical patterns are embedded in LLMs and may translate in different treatment in agentic settings.
[AI-40] he Smart Buildings Challenge: Learning Foundation Models for Real-World Optimization at Scale NEURIPS2026
链接: https://arxiv.org/abs/2610.18511
作者: Judah Goldfeder,Ayush Rai,Philippe M. Wyder,Na Li,Victoria Dean,Donna Vakalis,Jiong Lin,Bing Dong,Michael Rossetti,Nathan Kutz,Sohei Arisaka,John Sipple,Hod Lipson,Yoshua Bengio
类目: Artificial Intelligence (cs.AI)
备注: This is for a Neurips 2026 competition
Abstract:Buildings cause nearly 40% of carbon emissions in the US. Of total building emissions, HVAC systems are responsible for approximately 40-60% and account for 15% of global energy consumption. Despite advances in artificial intelligence, most buildings still operate as inefficiently as they did decades ago. Building optimization, perhaps more so than any other optimization problem, has enormous potential for Machine Learning (ML) driven impact on climate change, in both the short and long term. One of the biggest obstacles for building optimization is the lack of policy transfer. Most building control systems, whether simulation or learning based, don’t generalize: they only work well for the building they were developed for, but not for other buildings. To solve this problem, we propose a challenge for the development of building foundation models that easily adapt to new buildings with minimal configuration and thereby enable building optimization to scale across the globe. Participants will build control policies on a set of simulated buildings, and have their policies evaluated on an unseen set of buildings. The buildings vary in size, layout, weather, material properties, and HVAC system. Successful outcomes from this initiative promise a massive leap forward in global sustainability efforts and demonstrate the great potential of ML to combat climate change.
[AI-41] ReTerm: Recovery Training for Self-Improving Terminal Agents
链接: https://arxiv.org/abs/2610.18484
作者: Wei Xie,Kaituo Feng,Shiming Fu,Yong Liu,Xiangyu Yue,Shuangjia Zheng
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Software Engineering (cs.SE)
备注: Project page: this https URL
Abstract:Terminal agents have shown strong abilities in complex coding tasks. However, they often struggle to recover from errors during long-running execution (e.g., command errors, faulty implementations), eventually leading to task failures. To this end, we present ReTerm, a self-improving terminal agent trained by learning how to recover from its own failed terminal tasks. To support its training, we construct a customized data pipeline and produce ReTerm-10K, a high-quality dataset that provides effective recovery demonstrations for SFT cold start. Furthermore, we propose ReTerm-GRPO, an online RL algorithm that forks from the agent’s own failure states to generate multiple independent self-recovery attempts. By learning from whether these attempts successfully complete the task, the agent self-improves through its own recovery experience. Built on Qwen3.5-9B, our ReTerm-9B achieves 34.8% Pass@1 on Terminal-Bench 2.0, improving over the base model by 14.6 points. Further analysis shows that our method achieves consistently higher recovery rates across diverse error categories compared with standard GRPO training. All code, data and model are released at this https URL.
[AI-42] MemoryVLN: From Ego-Aware Memory Representation to Cross-Context Memory Reuse in Vision-Language Navigation
链接: https://arxiv.org/abs/2610.18478
作者: Xueyao Wei,Yifei Wang,Delun Li,Shaoan Wang,Junzhi Yu,Wenjie Song,Xingyu Chen
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Large vision-language models (LVLMs) have shown strong potential for embodied vision-language navigation (VLN), yet their performance degrades on long-horizon trajectories due to context-length limits, redundant historical observations, and unstable spatial alignment across time. These issues make it difficult to both compress long navigation histories and reliably reuse past spatial experience. To address these limitations, we propose \textbfMemoryVLN, a unified memory-driven navigation system built around memory construction and management. During memory construction, \emphLong Sequence Compression employs forgetting-curve-inspired temporal sampling to preserve dense recent evidence and sparse yet representative distant observations under a limited token budget. \emphEgo-Aware Memory Representation further integrates 2D semantic tokens, 3D geometric tokens, and 6D agent poses, allowing each memory to encode both what was observed and where it was observed. During memory management, \emphHierarchical Memory Bank maintains in-episode and context-level memories through cross-context warm-start and confidence-controlled write-back. \emphMemory Enhancement then retrieves the Top- K relevant memories according to the current navigation state and fuses them with the current visual tokens. Together, these mechanisms transform raw navigation history into compact, spatially grounded, selectively retrievable, and reusable navigation experience. Extensive experiments on R2R-CE and RxR-CE demonstrate consistent improvements over the state-of-the-art baselines by a substantial margin (i.e., 2% and 6.5% in successful rate).
[AI-43] Learning Perturbation-Specific Representations through Conditional Energy Prior for Cellular Response Prediction
链接: https://arxiv.org/abs/2610.18456
作者: Yuchen Yan,Yatao Bian
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:AI-based cellular perturbation prediction is essential to accelerate drug discovery and therapeutic identification. However, perturbation-specific signals are generally obscured by cellular complexity, making it difficult for existing methods to correctly capture perturbation effects. To address this problem, we propose PerturbEnergy, which introduces a conditional energy-based prior over perturbation representations. It assigns lower energy (higher probability) to representations compatible with the given perturbation and higher energy to others, thereby explicitly tilting the latent space toward perturbation-preferred regions. Besides, PerturbEnergy can be integrated into autoencoder architectures as we find its Evidence Lower Bound (ELBO) is equivalent to the ELBO of the complete autoencoder. We further prove that optimizing PerturbEnergy increases the lower bound on the conditional mutual information between representation and perturbation. Extensive experiments on four benchmark datasets demonstrate the effectiveness of our PerturbEnergy: it improves perturbation recognition especially in latent space with improvement of up to 83.2%; It also achieves state-of-the-art performance for predicting top-50/100/200 differentially expressed gene(DEG) response and remaining competitive on highly variable gene(HVG) prediction. The code is provided: this https URL
[AI-44] Autonomous Science Bench: Evaluating AI Agents for Autonomous Scientific Discovery
链接: https://arxiv.org/abs/2610.18449
作者: Yibo Wen,Han Liu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:AI agents for autonomous scientific discovery must decide which experiments to run under a limited budget. Most benchmarks for science agents provide fixed datasets, while others that support active experimentation largely rely on simulation. We introduce Autonomous Science Bench, an initial release comprising six budgeted design-build-test-learn campaigns grounded in published experiments across biology, chemistry, and materials science. Agents select batches of designs, receive measurements from a lab service, and use the results to guide subsequent experiments before delivering a predictive model. Success requires passing two calibrated gates: a discovery gate, which evaluates the designs the agent measured, and a learning gate, which evaluates its model’s ranking of held-out designs. In a pilot evaluation of eight agent-model configurations and 144 trials, the best configuration succeeds in 67% of trials, while the hardest campaign is solved in only three of 24 attempts. The two gates reveal different weaknesses: some agents build accurate rankings but find too few promising designs, while others find promising designs but fail to predict which unseen designs will perform well. These differences are obscured by overall success rates alone. We release version 0.1 of the campaigns, lab runtime, and public leaderboard at this https URL.
[AI-45] he Singularity Has a Distribution: Exact Timing Laws and Detection Limits for a Stochastic Intelligence Explosion
链接: https://arxiv.org/abs/2610.18445
作者: Xiaoyu Li,Zheng Gao,Chiwun Yang,Zhizhou Sha
类目: Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Probability (math.PR)
备注: 24 pages, 6 figures, 3 tables. Code and data: this https URL
Abstract:Automating AI research creates a positive feedback loop between software capability and research effort. We study a stochastic realisation of this loop in which independent, positive, multiplicative improvements arrive at a rate proportional to a power of current capability. Within this model, randomness leaves the deterministic explosion threshold unchanged but replaces the blow-up date by an explicit distribution. For equal-sized improvements the explosion time is a geometrically weighted sum of independent exponentials; its q -series representation gives its transforms, moments and tail behaviour, including a two-term small-time asymptotic. For Lomax-distributed improvement sizes the explosion time is exactly Gamma distributed. Self-similarity implies that, with parameters known, the relative uncertainty about the remaining time does not vanish as explosion approaches. We formulate a fixed-sample test between constant and geometrically increasing idea-arrival rates. If the baseline rate and the alternative log-rate increment \ell are known, the minimum sample size at error level \alpha is asymptotic to (12z_\alpha^2)^1/3\ell^-2/3 , where z_\alpha is the upper- \alpha standard normal quantile. Uncertainty about the growth exponent near its critical value produces a 1/t tail with an explicit constant, and hence an infinite conditional mean waiting time, although explosion is certain given a positive exponent. Separate extensions show that a fixed deployment delay prevents explosion in the specified delayed-feedback models, that power-law diversion of effort shifts the growth threshold, and that in a deterministic oversight race a fixed allocation share does not change the asymptotic regime, except at a critical equality of exponents. Numerical illustrations, which are not empirical forecasts, show how timing and detectability depend on the granularity of improvements.
[AI-46] FedGraph: Defending Federated Large Language Model Fine-Tuning Against Backdoor Attacks via Graph-Based Aggregation ICLR2026
链接: https://arxiv.org/abs/2610.18430
作者: Xi Chen,Chunyi Zhou,Rui Zeng,Xiaogang Xu,Zhe Liu,Shouling Ji
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: Accepted at the ICLR 2026 Workshop on Principled Design for Trustworthy AI. This is an author-owned preprint, not the official proceedings version
Abstract:Federated fine-tuning of large language models (LLMs) allows multiple clients to collaboratively adapt models without sharing raw data, offering strong privacy benefits. However, this setting is highly vulnerable to backdoor attacks, in which malicious clients manipulate their updates to induce targeted behaviors while maintaining normal performance on benign inputs. In this work, we make a key observation specific to federated LLM fine-tuning: malicious clients tend to exhibit strong consistency in their model updates, whereas benign clients produce more diverse and dispersed updates. This structural difference is not well captured by existing defense methods, which typically rely on direct anomaly detection in high-dimensional parameter space and therefore struggle in the presence of entangled LLM updates. Motivated by this observation, we propose FedGraph, a graph-based aggregation framework that explicitly models relationships among client updates. By representing clients as nodes in a dynamically constructed graph and extracting structural features that characterize their relative positions and interactions, FedGraph derives compact client representations that effectively separate malicious and benign behaviors. These representations are then used to identify and suppress malicious contributions during aggregation. Extensive experiments demonstrate that FedGraph significantly improves backdoor defense. It consistently reduces attack success rates to below 10% while achieving high detection accuracy (95.5% on average) and a low false positive rate (2.33%), substantially outperforming existing federated learning defenses. Comments: Accepted at the ICLR 2026 Workshop on Principled Design for Trustworthy AI. This is an author-owned preprint, not the official proceedings version Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI) Cite as: arXiv:2610.18430 [cs.CR] (or arXiv:2610.18430v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2610.18430 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Xi Chen [view email] [v1] Tue, 13 Oct 2026 14:08:49 UTC (4,085 KB)
[AI-47] Delegating Earth science model calibration to knowledge-guided LLM agents
链接: https://arxiv.org/abs/2610.18378
作者: Zhanliang Zhu,Ziwei Li,Liujun Zhu,Yuchen Liu,Junliang Jin,Jianyun Zhang
类目: Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE)
备注:
Abstract:Applying process-based Earth science models to new sites and questions often requires expert calibration decisions. Reliable delegation to language-model agents must preserve scientific competence and reproducibility. Here we develop a general calibration harness (GCH) for end-to-end delegation from a calibration request to assessed results. Constrained by a knowledge infrastructure (KI) and a validation convention, agents define parameter domains, observation mappings and objectives, and generate model-specific code. GCH checks the formulations against the same declarations and freezes them before numerical search and assessment, which require no further agent judgement. Agent-authored workflows executed across six model families, with assessment identifying unmet criteria. In snow-model tests, five of five knowledge-guided specifications executed on the first attempt, compared with none of five given only the model name, achieving validation Nash-Sutcliffe efficiency (NSE) of 0.81-0.94 against 0.57 for defaults. Under a common objective, independent GR4J formulations yielded scores within 0.001 NSE; re-evaluating eleven saved parameter vectors reproduced their metrics. In ten paired VIC tests with matched budgets, fixed search was non-inferior to continuing agent proposals (margin, -0.02 NSE; median difference, -0.003). This separation enables end-to-end calibration delegation with inspectable scientific choices and reproducible numerical evaluation.
[AI-48] ABot-ER: Unified Embodied Reasoning through Agent ic Data-Model Co-Evolution
链接: https://arxiv.org/abs/2610.18355
作者: Qiuyu Ding,Heng-Da Xu,Changda Xia,Jiajun Shen,Zihao Guan,Zhengchen Yuan,Jianfang Zhang,Di Yang,Jiayi Tian,Yuting Xu,Fei Wang,Shiao Liu,Wei Zhang,Xiaolong Wu,Wenbin Tang,Zedong Chu,Mu Xu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:We introduce \textbfABot Embodied Reasoning (ABot-ER), a unified embodied reasoning foundation model that breaks the tension between breadth and depth in embodied reasoning. Embodied intelligence demands the joint mastery of perception, planning, and execution, yet reconciling generality with specialization across this spectrum remains an open challenge. Within a single unified model, ABot-ER commands specialist-level depth across all three competencies, spanning spatial, scene, and temporal reasoning in perception; long-horizon action organization in planning; and point grounding and affordance in execution. The \textbfAgentic Data-Model Self-Evolution Engine continuously adapts data strategies to the model’s evolving capability state. The \textbfTask-Conditioned Adaptive Training Strategy uses task tokens to activate lightweight low-rank residuals over a shared backbone, separating updates from heterogeneous output formats while retaining shared embodied knowledge. Across a systematic evaluation on twelve benchmarks covering three core competencies and five capability dimensions, ABot-ER achieves \textbfstate-of-the-art overall performance across perception, planning, and execution within a unified parametric framework, demonstrating that breadth and depth can coexist in unified embodied reasoning.
[AI-49] Let the Mind Conceive Let the Hand Achieve: Following Multimodal Instructions for Long-Horizon Robotic Manipulation
链接: https://arxiv.org/abs/2610.18352
作者: Youpeng Wen,Gang Wang,Tianyu Huang,Weiwei Zhang,Zhijian Huang,Wei Chen,Xieyuanli Chen,Yunhui Liu
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:A robot following a multi-step instruction needs to act on the right objects in the right order, even as the scene changes. It must also maintain awareness of task progress and determine when to advance. To address these challenges, we introduce Orchestra, an agent harness that separates task planning and execution management from action generation. Orchestra decomposes a complete multimodal instruction into ordered subtasks, tracks completed steps, and updates the locations of relevant objects and destinations during execution. It maintains a shared context containing the active subtask, spatial grounding, and recent observations. An action policy uses this context to generate actions, while a separately trained progress evaluator estimates when to advance to the next subtask. To evaluate whether robots trained on atomic instructions and short trajectories can execute longer subtask sequences, we introduce a benchmark named RMBench-Compose. Orchestra achieves a mean success rate of 68.22% across nine RMBench tasks, outperforming Mem-0 by 26.22 percentage points. On RMBench-Compose, it achieves an Average Success Length (ASL) of 7.35 subtasks, 2.46 times that of HarnessVLA. Across four real-world multi-step tasks, it improves mean ASL by 25.0% over Orchestra using the base \pi_0.5 policy. These results show that Orchestra helps robots carry out longer sequences of subtasks while preserving the requested subtask order and acting on the specified objects. Our code and project page are available at this https URL.
[AI-50] AssetForge: Asset-Augmented Task Synthesis for Business Automation Agents
链接: https://arxiv.org/abs/2610.18330
作者: Xiangrui Liu,Shuo Tang,Siheng Chen
类目: Artificial Intelligence (cs.AI)
备注: 22 pages, 5 figures
Abstract:Training business automation agents requires executable tasks that connect a natural-language request to an application state and a verifiable end state. Traditional Rubrics requires Authors to recreate application interfaces, initial records, policy evidence, and outcome checks for each task, entangling business logic with repeated infrastructure work. We introduce AssetForge, an asset-augmented task-construction pipeline that pairs a reusable construction-asset library with task-specific Rubrics. The assets provide application interfaces, world-state components, policy/evidence carriers, and native assertion definitions; the Rubric directs their instantiation for a particular business task. Executable checks and an additional model-based review validate each candidate, return targeted feedback, and support bounded repair before acceptance. On AutomationBench v1.0.6 public-600, fine-tuning Qwen3.6-35B-A3B on 6,000 AssetForge examples yields 41.17% on the official strict pass@1 metric, 4.4 times the 9.33% base score (+31.83 points). With the same backbone and GLM-5.3 supervision, AssetForge reaches 41.17% versus 30.33% for traditional Rubrics, a 10.84-point gain (35.7% relative) at the same accepted-example scale (6,000 versus 6,000 examples). Both DeepSeek- and GLM-supervised AssetForge models improve AppWorld completion over the base model on Test-Normal and Test-Challenge. These results establish the advantage of AssetForge’s complete construction pipeline on AutomationBench public: the coupled asset-Rubric interface directs Authors toward business logic, while executable checks and model-based review validate their tasks.
[AI-51] Complex materials engineering via Provenance Action Graphs: potential structure and demonstration
链接: https://arxiv.org/abs/2610.18314
作者: Yevgeny Rakita,Paul Todd,James R. Neilson,Simon J. L. Billinge
类目: Artificial Intelligence (cs.AI); Materials Science (cond-mat.mtrl-sci)
备注: 13 pages, 8 figures
Abstract:This work discusses and tests the potential of reaction graphs for material synthesis, in which events consist of reaction steps (actions) and their outputs (observations). The importance comes from the requirement to synthesize technologically useful materials, especially those that are kinetically stabilized. Often, it is a science that follows years of trial and error, culminating in (published) successful recipes, but is truly driven by far more (unpublished) failures. The entry into the era of AI provides an opportunity to use learning algorithms to encode rules from multi-dimensional synthesis pathways, regardless of their technological usefulness, and to decode both forward prediction and inverse learning modes. The main incentive is to accelerate material discovery by generating a global knowledge base that can be shared among data agents and contributed to by them. We will focus on several scenarios in the common synthesis process during solid-state synthesis, primarily because these reactions are kinetic in nature and require closer monitoring and control of the reaction history.
[AI-52] Efficient Safety Guardrails through Latent Dual-Hypothesis Reasoning with On-Demand Trace Reconstruction
链接: https://arxiv.org/abs/2610.18304
作者: Md Asiful Islam,Fahmida Alam,Mihai Surdeanu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Guardrail approaches that rely on reasoning-based large language models (LLM) improve safety decisions, but come with substantial inference overhead due to the explicit reasoning. We introduce LR-Guard (Latent Reasoning Guardrail), an efficient and explainable guardrail that jointly learns structured latent reasoning and on-demand reasoning reconstruction. Our method distills a frozen teacher’s analysis of safe and unsafe prompt interpretations and their comparison into compact latent representations. Joint training optimizes these representations for both safety prediction and reconstruction of the teacher’s explicit reasoning traces, enabling human-readable explanations without requiring intermediate text generation for every decision. Experiments show that LR-Guard outperforms its explicit-reasoning teacher in most in-domain and out-of-domain settings on safety classification and explanation extraction, while achieving a 19.4x inference speedup at batch size one. Further, reconstructed reasoning traces from the latent information exhibit high semantic similarity to the teacher’s explicit reasoning.
[AI-53] A dataset of formalised theorem statements from the Annals of Mathematics NEURIPS2026 NEURIPS
链接: https://arxiv.org/abs/2610.18301
作者: Thomas Browning,Katerina Hristova,David Ledvinka,Justus Springer,Hang Lu Su,Jack McKoen,Kevin Buzzard
类目: Artificial Intelligence (cs.AI)
备注: Accepted at AI-MATH@NeurIPS, the 6th workshop on mathematical reasoning and AI, NeurIPS 2026
Abstract:Recent progress in autoformalisation of mathematics using AI has been impressive. Measuring this progress requires datasets of formal statements that are both hard and trustworthy. We present AnnalsChallenge: 50 formalised theorem statements written by hand in Lean 4 using Mathlib. Each is the main result of a paper published in the Annals of Mathematics since 2020. Every formalisation was manually reviewed and then audited by AI. We release them without formalised proofs.
[AI-54] Output Discipline Sets the Floor: A Falsification-Guided Model-Agnostic Agent Protocol for Auditable Hypothesis-Driven Decision Making
链接: https://arxiv.org/abs/2610.18298
作者: Jing Liu,Yongxing Qi,Ping Yao,Ting Wu
类目: Artificial Intelligence (cs.AI)
备注: 24 pages, 5 figures, 12 tables
Abstract:What binds large language models on auditable, hypothesis-driven decision tasks, where every candidate can be checked against the evidence? On the public TCMEval-SDT benchmark for Traditional Chinese Medicine syndrome differentiation (200 cases), the binding constraint is output discipline and reasoning structure, not model capability alone: two years of model scaling move zero-shot scores by only -0.03 to +0.08, while output-discipline failures (non-terminating reasoning, format non-compliance, large run-to-run spread) persist even in frontier models. We test this with a model-agnostic agent framework formalizing decision making as a five-stage closed loop, enforced by a candidate x evidence matrix falsification protocol: every candidate is judged against every evidence item, at least half of the candidates must be falsified before a conclusion is licensed, and survivors are emitted in a machine-verifiable single-line format. Across five base models, the effect is sign-conditional on zero-shot output discipline: it rescues models whose discipline fails (DeepSeek-V4-Flash: +0.267/+0.271, entirely termination rescue), reduces a high-variance model’s run-to-run standard deviation from 0.218 to 0.007 (Kimi K3), and significantly costs already-disciplined models (-0.087 to -0.120 on budget-matched Qwen3-4B/32B). Ablations attribute the cost to the candidate-bounded format layer and show the effect is independent of injected knowledge content (protocol-knowledge decoupling). The protocol also transfers to misinformation-robust mathematical reasoning, detecting injected false premises at 89.8% recall with a 6.8% false-positive rate (MCC 0.54; a flag-everything baseline scores 0) and restoring accuracy to the no-misinformation ceiling. Where zero-shot output discipline is weak, the reasoning protocol sets the floor; base-model capability always sets the ceiling.
[AI-55] Which Optimizer Leaks Most? Optimizer Implicit Bias as a Training-Data Privacy Controller
链接: https://arxiv.org/abs/2610.18294
作者: Sagi Habani,Yehonathan Refael,Ofir Lindenbaum
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Releasing a neural network checkpoint can leak information about its training data, and existing reconstruction results rely on gradient-descent training. We ask whether the choice of optimizer affects the ability to recover the training data. Extending work connecting Karush–Kuhn–Tucker conditions to training-data reconstruction, we show that the implicit biases of popular optimizers (SGD, Adam, Muon) leave very different fingerprints of the training data. We argue that the key distinction is the \emphweight-specificity of these implicit biases, namely how much continuous information about the final weights the optimizer-dependent target of the stationarity conditions encodes. The limiting stationarity conditions for models trained with SGD and Muon retain enough weight-specific information to reconstruct training examples in practice. For Adam in the \ell_\infty implicit-bias regime, the optimizer-dependent part of these conditions encodes only a discrete pattern of maximal coordinates and their signs, and at nondegenerate KKT points the conditions are too few to identify the training set to first order. Experiments on MNIST and CIFAR-10 support these claims: models trained with Muon can yield stronger reconstructions at much lower training cost than models trained with SGD, although this advantage depends on the training regime and architecture, and models trained with Adam appear consistently more resistant to the evaluated attacks.
[AI-56] Beyond Answer Sharing: TrueMAD for Targeted Peer Engagement in Multi-Agent Debate
链接: https://arxiv.org/abs/2610.18240
作者: Dezheng Bao,Sihang Du,Renhong Huang,Zhizhang Yuan,Yifei Sun,Yang Yang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Multi-agent debate (MAD) aims to improve large language model reasoning by exposing agents to alternative solutions and enabling iterative revision. However, standard broadcast-style MAD makes peer answers visible without requiring agents to engage with a particular peer claim. As a result, reliable reasoning may be suppressed by conflicting or majority responses, limiting achievable performance, while erroneous reasoning may persist and degrade final predictions without actionable feedback. We propose TrueMAD, a backbone-agnostic test-time plug-in that introduces one-to-one, free-form peer advice between answer broadcasting and self-revision. Each source agent responds to a target’s current reasoning through support, rebuttal, refinement, or further questions, while the target retains control over how to revise. The same mechanism can augment standard MAD and its improved variants without parameter updates. Across AIME, GPQA-Diamond, and TruthfulQA with Qwen3-30B-A3B, TrueMAD improves agent-level and strict-majority accuracy over MAD by 3.62 and 3.52 percentage points on average, while TrueDMAD improves DMAD by 4.72 and 4.06 points, respectively. Analyses at the utterance, agent-transition, and group levels further show more peer-specific deliberation, more correction-oriented revisions, and more favorable consensus dynamics.
[AI-57] Holding Agents Accountable with Verifiable Agent Insurance
链接: https://arxiv.org/abs/2610.18230
作者: Cheng Hua,Taro Tsuchiya,Nicolas Christin,Giulia Fanti
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 21 pages, 1 figure, 7 tables
Abstract:While AI agents are useful for many tasks, they regularly misbehave in destructive ways, such as deleting important data. While prior work has focused on preventing such misbehavior, we consider how to recover after an agent misbehaves. We propose verifiable agent insurance (VAI), a framework for agent providers to commit financially to promises, a list of programmatically-specified policies that an agent is meant to follow. These promises are embedded in a smart contract on an immutable ledger (e.g., a blockchain) to prevent tampering, along with a monetary escrow. A user who detects a violation of committed promises can prompt a verifier to review agent execution traces, which are logged through the agent harness’s shared tool-execution path, without instrumenting each tool separately. Upon verification of the violation, the agent provider’s escrow funds are paid to the reporting user. We instantiate VAI as an end-to-end implementation on an Ethereum layer-2 network, and evaluate its coverage on three prominent agent benchmarks, two of which we re-annotated to obtain task-level policy-compliance labels. Currently, many agents do not emit user authorizations and security-relevant information as explicit, structured data objects – rather, they embed the data needed to verify policy violations in natural language exchanges. Even so, VAI manages to detect violations in 42.2% of runs containing violations across three benchmarks. By augmenting agents to explicitly emit such objects, violation detection rises to 83.0% of traces. Our findings suggest that for improved auditability, agents should be co-designed with their security policies to emit explicit, structured records of user authorization and security-relevant actions.
[AI-58] Reset Early Reset Often Eliminate Models ATC
链接: https://arxiv.org/abs/2610.18204
作者: Martin S. Fixman,Fredrik Rømming,Sean B. Holden
类目: Logic in Computer Science (cs.LO); Artificial Intelligence (cs.AI)
备注: 13 pages, 6 figures, 5 tables. Presented at Vampire 2026. Code: this https URL
Abstract:Connection-tableau provers typically construct closed tableaux by backtracking search. This gives lightweight proof procedures, but their low-memory search state leaves little derived information to reuse after a failed attempt. SATCoP addresses this by retaining grounded tableau clauses in an incremental SAT solver, allowing the prover to refute the problem once the accumulated ground clauses become propositionally unsatisfiable. We introduce SATResetCoP, which relies more directly on this persistent SAT state by not backtracking on unproductive tableaux and instead starting new ones to generate further ground clauses for the SAT solver. On TPTP, SATResetCoP improves over our leanCoP-style baseline by 54% and our SATCoP-style baseline by 18%. On the bushy MPTP2078 benchmark, it solves 616 problems, a 49% improvement over our leanCoP-style baseline and a 12% improvement over our SATCoP-style baseline. These results suggest that changing tableau guidance to accumulate useful ground clauses faster improves the performance of the prover.
[AI-59] How Reliable Are Large Language Models for Soil Organic Carbon Assessment? An Australian Benchmark
链接: https://arxiv.org/abs/2610.18201
作者: Yingxuan Cheng,Tong Li,Yayong Li,Yuchuan Fan,You-Gan Wang,Jinran Wu,Scott Chapman,Geoffrey J. McLachlan
类目: Performance (cs.PF); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注:
Abstract:Soil organic carbon (SOC) is an indicator of agricultural soil condition and carbon storage, but numerical outputs from large language models (LLMs) require validation before use in monitoring or management. We evaluated seven general-purpose LLM configurations on three direct-response tasks: estimating 0–10 cm SOC concentrations for 308 Australian paddocks under two information conditions, calculating SOC stock in 120 matched scenarios, and comparing 300 paddock pairs. References were measured SOC records or deterministic calculations; response validity, repeated-call dispersion and estimated API cost were assessed separately. Additional measured soil attributes reduced concentration mean absolute error by 13–62% across configurations, although high-SOC underprediction and paddock-specific errors persisted. Six configurations were highly accurate in stock calculations using equation-native units without coarse fragments, but accuracy declined with physically equivalent alternative units. Pairwise accuracy was 90.5–94.6% overall and 59.1–70.9% for 52 closely separated pairs; some configurations were sensitive to presentation order. Nearly all outputs were parseable, without necessarily being correct. The benchmark identifies three distinct requirements for AI-mediated SOC assessment: agreement with measured concentrations, unit-consistent stock calculations and reliable discrimination between sites. These findings support task-specific validation and auditable calculations, rather than unverified use of LLM outputs as ecological evidence.
[AI-60] A Few More Tokens Ahead: Reinforcing Parallelism in Diffusion LLM s
链接: https://arxiv.org/abs/2610.18200
作者: Jiahao Kuang,Yangfan Gao,Tao Gui,Yuanbin Wu,Qi Zhang,Xuanjing Huang,Tao Ji
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Parallel generation is widely regarded as an inherent advantage of diffusion large language models. Yet our fixed-quota evaluation of four representative models reveals severe degradation across mathematics and code, with all scores below 15% at 16 tokens per step. Token-level probes identify local predictive potential near the observed prefix, especially in mathematical text. We propose RePara, which reinforces short consecutive token blocks through group-relative policy optimization and reduces rollout and log-probability recomputation costs. On LLaDA and SDAR, \ours improves GSM8K and MATH-500 accuracy over JustGRPO under reduced denoising budgets, with a 1.96x training-step speedup in our LLaDA/MATH setup.
[AI-61] QORH: Learning Query-Specific Ordered Rule Hypergraphs for Inductive Knowledge Graph Link Prediction
链接: https://arxiv.org/abs/2610.18198
作者: Naixing Xu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Knowledge graph link prediction involving unseen entities and relations requires learning transferable reasoning patterns from observed graph structure. Existing methods based on relation graphs primarily capture pairwise dependencies, limiting their ability to preserve the compositional semantics, relation ordering, and role information of complete path rules. To address this limitation, we propose QORH (Query-specific Ordered Rule Hypergraph), which jointly encodes rule heads and ordered rule bodies as hyperedges conditioned on the query relation, explicitly modeling joint dependencies among multiple relations. QORH performs learnable rule selection using statistical evidence, such as rule confidence and coverage, and enhances relation representations through role-aware hypergraph message passing to guide entity-level reasoning. The model’s shared parameters are independent of specific entity or relation identifiers, enabling inductive prediction for unseen entities and relations. Extensive experiments on entity-inductive and fully inductive benchmarks demonstrate that QORH significantly outperforms the state-of-the-art baselines included in our evaluation, validating the effectiveness of rule hypergraph modeling for inductive link prediction.
[AI-62] Neuro-Symbolic Graph Transformers for Hierarchical Intention Recognition from Multimodal Human Activity Sequences
链接: https://arxiv.org/abs/2610.18176
作者: Farnaz Soleimani(LISSI),Abdelghani Chibani(LISSI),Yacine Amirat(LISSI),Ghazaleh Khodabandelou(LISSI)
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Proactive human-robot collaboration requires autonomous systems to infer not only immediate actions but also the underlying hierarchical intentions governing human behavior. Current deep learning approaches predominantly focus on flat action classification, failing to capture the structural dependencies connecting kinematic motion to high-level goals. A Neuro-Symbolic Graph Transformer (NSGT) is proposed within this multimodal framework, a Human Reasoning Graph (HR-G) is integrated into the self-attention mechanism via learnable topological biases. Furthermore, the architecture is augmented with a differentiable First-Order Logic (FOL) loss, acting as a structural regularizer by which ontological consistency is enforced across four levels: atomic actions, composite activities, Low-Level Intentions, and High-Level Intentions. Evaluation is performed on the synthesized NTU-Hierarchy dataset, a four-level ontological extension of NTU RGB+D 120, wherein existing sequences were mapped into 119,383 complex hierarchical episodes (85 actions, 16 activities, 8 LLIs, 4 HLIs). To prevent circular supervision, strict isolation between label generation and FOL constraints was maintained. In the base configuration, 91.65% top-1 action accuracy is achieved by the NSGT. When the FOL loss is incorporated into the baseline, logic consistency is improved by 3.26 percentage points to 96.67%. Through the HR-G biases, a tunable trade-off between accuracy and structural coherence is established. In the primary consistency-maximizing configuration ( \lambda = 0.5), a logic consistency of 94.59% is reached and ontological contradictions are reduced by 33%, while a robust 89.90% top-1 action accuracy is retained. Finally, interpretable reasoning evidence is provided by the learned HR-G weights, wherein transition edges are strongly exploited ( \beta trans \approx +2.6) without explicit supervision.
[AI-63] Looped Reasoning Finishes Earlier Than You Think
链接: https://arxiv.org/abs/2610.18174
作者: Sumin Park,T. Konstantin Rusch
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 25 pages
Abstract:Looped reasoning with dynamic depth can be viewed as relaxation toward solution attractors in a learned dynamical landscape. In practice, however, latent trajectories often oscillate, cycle, or drift rather than cleanly converge to stable fixed points. How these non-stationary dynamics contribute to reasoning and constrain convergence remains unclear. We show that slowly decaying oscillatory modes can impose their convergence rate on coupled latent dynamics, creating a bottleneck. Yet damping these modes indiscriminately can disrupt the reasoning computation they support. This tension motivates an architectural approach: introducing rotational degrees of freedom along which symmetry-generated rotational motion leaves the reasoning-relevant quotient dynamics unchanged by construction. We propose QRR (Quotient Recurrent Reasoner), a gauge-equivariant recurrent architecture with an exact per-token phase-rotation symmetry. Motion along these symmetry directions leaves the quotient representation unchanged and is dynamically decoupled from reasoning-relevant dynamics. Empirically, QRR achieves stable and faster convergence to solution roots in quotient space, without forming a dominant slow oscillatory bottleneck, while matching or surpassing recent recurrent reasoning models on Sudoku, Maze, and Mini-ARC.
[AI-64] Back to the Future: Reverse-Consequence Modeling for Long-Horizon Particle Dynamics
链接: https://arxiv.org/abs/2610.18168
作者: Leyan Zhang,Nantian He,Jiajun Ma,Yanpeng Zhao,Yunbo Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Accurate prediction of object trajectories is fundamental to mechanical understanding and robotic control. However, autoregressive models inherently suffer from cumulative errors, leading to severe drift in long-horizon rollouts. Observing that in many physical systems with dissipative or convergence-to-rest behavior, the final states are often more predictable than transient dynamics (e.g., a sliding object resting under friction or a pendulum stabilizing at equilibrium), we propose ReStep, a cascaded approach that reverses the conventional forward-prediction paradigm. ReStep first employs a stochastic model to directly predict the final state, then leverages Anchored Diffusion to generate goal-conditioned intermediate trajectories. By conditioning on both ‘‘boundary states’’, it effectively constrains dynamics and reduces uncertainty, producing physically plausible and temporally coherent transitions. We further integrate ReStep dynamics as the physical engine within a 3D Gaussian Splatting pipeline, enabling a unified physical-visual framework for intuitive inference from images. Extensive experiments on diverse rigid and non-rigid collision datasets show that ReStep achieves superior long-term accuracy, substantially mitigating error accumulation via reverse-consequence modeling.
[AI-65] When and How Order-Flow Interventions Work After Dealers Adapt
链接: https://arxiv.org/abs/2610.18158
作者: Guanghao Wu,Zhuo Cai,Zhidong Li,Shoujin Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Learning dealer agents can settle on persistently wide quotes, raising what customers pay to trade. A venue can route more orders to tighter quotes, but a wide quote may reflect real risk, and once dealers adapt, an end-to-end comparison cannot show why a rule works: because it intervenes more, picks better moments, is announced in advance, or rewards the right quote. We propose SOFA (State-aligned Order-Flow Allocation), a public order-flow contract with three parts: a State Scorer ranks market states from lagged public data, a Commitment Scheduler announces a budgeted intervention strength before dealers quote, and a Quote-to-Flow Allocator shifts a bounded share of fixed customer demand toward tighter quotes. Our paired re-adaptation protocol gives competing rules the same intervention strengths, only re-timed, lets dealers re-adapt to each rule from a shared checkpoint, and breaks one link at a time. Across 20 independently trained dealer populations in an ABIDES market, concentrating the intervention in states that the score ranks as low-risk lowers low-risk execution cost relative to random timing in every population, by 0.040 tick on average, while a stale announcement or a broken quote-to-flow link raises this cost. In the primary setting the gain comes at the expense of high-risk trades, whose costs rise, so state alignment pays off only for a venue that weighs low-risk costs more than 1.38 times as heavily as high-risk costs. Our code is available at this https URL.
[AI-66] When Do Learned Blocks Preserve Causal Meaning? Block Identifiability and Constructive Causal Abstractions
链接: https://arxiv.org/abs/2610.18155
作者: Karim Zaghw,Wieland Brendel,Patrik Reizinger
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Standard identifiability theory assumes one-dimensional latent factors, but learned representations can encode latent information across multidimensional blocks whose internal coordinates are not uniquely determined. When such blocks are used to represent and intervene on higher-level concepts, it is unclear whether their causal meaning survives the transformations left unresolved by block-identifiability. We connect block-identifiability to constructive causal abstraction. We show that block-diagonal invertible transformations followed by compatible coordinate permutations form the maximal class that preserves every constructive causal abstraction built on those blocks. This guarantee assumes that the correspondence between blocks and concepts is updated with the representation. A small synthetic experiment with additive autoencoders shows that high-level interventions can be transferred between independently learned representations while approximately preserving their intended effects.
[AI-67] RadixFlow: Filling Execution Bubbles in Prefix-Cached Hybrid LLM Serving
链接: https://arxiv.org/abs/2610.18143
作者: Huihao Jing,Wenbin Hu,Shaojin Chen,Haochen Shi,Haoyu Huang,Sirui Zhang,Yuxuan Liu,Zhongwei Xie,Haoran Li
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Prefix reuse improves LLM inference by sharing the KV cache of common prompt prefixes. This helps only if the scheduler can execute a matched branch in the current worker round. Schedulers assume that a matched prefix is ready, which holds for full attention but fails for hybrid LLMs. There, a branch (the consumer) may need recurrent state that another branch (the producer) is still computing in the same round. The native scheduler stops its waiting-queue scan at this blocked consumer, and the resulting head-of-line blocking creates \emphexecution bubbles: idle batch slots that independent work could fill. We introduce RadixFlow, which decouples prefix matching from state readiness. Its state-ready traversal turns a global scan barrier into a local wait: only the consumer is blocked until the next round, while feasible independent branches keep filling the batch under the native state check. On five hybrid models, RadixFlow lowers mean latency by 3% to 24% and worker rounds by 11% to 44% at almost unchanged reuse. On the larger Qwen3.5 models, ablations show that most of the gain comes from traversal, which adds little on a full-attention control. Heavier rounds limit how much of this saving reaches latency.
[AI-68] Mitigating Cross-Modal Hallucinations by Correcting Modality Misrouting in Omnimodal LLM s
链接: https://arxiv.org/abs/2610.18141
作者: Yu Mi,Licheng Pan,Jiayi Zhu,Qirui Liu,Yitong Qiao,Changtong Xu,Tao Wei,Fei Sun,Zhixuan Chu,Kui Ren
类目: Artificial Intelligence (cs.AI)
备注: 21 pages, 10 figures
Abstract:Omnimodal large language models (OLLMs) are susceptible to cross-modal hallucinations under audio-visual interference. We find that, for a substantial fraction of incorrect joint predictions, the question-relevant unimodal branch can still recover the correct answer, suggesting that task-relevant evidence can remain recoverable from the relevant modality even when it is not reflected in the joint prediction. We refer to this recoverable failure pattern as \emphmodality misrouting. We propose CaRe (Conflict-aware Residual Intervention), a training-free inference-time method for identifying and correcting modality misrouting. CaRe uses a text-only modality probe to identify a candidate modality-specific counterfactual branch, and flags potential misrouting when its prediction conflicts with the original joint prediction. For such cases, CaRe computes a residual direction from the discrepancy between the counterfactual and fused representations and uses it to steer the original multimodal state toward the relevant modality evidence, while retaining both modalities in the original audio-visual inference path. The intervention strength is adapted based on the confidence of the counterfactual branch. Experiments on CMM and AVHBench across two OLLMs show that CaRe consistently outperforms the evaluated inference-time hallucination-mitigation baselines.
[AI-69] Certifying Almost-Sure Reachability in Unknown MDPs
链接: https://arxiv.org/abs/2610.18119
作者: S Akshay,Suguman Bansal,Gaurav Mahla
类目: Logic in Computer Science (cs.LO); Artificial Intelligence (cs.AI)
备注:
Abstract:\em Specification-Guided Reinforcement Learning (RL) has largely concerned itself with \emphlearning an optimal policy against a formal specification in an unknown \em Markov Decision Process (MDP). We take up the dual question. Given a policy already in hand, can one \emphverify that it is correct against a formal specification in an unknown MDP, using sampling access alone? Such a question arises naturally in safety-critical systems. We investigate it for the \em almost-sure reachability property, whose correctness is witnessed by a \em ranking supermartingale (RSM) certificate. Our first finding is negative. In the absence of any further information about the MDP, one can neither learn such a certificate nor verify a candidate one with probabilistic guarantees. On the positive side, we show that, under a mild restriction, a known lower bound p_\min on nonzero transition probabilities, a correct certificate \emphcan be learned with PAC-style guarantees. To the best of our knowledge, this is the first study of learning and verifying certificates for almost-sure reachability in unknown MDPs from sampling access alone. Finally, our experiments show that the p_\min -based sufficient condition, while sound, is conservative: the learned certificate often remains valid on the true MDP even when the condition fails.
[AI-70] When Missingness Follows the Graph: Beyond Feature Propagation for Missing Features
链接: https://arxiv.org/abs/2610.18113
作者: Enrico Chialà,Francesco Ferrini,Antonio Longa,Bruno Lepri,Filippo Maria Bianchi,Manfred Jaeger,Veronica Lachi
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 29 pages, 3 figures
Abstract:Graph Neural Networks (GNNs) commonly assume that node features are fully observed, despite missing features being pervasive in real-world applications. By leveraging the dependence between node features and graph structure, existing graph-specific methods handle missingness by propagating observed feature values along graph edges. However, these methods have mainly been evaluated in overly optimistic settings, where feature entries are removed independently and uniformly across the graph. This overlooks realistic scenarios in which missingness propagates among connected nodes, as can occur during localized sensing failures in sensor networks. We formalize this complementary notion as graph-structured mask dependence, where the joint distribution of missingness indicators is shaped by the graph, and introduce a controlled mechanism that instantiates this setting. We study propagation-based methods theoretically and empirically in this setting, showing that their effectiveness changes substantially compared with the uniformly random regime. Motivated by this limitation, we propose PEMix, designed to remain effective under challenging graph-structured missingness. Rather than propagating observed values, PEMix uses global structural representations to condition a probabilistic model of incomplete node features before message passing. Experiments under controlled graph-structured missingness and naturally occurring missingness show that PEMix remains robust across complex missingness regimes, with particularly strong performance as missingness becomes severe.
[AI-71] A CaLLM Judge is a Better Judge: Mitigating Bias in LLM -as-a-judge
链接: https://arxiv.org/abs/2610.18108
作者: Cesare Barbera,Luca Stradiotti,Lorenzo Cascioli,Lorenzo Perini,Niek Tax,Daniel Haimovich,Jesse Davis,Andrea Passerini,Andrea Pugnana
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) are increasingly used as judges to predict which of two candidate answers a human evaluator would prefer, providing a scalable alternative when human evaluation is too costly. Unfortunately, LLM judgments can exhibit systematic biases. Prior work has documented position, verbosity or model family of the candidate answers influencing the judge decision rather than their quality, while additional biases may remain undocumented. \textitMulticalibration provides a natural framework to address this problem: each bias corresponds to miscalibration on an identifiable group, \textite.g. , family bias manifests as overconfidence in answers generated by models belonging to the same family of the judge. The challenge is to fix miscalibration simultaneously across many, potentially unknown groups. To this end, we introduce \mathrmCaLLM , a post hoc framework that multicalibrates an LLM judge, thereby bounding all such biases simultaneously. We prove that \mathrmCaLLM bounds the bias of every group whose error structure is captured by an available representation, and that augmenting it with explicit bias groups further reduces the residual bias. We then derive theoretical conditions under which \mathrmCaLLM improves the judge’s accuracy, rather than trading it for calibration. Finally, we evaluate \mathrmCaLLM across pairwise preference benchmarks, showing that it reduces the multicalibration error on both specified and unspecified bias groups, while improving predictive performance of the judge.
[AI-72] LifeSearch: Benchmarking Deep Research Agents for Real-World Life and Productivity
链接: https://arxiv.org/abs/2610.18092
作者: Yibo Wang,Senjie Jin,Yansong Ning,Xinbing Liang,Haotian Luo,Jiazheng Zhang,Yuxuan He,Will Wu,Clarenceai
类目: Artificial Intelligence (cs.AI)
备注: Preprint
Abstract:Deep research (DR) agents address open-ended information needs through multi-step reasoning, long-horizon web search, and evidence synthesis. However, existing benchmarks primarily focus on academic research or synthetic tasks, leaving DR agents’ capabilities in real-world life and professional productivity settings insufficiently evaluated. To bridge this gap, we introduce LifeSearch, a challenging benchmark comprising 203 Chinese and English tasks derived from real-user query seeds across 7 concrete life and productivity domains. With more than 4,500 hours of human annotation, we construct over 4,000 expert-written, fine-grained rubrics, covering both report-content criteria for coverage and accuracy and concrete criteria specifying values, units, dates, and URLs. We conduct extensive experiments on 12 advanced DR agents under a unified evaluation harness, systematically analyzing their performance patterns and failure modes. We find that even the best-performing frontier agent, GPT-5.6 Sol, achieves a score of only 63.4%, leaving substantial room for improvement on realistic life and professional DR tasks. We select the grading configuration by comparing candidate LLM judges against human rubric labels, and a human ranking study shows strong alignment between aggregate benchmark scores and human judgments (Spearman’s \rho = 0.89 , Kendall’s \tau = 0.81 ). LifeSearch provides a rigorous evaluation framework and foundation for advancing DR agents toward real-world life and productivity.
[AI-73] AM-Bench: Understanding Agents under Task Underspecification across Models and Harnesses
链接: https://arxiv.org/abs/2610.18090
作者: Kaiyu Huang,Yiting Li,Angyang Gu,Pinjia He,Minchen Yu,Qingjiang Shi
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Ambiguous or incomplete task specifications complicate both execution and clarification for language-model agents. Yet agents’ responses to these information gaps remain poorly characterized across models and execution harnesses. We present AM-Bench, a benchmark built from 48 tasks adapted from Terminal-Bench 2.0 and Harness-Bench. Each task includes a fully specified reference and six variants spanning ambiguous or missing goals, context, and constraints. We evaluate every pairing of five models and four harnesses over 20,160 runs. Our results reveal distinct execution deficits: goal gaps hinder completion, context gaps increase token use, and constraint gaps weaken compliance without necessarily preventing completion. Missingness lowers success more than ambiguity in all three domains. These effects also reshape system comparisons: advantages under complete information need not persist, and rankings depend on model \times harness pairings and the evaluation criterion – completion, success, or repeat consistency. Estimates based on observed clarification requests and matched fully specified runs further suggest that clarification could recover unfinished work while leaving most constraint violations unresolved. Together, these findings motivate joint model-harness evaluation across information conditions and targeted clarification of unresolved requirements before dependent actions. To support future research on agents under task underspecification, we provide the AM-Bench project repository at this https URL.
[AI-74] One-Sided Verification Under Limited Expert Review
链接: https://arxiv.org/abs/2610.18089
作者: Fan Nie,Haotian Ye,Shurui Liu,Zihao Wang,Stefano Ermon,James Zou
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:AI has become increasingly helpful in scientific discovery, yet human expert review remains essential for those problems without reliable automated evaluation, posing a fundamental challenge as candidate generation scales beyond available review capacity. This paper studies verification under limited human effort, where rare valid solutions need to be prioritized while invalid candidates should be filtered out to reduce human workload. We adopt the one-sided verification objective that maximizes the rejection rate of invalid solutions (TNR) under a constraint on the false-rejection rate of valid ones (FNR), making human review cost an explicit term to minimize. Drawing on the intuition that verifying a specific error can be easier than discovering one, we introduce OSVerify, an algorithmic framework that decouples error proposal from error validation. The error proposer searches for errors in solutions across different granularities by hierarchical decomposition, while the validator distinguishes substantial errors from minor or hallucinated ones. Experiments on competition- and research-level benchmarks show that OSVerify reduces the number of human reviews of invalid solutions by 46% on average compared to voting baselines under the same empirical low FNR constraint. Our post-training of the validator further improves the reduction to 57%. As high-impact real-world use cases, we apply OSVerify to identify errors in published research papers that are subsequently confirmed by the original authors and help resolve three Langlands open conjectures that are confirmed by experts, demonstrating its utility to facilitate human-AI research collaboration.
[AI-75] ResearchFlow-Eval: Tracing the Execution Gap in End-to-End Multi-Claim Autoresearch
链接: https://arxiv.org/abs/2610.18088
作者: Peng Jiang(Tsinghua University),Dongyang Li(Tsinghua University),Quanying Liu(Southern University of Science and Technology),Xiaoxuan Jia(Tsinghua University)
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 24 pages, 6 figures
Abstract:Autoresearch systems increasingly undertake entire research projects. In practice, scientific research requires systems to pursue multiple interdependent claims through a continuous, multi-stage workflow and revise experiments as evidence accumulates. Existing evaluations of autoresearch systems, however, typically score isolated tasks or final artifacts, leaving these process-level capabilities largely unmeasured. We therefore introduce ResearchFlow-Eval, a fine-grained framework for assessing continuous scientific execution and iteration at both the stage and claim levels. Each evaluation is grounded in a published anchor paper, from which we derive stage-specific rubrics that decompose each research stage into dimensions and separately scored rubric items. Using these rubrics, we score each research stage and examine experimental execution in detail, focusing on multi-claim completion and revisions across attempts. We apply the evaluation framework to general-purpose agent harnesses, skill-augmented harnesses, and systems with research-specific orchestration. Across all evaluated systems, experimental execution consistently lags behind design and planning. Within execution, systems struggle to produce high-quality results and meet claim requirements, and repeated attempts rarely yield measurable improvements. Moreover, neither skill augmentation nor research-specific orchestration reliably yields higher scores. ResearchFlow-Eval thus extends autoresearch evaluation beyond isolated, single-objective runs to continuous workflows in which systems pursue multiple claims and refine their work across iterations.
[AI-76] AReaL-RSI: Adaptive Scaling and Coordination of Recursive Self-Improvement
链接: https://arxiv.org/abs/2610.18083
作者: Wenhao Zhou,Jiajun Song,Haitao Wang,Zipeng Qiu,Yingqi Peng,Ruida Xu,Jiale Li,Wenshuang Li,Shusheng Xu,Wei Dong,Hao Dai,Tongkai Yang,Yi Gao,Binhang Yuan
类目: Artificial Intelligence (cs.AI)
备注: 12 pages
Abstract:Scaling recursive self-improvement requires additional computation to produce useful research progress in reaching persistent objectives. Intuitively, more researcher agents can explore more hypotheses, but early evaluation scores might misrepresent their promise, shared findings could lose their applicability conditions, and useful procedures could remain confined to completed investigations. In this paper, we present AReaL-RSI, a framework for organizing a persistent research team of agents without updating backbone LLM weights. We formulate the scaling problem in terms of evaluated outcomes and whole-system resource expenditure. Concretely, a leader agent selects task-appropriate research orientations and guides ongoing investigations; conditional memory sharing distinguishes unresolved problems from supported findings and enables recipient adaptation. During recursive evolution, after each campaign, evidence-based review revises researcher orientation, knowledge, and skills for later campaigns. Experiments compare AReaL-RSI with Codex-Goal on downstream harness optimization, GPU kernel optimization, and AI research tasks using two backbones. Under the reported resource allocations, AReaL-RSI improves Terminal Bench performance by 2.2 percentage points for each backbone and increases mean KernelBench speedup by 32.2-36.8%; further scalability comparisons favor the design of AReaL-RSI over native scaling: combining experience sharing with specialized researcher roles yields 1.9-27.5% lower kernel latency than independent exploration and 1.3-33.9% lower latency than sharing alone across all seven comparable task-backbone pairs.
[AI-77] WorkRover: Scaling Long-Horizon Cowork Agent Training Via Recursive Environment Evolution KR
链接: https://arxiv.org/abs/2610.18067
作者: Hang He,Linting Wang,Guanming Liu,Bo Mao,Jinxiu Liu,Yiran Suo,Maosen Zhou,Zhiheng Xi,Guanting Dong,Zhou Shao,Rui Zheng,Jie Zhou,Chengcheng Wan,Tao Gui,Qi Zhang,Xipeng Qiu,Xuanjing Huang
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Project page: this https URL
Abstract:Scaling interactive environments is important for training long-horizon agents through reinforcement learning. These environments often require further testing and refinement after initial generation, and correcting defects in a task may also require updating its work context and verifier. However, existing construction pipelines offer limited support for such iterative refinement while keeping these components aligned. We introduce WorkRover, a framework for training long-horizon cowork agents through recursive environment evolution. Its harness-driven synthesis enables construction agents to jointly build and validate semantically aligned work contexts, executable tasks, and evidence-grounded verifiers, adapting their actions to workspace state and execution feedback rather than following a fixed construction workflow. To support environment evolution, we introduce Factorial Compatibility Replay (FCR), which uses rollout feedback to evaluate combinations of component revisions through dependency-closed replay and selects the smallest validated repair among tested candidates while preserving task requirements and evaluation constraints. To provide finer-grained training signals, we propose Verifier-to-Turn Credit Assignment (VTCA), which augments trajectory-level GRPO advantages with signals from deterministic workspace-progress checks and assigns a bounded share of positive-progress credit to prerequisite turns using recorded tool-output and artifact dependencies. Using WorkRover, we post-train Qwen3.5-35B-A3B to obtain WorkRover-35B-A3B and evaluate it on several cowork-agent benchmarks. Its reported scores exceed those of Qwen3.5-35B-A3B by 26.27 percentage points on JobBench, 29.32 percentage points on OfficeQA Pro, and 413.25 Elo points on GDPval v2. These results suggest that WorkRover offers a promising foundation for scalable training of long-horizon cowork agents.
[AI-78] Uncertainty Evaluation: Bridging Machine Learning and Measurement Science
链接: https://arxiv.org/abs/2610.18066
作者: Dario Petri,Luca Mari
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Reliable uncertainty evaluation is essential for assessing the trustworthiness of machine learning (ML) predictions, particularly in safety-critical applications. This paper explores conceptual analogies between uncertainty evaluation in ML and measurement science, drawing on the evolution of metrology from error-based approaches to the framework established by the Guide to the Expression of Uncertainty in Measurement (GUM). Particular attention is devoted to the distinction between process-level and result-level quality and to the limitations of the conventional classification of ML uncertainty into aleatoric and epistemic components. It is argued that this dichotomy, similarly to the traditional distinction between random and systematic measurement errors, is inherently model-dependent and provides an insufficient basis for a unified uncertainty evaluation framework. Further analogies between machine learning and measurement field of knowledge are identified concerning distribution shifts, anomalous operating conditions, and the validity of uncertainty evaluation. Building on the proposed considerations, the paper advocates a metrology-inspired approach based on the systematic identification, quantification, and propagation of uncertainty contributions. By extending the methodology already established in measurement science to ML, this approach could offer a potentially rigorous, consistent, and transferable framework for evaluating the trustworthiness of individual predictions.
[AI-79] From Policy Evaluation to Off-Policy Control: Provable Emergence of In-Context Q-Learning
链接: https://arxiv.org/abs/2610.18057
作者: Yingjun Mo,Yajie Zhao,Fang Kong
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Transformers can implement reinforcement-learning algorithms in context, yet how off-policy control emerges within Transformers remains poorly understood. We study Q-learning, a foundational model-free control algorithm that combines off-policy bootstrapping with greedy policy improvement. We prove that a two-layer attention-only Transformer can implement a finite-temperature form of Q-learning, where the temperature \tau0 controls the approximation to greedy action selection, recovering window Q-learning as \tau\to0 . We train the structured Transformer on randomly generated MDP tasks using output-only supervision. Experiments show that the theoretically predicted parameter structure emerges during training, including the internal greedy action-selection mechanism, and that the trained Transformer achieves performance comparable to Q-learning on unseen tasks. Our theoretical analysis further shows that the equivalent realizations form a solution manifold and that, under suitable task-diversity conditions, the training iterates converge locally to this manifold at an exponential rate. For the random-task model analyzed here, the local worst-case number of gradient steps needed to reduce the distance to the manifold by a factor \epsilon is \Theta\left(\tau^-5\log(1/\varepsilon)\right) , quantifying the increasing optimization cost as the learned update approaches window Q-learning.
[AI-80] Can Agents Forget When Asked? A Benchmark for Memory Unlearning in LLM Agents
链接: https://arxiv.org/abs/2610.18052
作者: Bofan Xie,Chaokun Wang,Guanchen Ge,Songyao Wang,Fang Niu,Hang Zhang,Wei Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Long-term memory has become a core component of large language model (LLM) agents. As agents accumulate personal information over long-term use, users increasingly ask them to delete sensitive or outdated facts. A memory system should honor such requests. It should remove exactly what was requested and keep the rest of its memory intact. However, existing memory benchmarks evaluate only memory writing, retrieval, and organization, so whether agents can forget at the user’s explicit request remains under-evaluated. To address this gap, we propose Bemula, the first benchmark dedicated to memory unlearning. It comprises 50 multi-session dialogue instances, 135 forget requests of six types, and 1,742 test questions. Each dialogue has a control version without the requests, so that only systems that have stored the target can earn credit for forgetting. Across the paired versions, five test suites measure storage, removal of the requested information, preservation of unrelated memories, relearning of similar facts, and resistance to extraction attacks. On Bemula, we evaluate six mainstream memory systems under three backbone LLMs, together with the Full Context LLM on two of them. The results are negative. Across all evaluated systems and backbones, no system passes more than 1.5% of forget requests end to end, and most pass none at all. These systems either leave the requested information extractable under simple attacks or delete far more than requested. Memory unlearning therefore remains an open challenge for agent memory design. We release our data and code at this https URL.
[AI-81] Beyond Black Boxes: Visual Grounding for Text Embeddings
链接: https://arxiv.org/abs/2610.18047
作者: Slimane Larabi
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:In this paper, our goal is to represent text embeddings as sketches for three main reasons. First, to visualize text representation and to explore their potential their potential for AI explainability. Second, to investigate the potential benefits of the sketch-based representation for information retrieval. Third, to explore their potential for reasoning, drawing inspiration from the concept of mental imagery. We propose a sketch-based visual grounding framework that transforms text representations into visual topographic sketches. Our approach aims to emphasis the semantics content encoded in the text embeddings through the corresponding sketch. Evaluations are conducted on the MS-COCO dataset to examine the utility of these sketches for text retrieval. The comparative study of SBERT and Sketch-based method reveals that SBERT retains an advantage in strict top-1 retrieval accuracy while achieving comparable performance to the sketch-based method in top-5 retrieval. A cascaded approach combining SBERT and the sketch-based method outperforms SBERT alone, suggesting that the Sketch-based representation could offer significant potential for information retrieval if trained directly on text. Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2610.18047 [cs.AI] (or arXiv:2610.18047v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2610.18047 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-82] ChampionKV: Head-Wise Adaptive Budget Allocation for KV Cache Compression
链接: https://arxiv.org/abs/2610.18026
作者: Yuechuan Chen,Yanda Guo,Xiaoxi Liu,Weilin Ruan,Huanbo Xu,Ruyi Wang,Wei Zhu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Some observation-window-based KV cache compressors use prompt attention to identify important tokens but allocate uniform or predefined budgets across heads, overlooking input-dependent differences in budget demand. We propose ChampionKV, an input-adaptive head-wise budget allocator for long-context KV cache compression. It uses the maximum aggregated observation-window attention score as a lightweight empirical proxy for attention concentration, assigning larger budgets to heads with dispersed attention and smaller budgets to concentrated heads. An abnormal-head mask and a floor-budget mechanism improve robustness under weak attention responses. ChampionKV plugs into compressors such as SnapKV and PyramidKV without changing their token-selection procedures. On LongBench and RULER, both Champion variants consistently improve over their SnapKV and PyramidKV base compressors; the improvements are largest and statistically significant on Llama3.1-70B-Instruct, especially under tight budgets, whereas results on the tested 7B/8B models remain competitive but mixed. ChampionKV does not uniformly outperform Ada-KV, whose global cross-head Top-K selection retains an advantage in several settings. Controlled analysis further shows improved retained attention mass over the corresponding base compressors at negligible overhead.
[AI-83] SHELDON: Reasoning through Dual Search over Hypothesis and Observation Spaces
链接: https://arxiv.org/abs/2610.18025
作者: Harshit Joshi,Priyank Shethia,Monica S. Lam
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 16 pages main text, 34 total. Preprint
Abstract:Test-time scaling has enabled LLMs to solve increasingly complex reasoning, discovery, and optimization tasks. However, existing approaches often gather observations while seeking evidence for current hypotheses, then generate new hypotheses conditioned on those hypotheses and their associated evidence. This couples observation discovery to existing hypotheses, favoring local refinements over exploration of unknown hypotheses regions. Inspired by the hypothetico-deductive method (Hempel, 1945) and dual-space models of scientific discovery (Klahr Dunbar, 1988), we introduce SHELDON, a scientific reasoning harness that separately represents and jointly explores hypothesis and observation spaces. By actively seeking new observations along with new hypotheses, SHELDON can discover observations that its current hypotheses do not explain. These unexplained observations expose gaps in its understanding, motivate new investigations, and guide the generation of hypotheses beyond refinements of existing ones. Across diagnosis, abstract reasoning, and optimization benchmarks, SHELDON outperforms evaluated baselines by up to 26.7 points. Additional analysis shows that SHELDON generates up to 4.5x more novel solutions.
[AI-84] Correcting or Continuing? What On-Policy Distillation Learns from Student Errors
链接: https://arxiv.org/abs/2610.18011
作者: Shuo Yang,Zheyu Zhang,Mahito Sugiyama,Gjergji Kasneci
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 10 pages
Abstract:On-policy distillation (OPD) trains a student on its own rollouts under a frozen teacher, mitigating the distribution mismatch of off-policy supervision. While prior work has primarily trained on problems a student can largely solve, we ask whether OPD can improve a student when training is restricted to problems it currently fails. We evaluate five representative OPD mechanisms with Qwen-3 and Phi-4 models under a unified competition-mathematics protocol. We find that OPD does not improve accuracy on this hard subset, with most mechanisms instead degrading performance despite continued decreases in training KL divergence. All eight configurations show a decline, and the two-sided sign test gives p=0.0078. In contrast, the same pipeline improves accuracy by 4.6 percentage points when training is not concentrated on failures, showing that the failure is conditional on the training regime rather than systematic. These findings suggest that OPD is effective when supervision is dominated by correct student prefixes, but can reinforce existing errors when training is concentrated on problems the student currently fails.
[AI-85] Cost-Efficient Privacy Enforcement for Email-Generating LLM Agents under Strategic Information Requests
链接: https://arxiv.org/abs/2610.18007
作者: Qikun Cai,Tao Huang,Chen Hou
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 24 pages
Abstract:LLM agents now act as delegated communication intermediaries: they read private context, draft messages, and send externally visible emails on behalf of a user. Existing checker-repair defenses can reduce inappropriate disclosure, but they rely on repeated calls to a remote action LLM (the model that drafts and repairs the email) and a remote checker LLM whenever a draft fails validation. This creates a cost problem as well as a privacy problem, especially when a requester strategically asks for exact values, raw records, source evidence, or other details that induce leakage-heavy drafts. We propose local contract-conditioned rewriting as a cost-efficient enforcement layer for email-generating agents. The mechanism intercepts an outgoing draft, constructs an information-release contract that specifies what to preserve, generalize, and forbid, rewrites the draft with a local Gemma rewriter, and selectively invokes a final PrivacyChecker. Across extensive email tasks under normal prompts and attack prompts that operationalize strategic information requests, Gemma + Checker reduces normal-prompt leakage from 11.41% to 4.03% while improving UsableRate (satisfied or partially satisfied outputs) from 85.2% to 92.6%. Under the strategic attack-prompt condition, it reduces leakage from 28.19% to 18.79% and reduces checker-only cost amplification from 1.32x to 1.02x. Offline replay ablation shows that the full contract reduces leakage relative to generic Gemma rewriting by 36.3% under normal prompts and 54.1% under attack prompts. These results support a local-first privacy enforcement design: move repair to a cheaper trusted local layer when possible, but retain the checker as the final semantic gate for high-risk sends.
[AI-86] SWEPatchTest: Evaluating Software Tests Beyond Reproduction
链接: https://arxiv.org/abs/2610.18004
作者: Hua Lu,Pengcheng Wang,Han Li,Zichen Tian,Yanyu Chen,Huang He,Bingjin Chen
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:Coding agents use generated tests to decide whether a proposed patch satisfies an issue’s requirements. A suite that fails on the buggy program and passes after applying the reference patch can still accept an incorrect patch or reject a correct patch. We introduce SWEPATCHTEST, a test-generation benchmark with 619 tasks from SWE-bench Pro, SWE-bench Multilingual, and DeepSWE and 82,579 candidate patches collected from coding-agent trajectories. By replaying each suite on the buggy and patched codebases, \benchmark jointly measures issue reproduction, acceptance of correct patches, and rejection of incorrect patches. Joint Validation Rate (JVR), which requires all three criteria on the same task, exposes failures hidden by single-reference evaluation: the best-performing configuration achieves only 9.21% JVR despite reproducing the reference transition on 59.29% of tasks. We further use validation scores to select and scale demonstrations for supervised fine-tuning (SFT). The resulting training recipe approximately triples JVR and improves downstream repair across three SWE-bench benchmarks. Patch discrimination thus provides both a diagnostic measure of test quality and an execution-based signal for curating agent training data.
[AI-87] Seeing Depth but Choosing Cues in Vision-Language Models
链接: https://arxiv.org/abs/2610.17976
作者: Zhengtao Zou,Pekka Marttinen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Do vision-language models use depth information to judge which object is closer, or rely on image-plane cues? We investigate this question through controlled depth–cue conflicts, representation analysis, and interventions. Among the many image-plane cues relevant to depth perception, we focus on apparent size and vertical position. We compare depth judgments under cue-consistent and cue-conflict conditions in both rendered scenes and real-world photographs. Across model families and scales, depth judgments systematically shift toward misleading cues. Yet the correct depth order remains decodable from the queried objects’ visual representations, including when the model answers incorrectly. Image and category controls support interpreting this signal as image-dependent depth information beyond the two target cues. Strengthening the model’s own depth readout through object-level write-back can correct conflict decisions without updating model parameters or using ground-truth depth annotations. Interventions show that the effect of this correction depends on the direction and target tokens of the written signal, while its strength varies across architectures and domains. Together, these findings reveal a mismatch between a model’s ability to represent depth and use this information when deciding the answer. VLMs can retain useful depth information while allowing competing image-plane cues to exert greater influence on their answers.
[AI-88] BELT : BEhavior-Aligned Low-Rank KV Cache Compression with Learned Top-k Rank Allocation
链接: https://arxiv.org/abs/2610.17966
作者: Yaoyang Liu,Shuwei Shen,Jincheng Tang,Jinguang Liu,Jinbao Xue,Hailong Yang,Ke Zhang,Yinjun Wu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Long-context inference makes the key-value (KV) cache a major memory bottleneck for large language models. We propose BELT, a low-rank KV cache compression framework. Its key-side objective targets attention distributions rather than raw keys or pre-softmax logits, yielding a quadratic surrogate with a closed-form low-rank solution. A learned global rank allocator uses differentiable soft selection during training and hard selection at deployment to meet a user-specified component budget. Experiments demonstrate improved model quality under aggressive compression and increased decoding throughput at comparable accuracy. Evaluated across three LLMs and multiple benchmarks, BELT achieves up to 70% low-rank KV cache compression while outperforming the strongest baseline by up to 23.3 points on LongBench, and up to 17.8x overall KV cache reduction when combined with quantization. With Triton-based GPU kernels and a quantized cache, BELT delivers up to 2.7x speedup for the attention module and 3.3x end-to-end decoding throughput.
[AI-89] Better Capturing Interactions Between Products in~Retail: Revisited Negative Sampling for~Basket Choice Modeling
链接: https://arxiv.org/abs/2610.17963
作者: Jules Désir,Vincent Auriau,Martin Možina,Emmanuel Malherbe
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Brick-and-mortar retailers face many different challenges that involve understanding thoroughly its products catalog and customer preferences. In particular, assortment optimization - proposing the ideal mix of products - and promotion planning hold a pivotal role in their strategy. By leveraging sales data, retailers can make informed decisions on which products to sell and how to manage inventory, based on customer preferences as well as regional and seasonal trends. It is especially crucial to capture interactions between products, in order to minimize the number of items that cannibalize each other’s sales and to ensure that complementary products, which are often purchased together, are conjointly available and sold across all stores. In this paper, we propose a model of shopping basket that learns embeddings to represent interactions between products, prices and stores. Our model is built to uncover sales patterns from very large transaction datasets. In particular, the optimization loss is computed with random negative samples in order to overcome the computational bottlenecks that arise with large number of items. Our experiments on synthetic data show the efficiency of drawing such negative samples based on the actual assortment of available products, with better results than approaches from the literature. We also validate our approach by training and evaluating our model on a dataset composed of millions of transactions from Fortenova Group, one of the leading retail company in Southeast Europe. Our model showcases promising applications in the sector of retail, with enriched interfaces to efficiently support category managers.
[AI-90] Learning How to Remember: Memory Modality Selection for GUI Agents
链接: https://arxiv.org/abs/2610.17928
作者: Taeyun Roh,Suhyeong Park,Gijin Kim,Wonjune Jang,Jaewoo Kang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Recent advances in vision-language models (VLMs) have substantially improved the capabilities of GUI agents, while memory augmentation has emerged as a promising approach for reusing knowledge from past interactions. However, existing memory augmented GUI agents typically adopt a fixed representation strategy, providing retrieved experience as a static modality. It remains unclear whether a single memory modality is consistently optimal, or how modality preference changes across interaction steps. In this work, we conduct a systematic study of memory modality in VLM based GUI agents. Across GUI benchmarks, we find that the preferred memory representation is strongly step dependent. Motivated by the finding, we introduce \textbfSwitchMem, a routing framework that dynamically selects between textual and visual memory representations at each step. Across three GUI benchmarks, SwitchMem outperforms memory augmented baselines that rely on static memory representations. Our findings validate our claim that memory modality is a step dependent decision rather than a fixed design choice. Code and project resources are available at this https URL
[AI-91] GLST: Global-Local Spatio-Temporal Memory Decomposition Framework for Online Chained-Goal Navigation
链接: https://arxiv.org/abs/2610.17927
作者: Yousef Liang,Ryan Ardalan,Yarden Goraly,Chenhao Zhu,Yue Yin,Jonas Frey,Claire Tomlin,Marco Pavone,Shankar Sastry
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: Accepted at the Conference on Robot Learning (CoRL) 2026
Abstract:Autonomous navigation in real-world environments often requires long-horizon operation under partial observability and evolving goals. Existing mapless navigation approaches predominantly assume fixed goals throughout an episode, limiting applicability to practical scenarios where goals change online during execution. We study online chained-goal navigation, where goals evolve through existing-goal redirection or new-goal emergence without explicit transition signals. Under this setting, we find navigation may require either continued exploration or strategic revisitation, creating conflicting memory requirements. We propose Global-Local Spatio-Temporal Belief Learning (GLST), a memory-decomposition framework that separates latent beliefs into global-local spatial and long-short temporal representations for long-horizon reasoning. We further introduce goal-conditioned rewards to balance exploration and exploitation under evolving goals. Extensive experiments in simulation across diverse terrains demonstrate improved chained-goal navigation performance over baselines. Real-robot experiments using the Spot quadruped further validate our memory decomposition framework.
[AI-92] SpecCloak: Probing Harmful Code Generation through Engineering Specifications
链接: https://arxiv.org/abs/2610.17907
作者: Wenbin Chen,Jiexin Wang,Guanting Wu,Jiaxi Liang,Yi Cai
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: 9 pages, 7 figures
Abstract:The code generation capabilities of large language models (LLMs) can equally be misused to implement harmful functionality, underscoring the need to understand their safety behavior in response to programming requests. While request phrasing is known to influence refusal, it remains unclear whether the form in which coding task requirements are stated affects the refusal process for malicious code generation. We introduce PseudoSpec, a template that organizes functional requirements as engineering specifications, and investigate its effects through controlled comparisons, representation analysis, and activation interventions. Across five open-weight models, removing explicit risk keywords only partially reduces refusal, whereas specification prompting substantially lowers refusal even when these keywords are retained. This reduction accompanies consistent shifts toward code-output-related representations, and interventions along condition-difference directions alter refusal behavior. Output analysis further distinguishes code-related responses from generation degeneration. Building on these findings, we propose SpecCloak, a black-box red-teaming framework combining PseudoSpec with discriminator-guided optimization, achieving an average execution-verified attack success rate of 28.91% across eight target models on RedCode-Gen. We also introduce SpecReveal, a defense prototype combining natural-language intent reconstruction with dual-view detection without target-model fine-tuning. It achieves a 96.88% attack detection rate in a separate prompt-detection evaluation, with false positives assessed on benign programming and specification-style inputs. These results connect specification-induced response differences to practical security consequences and identify intent reconstruction as a complementary detection approach.
[AI-93] PaceReason er: Label-Free Learning to Answer Briefly from Long Reasoning
链接: https://arxiv.org/abs/2610.17900
作者: Changguo Jia,Zhiyou Xiao,Xinbei Ma,Geng Chen,Weiming Zhang,Yiming Xu,Tianqi Zhao,Minghui Zhou
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Extended chain-of-thought reasoning improves problem-solving performance but increases inference cost and latency, motivating more accurate answers with shorter reasoning. We introduce PaceReasoner, a label-free test-time reinforcement learning framework that uses a model’s LONG responses to improve its SHORT-mode accuracy. Both modes share a trainable policy, creating a need for reliable supervision, sufficient SHORT training samples, and continued LONG responses to guide learning. PaceReasoner combines all-response and LONG-only answer consensus through progressive, confidence-weighted supervision. A behavior reward gradually shifts from encouraging SHORT exploration to rewarding agreement with LONG answers, while a ratio controller regulates the balance between modes. Across three model backbones and four benchmarks, PaceReasoner improves average SHORT accuracy over the baselines by 7.4–13.5 percentage points, with fewer than 25 reasoning tokens per response on average. Ablations support the key components, and cross-dataset results demonstrate transfer beyond the adaptation set.
[AI-94] Interaction-as-a-Judge
链接: https://arxiv.org/abs/2610.17895
作者: Wayne Chi,I-Hung Hsu,Li Chen,Wei-Lin Chiang,Ion Stoica,Anastasios Nikolas Angelopoulos,Ameet Talwalkar,Chris Donahue,Cho-Jui Hsieh
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:
Abstract:Coding agents are tasked with producing sophisticated interactive applications such as websites, desktop applications, and games. While humans assess these applications interactively, current automated LLM evaluations primarily focus on code analysis and unit tests. This mismatch can cause LLM judges to miss failures that are immediately apparent to human users. We study Interaction-as-a-Judge, an evaluation paradigm where computer-use agents (CUAs) directly interact with an application. We demonstrate that Interaction-as-a-Judge can improve both \emphevaluation and \emphgeneration of interactive applications. To this end, we collect and release two complementary datasets: 6,858 pairs of applications with human preferences and corresponding agent traces, and 222 applications with human-verified requirement labels. Relative to standard non-interactive approaches, we find that Interaction-as-a-Judge improves evaluation in terms of predicting human preferences ( 65.6% \to 69.1% ) and requirement verification ( 77.0% \to 81.6% ). We also observe scaling benefits: across both preference alignment and requirement verification, higher-quality interactions improve performance. Finally, through both inference- and training-time experiments, we demonstrate that Interaction-as-a-Judge can also improve the quality of generated artifacts. At inference time, Interaction-as-a-Judge repairs 75.5% of broken requirements within five attempts, compared to 55.4% for code-only self-feedback. At training time, using Interaction-as-a-Judge as a reward signal improves win-rate over the base model by 6.7 pp while reducing both generated code length by 35.7% and build failure rates by 7.2 pp compared to a code-only reward signal.
[AI-95] RT-J: Large-Scale Pretraining of Relational Transformers for Context-Efficient Predictions NEURIPS2026
链接: https://arxiv.org/abs/2610.17867
作者: Rishabh Ranjan,Vignesh Kothapalli,Harshvardhan Agarwal,Charilaos Kanatsoulis,Roshan Upendra,Tom Palczewski,Carlos Guestrin,Jure Leskovec
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Databases (cs.DB)
备注: Accepted to NeurIPS 2026. Website: this https URL
Abstract:Foundation models for tables and relational databases are increasingly used for predictive tasks across various scientific and industrial domains. However, adapting existing models to unseen datasets requires tens of thousands of examples. Here we introduce RT-J, an 85M-parameter relational foundation model (RFM) that makes accurate predictions even with hundreds of in-context examples or fewer. To achieve this, we first assemble the Join, a large collection of 639 web-crawled databases with 6,023 LLM-generated forecasting tasks. We pretrain with long contexts of up to 8,192 database cells, using in-context examples from a novel lightweight random-walk retriever. Finally, we leverage context tuning and ensembling to scale up test-time compute. On in-context regression tasks from RelBench, RT-J requires 10x fewer examples on average to match strong feature engineering pipelines with LLM-agent or RDBLearn featurizers and TabICLv2 or LightGBM predictors. Overall, our work pushes the frontier for context-efficient predictions on relational data and highlights the importance of real-world semantics, context engineering and large-scale pretraining for RFMs.
[AI-96] Struct: Learning Shared Temporal Structures for Long-Term Time-Series Forecasting NEURIPS2026
链接: https://arxiv.org/abs/2610.17864
作者: Yangze Li,Zhengyang Zhou,Qihe Huang,Pengkun Wang,Binwu Wang,Yang Wang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: This paper has been accepted to NeurIPS 2026(Main Track)
Abstract:Long-term time series forecasting requires temporal dependency modeling that remains stable and generalizable under temporal distribution shifts. However, real-world time series are often noisy and non-stationary, making deep learning models prone to fitting incidental temporal interactions in the training data. A key question is whether there exist temporal relationship structures that recur across different temporal positions and variables. We analyze temporal relationship matrices on commonly used time series datasets and find that samples from different temporal positions and different variables exhibit shared dominant temporal relationship structures. This indicates that temporal relationship structures contain patterns that recur stably across samples and variables. To learn such shared structures, we propose \textbfTemporal \textbfStructure Network (\textbfTStruct), a simple yet effective network for directly learning shared and generalizable temporal dependencies. Specifically, TStruct directly parameterizes a set of shared temporal dependency matrices and generates normalized combination weights using dynamic sample information and static variable priors. These weights represent a soft allocation of each sample-variable pair over shared temporal dependency structures, enabling the model to generate adaptive temporal dependency matrices under shared structural constraints, rather than fitting incidental interactions in a fully unconstrained manner. Notably, this simple structure-aware design consistently outperforms several strong baselines across all datasets, achieving superior accuracy while maintaining high computational efficiency.
[AI-97] Socratic Skill Learning: Interleaving Teaching and Practice for Transferable Agent Skills
链接: https://arxiv.org/abs/2610.17860
作者: Letian Zhu,Han Li,Yifan Yao,Xinping Lei,Xinyu Che,Jiahao Wang,Dianyi Wang,Yucheng Fu,Jiakai Wang,Zhiqi Bai,Wenbo Su,Jiaheng Liu
类目: Artificial Intelligence (cs.AI)
备注: 25 pages. Code and benchmark: this https URL
Abstract:Agent Skills improve LLM agents by making task knowledge reusable across sessions. Learning transferable Skills requires evidence about a procedure and its conditions of use, which a successful task attempt can leave untested. We introduce Socratic Skill Learning (SSL), which interleaves expert teaching with Student practice. An Educator guides investigation while the Student remains the sole environment operator and Skill author, able to question, test, and adapt advice. To test reuse beyond the learning task, we construct SkillTransferBench, with 35 task families and 408 executable cases, pairing public variation specifications with held-out evaluation in fresh solver sessions. On the main 22-family, 155-case cohort, Fixed-Window SSL reaches 0.801 utility, versus 0.723 without a Skill, 0.761 for adapted Trace2Skill (trajectory distillation), and 0.751 for adapted SKILL-KD (contrastive distillation). Across three model pairs on shared families, interleaved guidance outperforms expert authoring and post-hoc teaching. Process analysis links guidance to Student writing and package revision; a case study shows the Student testing and refining a proposed procedure for later reuse. As demand for new lessons falls, the cost of fixed-window review grows. Issue-Guided SSL tracks unresolved questions across Skill revisions to select and focus review, reaching 0.805 utility with 61.7% fewer Educator tokens and 16.9% fewer total acquisition tokens than Fixed-Window SSL on the main cohort.
[AI-98] MMTA: Benchmarking Multimodal Temporal Analysis with Time Series Text and Vision NEURIPS2026
链接: https://arxiv.org/abs/2610.17856
作者: Ziyang Zhang,Shenyi Li,Yilin Wang,Ziyun Cui,Bowen Zhou,Wen Wu,Chao Zhang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted at the NeurIPS 2026 Evaluations and Datasets Track. Project page: this https URL
Abstract:Time-series analysis is increasingly studied in multimodal settings, yet most benchmarks pair temporal signals with text or with visualizations derived from the same signal. This leaves open whether multimodal large language models (MLLMs) can integrate time series with real-world visual evidence. We introduce MMTA (multimodal temporal analysis), a benchmark of tri-modal tasks with diverse cross-modal complementarity patterns among time series, text, and vision. MMTA spans nine domains – robotics, autonomous driving, human action, manufacturing, AR assistance, healthcare, recommendation, finance, and climate – under video+TS and image+TS layouts, with a 3,524-sample test split covering classification and forecasting. The benchmark standardizes heterogeneous real-world sources into a shared schema with domain-specific prompts, provenance documentation, and success-weighted metrics that penalize both wrong answers and failed generations. Evaluating eight open-source and proprietary MLLMs reveals a shared bottleneck: representing time series as text creates very long token sequences and weakens fine-grained alignment with visual evidence, especially in long-video settings. We then introduce TimeOmni-v, a multimodal LLM extension with a dedicated time-series encoder, temporal positional alignment, modality-aware token layout, and a forecasting head. TimeOmni-v improves long-video classification and time-series inference efficiency, showing that MMTA not only exposes representation and grounding failures but also motivates targeted model design.
[AI-99] Learning Adaptive Deliberation Control for Reliable Recognition with Multi-Agent Language Models
链接: https://arxiv.org/abs/2610.17840
作者: Yuan Liu,Haoyue Peng,Fanliang Hu,Lei Huang,Wenli Cai
类目: Artificial Intelligence (cs.AI)
备注: 33 pages, including references and technical appendices. Supplementary data included in the ancillary files
Abstract:Multi-agent deliberation can improve recognition by criticizing and revising candidate answers, but fixed controls cannot respond to changes in disagreement or reasoning quality. We present an adaptive multi-agent controller (ADMA) that learns continuous policies for proposal, challenge, arbitration, and coordination while keeping the task model frozen. Shared trace features describe quality, agreement, and progress, while role-specific features support decentralized control during inference. One controller trained on legal and financial reasoning is transferred without policy updates to four textual and multimodal benchmarks and two successive task-model versions. With Qwen3.5 held fixed, ADMA-Full improves accuracy by 6.2-14.2 percentage points over Direct and by 2.8-3.0 points over ADMA-Fixed. Paired bootstrap ranges for the Full-Fixed comparison are positive on three tasks. Outcome decompositions show that recoveries exceed introduced errors, while process assessment records fewer false-consensus events and more supported correct outcomes. Transfer to Qwen3.6 remains task dependent, which bounds the generality of the learned policy. Median token and latency ratios are 14.3-18.7 and 7.8-16.5 relative to Direct. These results show that adaptive deliberation can improve fixed-model recognition when its task-dependent gains justify the additional computation.
[AI-100] ProtoArena: Can AI Agents Discover and Exploit Real-World Cross-Component Protocol Vulnerabilities?
链接: https://arxiv.org/abs/2610.17837
作者: Fasheng Miao,Zhaoyu Hu,Xiang Li,Changqing An,Jiutian Zeng,Longtao Huang,Haixin Duan,Jilong Wang
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 18 pages, 5 figures
Abstract:Large language model (LLM) agents are increasingly used for vulnerability analysis, yet existing benchmarks provide limited coverage of exploitation that depends on interactions among network protocol components. Such vulnerabilities arise from inconsistencies in message interpretation, timing, state, or trust assumptions across participants. We introduce ProtoArena, a black-box, multi-component, dynamic benchmark with 80 cross-component vulnerability-exploitation environments spanning 11 protocol and application domains and 7 attack categories. Tasks are primarily informed by published protocol-attack mechanisms. Agents interact with running systems through scoped interfaces to achieve security-effect objectives without target source code. Three information levels assess guided reproduction, exploitation with mechanism and topology context, and effect-directed black-box exploration. A hybrid verifier combines rule-based checks with LLM review of independently collected traffic and logs. We evaluate five LLMs. Claude Opus 4.8 achieves the highest observed success rates: 48.8% under L1 guidance and 12.5% under effect-directed black-box L3. In a separate DNSBomb optimization experiment, it achieves a peak downstream-response to average query-input byte-rate ratio exceeding 171,000 times. The results show that agents can achieve verifiable exploitation effects in some cross-component environments, but success with limited black-box information remains low and varies across domains and attack categories. All experiments use controlled, explicitly authorized targets; they do not test production systems or disclose previously unknown vulnerabilities.
[AI-101] Agent ic-MatriBO: Capability-Compiled Agent ic Bayesian Optimization for Materials and Chemical Design
链接: https://arxiv.org/abs/2610.17832
作者: Yu Zeng,Yuchuan Chen,Ziyue Zhang,Zijian Wei,Yanyi Su,Guojiang Zhao,Hongshuai Wang,Zhifeng Gao
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 28 pages, 9 figures, supplementary information included
Abstract:Materials and chemical optimization must choose the next synthesis, characterization, or high-fidelity simulation from sequential observations under a limited budget. Large language models can use task knowledge to propose adjustments, but composing surrogates, acquisition functions, and search strategies is constrained by compatibility, parameter dependencies, and evidence requirements. We present Agentic-MatriBO, a capability-compiled agentic Bayesian optimization (BO) framework that represents BO components and selection conditions as composable capabilities. A language policy proposes BO recipes and candidate plans; a compiler checks proposals against the registered capability catalog; and a deterministic execution layer validates and commits candidates. Oracle observations are stored in research memory and inform later rounds. We compare Agentic-MatriBO with eight baselines on four offline finite candidate pools—COF, PCE10, Buchwald–Hartwig (BH), and Suzuki–Miyaura—using five seeds per condition; results are reported descriptively. Across the four tasks, Agentic-MatriBO has the best mean endpoint quality after orienting each objective so that higher is better. On COF, it reaches the tabulated optimum in all five seeds both in the end-to-end comparison and after shared initialization, where it also retains an advantage in cumulative quality. Its normalized area under the best-so-far quality curve during the BO stage (nAUC-BO) is substantially higher on COF, near parity with the best baseline on PCE10 and BH, and lower on Suzuki. COF ablations show that removing the capability compiler or evidence gate coincides with lower cumulative quality and fewer exact-optimum hits. The framework provides an inspectable interface for evidence-conditioned BO adaptation.
[AI-102] Benchmarking and Training Reward Models for Cross-Source Evidence Grounding in Search Agents
链接: https://arxiv.org/abs/2610.17827
作者: Haochen Zhao,Minsheng Hao,Zheng Li,Yifan Li,Zikang Liu,Bowen Qin,Yuhao Cheng,Chuang Fan,Haosheng Zou,Jeff Chen,Sujian Li
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Existing approaches to training and evaluating search agents focus heavily on final-answer quality, largely neglecting how generated answers map back to retrieved evidence. Recent methods track citations to enforce grounding, but citation support alone does not establish source independence or whether an answer appropriately handles conflicting evidence. We argue that credibility in open-web retrieval depends on principled cross-source evidence grounding that accounts for source independence and systematically attributes and adjudicates conflicting evidence, while hierarchical judgments of these grounding decisions provide a unified signal for training and evaluation. To this end, we propose an evaluation framework, Cross-Source Evidence Grounding (CSEG), that employs a fine-grained, six-layer decision tree over the complete search trajectory. We introduce CSEG-Bench and trainable reward models to provide targeted supervision following this framework. Our approach helps eliminating the severe data skew of natural trajectories by providing targeted coverage over complex, deep-layer decision paths, such as cross-source conflict resolution and nuanced attribution, that standard rollouts fail to capture. Specifically, (1) we define the hierarchy at label, node, and path levels; (2) we construct 1,160 verified trajectories over nine canonical paths by rewriting answers on real retrieved document pools, with independent frontier judges reproducing the entire gold path and human experts deciding inclusion; (3) we train CSEG-RM, which reaches 83.9% exact accuracy on our benchmark while lying on the performance-token efficiency Pareto frontier. Experiments that use this reward alone to supervise a frozen-retrieval answer policy raise the independently judged mean label from 2.83 to 4.19, and show that learning the hierarchy generalizes across layers at a fraction of frontier-judge cost.
[AI-103] Same Attack Different Judge: A First Controlled Adversarial Comparison of Jev LLM Judges and Purpose-Built Safety Classifiers
链接: https://arxiv.org/abs/2610.17820
作者: Xiang Li,Aichun Huang,Mingming Zhang,Zheli Liu
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 21 pages, 10 figures
Abstract:TypeSafe AI’s Jev is a new kind of safety judge. It generates no text and no label: it answers questions the developer defines in advance and returns a probability for each answer. The vendor admits that adversarial input can change these answers but gives no numbers. In this paper, we present the first controlled adversarial evaluation of Jev. We build a test harness that sends every system the same payload under the same policy, and run it on four public-data tasks, from rule-based ones (input moderation, injection detection, routing) to a judgment-based one (phishing detection), with fifty attack variants from thirteen families. The targets are Jev, five GPT and GLM judges, and two local classifiers (Llama Guard 3, Prompt Guard 2). We find that Jev is the most robust system on input moderation: eight forgeries of its structured input bypass none of its decisions, and a sentence asking only why someone might make a harmful request bypasses 9% of its blocks, against 56% for GPT-6 Astra. On phishing detection. However, the same forgeries bypass 19-29% of the emails it flags, and even without attack Jev detects only 62% of phishing emails, against 77-81% for the GLM judges. Jev is also the fastest judge, but it flags the most benign requests (27%). The local classifiers fail differently: a training-example statement makes Llama Guard 3 miss 62% of the phishing it had flagged. These results suggest that Jev resists attacks on tasks with a clear rule to check but is more exposed on tasks that call for an overall judgment.
[AI-104] Empowering Edge Agents : A Systematic Analysis of Edge-Cloud Collaboration in LLM Agent Execution
链接: https://arxiv.org/abs/2610.17816
作者: Pengcheng Huang,Zhenghao Liu,Shixin Ren,Shi Yu,Sijia Yao,Yukun Yan,Chaojun Xiao,Maosong Sun
类目: Artificial Intelligence (cs.AI)
备注: 23 pages, 5 figures, 12 tables
Abstract:Edge-cloud collaboration combines the lower inference overhead and greater data locality of edge-side execution with the stronger capabilities of cloud-hosted models. However, as LLM agents increasingly operate in long-horizon scenarios, cloud participation becomes interleaved throughout execution rather than a one-shot decision for each query. Such interleaved participation introduces trajectory-level effects that remain insufficiently understood. To address this gap, we introduce EC2-Bench, a benchmark for evaluating edge-cloud collaboration in long-horizon agent execution. It provides executable digital-workspace agent tasks and a unified evaluation framework covering task performance, resource usage, and privacy exposure. Using this benchmark, we empirically study diverse forms of edge-cloud collaboration. Our study reveals three findings. First, collaboration achieves a favorable balance among task performance, resource usage, and privacy exposure compared with edge-only and cloud-only execution. Second, as edge models become stronger, collaborative execution approaches cloud-only performance, and performance differences across strategies shrink, while privacy exposure remains strongly strategy-dependent. Third, stronger edge capability shifts preferred collaboration from externally controlled cloud participation and direct cloud execution toward more edge-controlled collaboration. Further analyses identify cloud-invocation decision-making and instruction following as important edge-side abilities for effective collaboration. These findings provide a systematic understanding of edge-cloud collaboration in long-horizon agent execution and insights for designing future edge agents that effectively collaborate with cloud models. Code and resources are available at this https URL.
[AI-105] GIFT-Eval-CTX: A Benchmark For Contextual Forecasting Model Evaluation
链接: https://arxiv.org/abs/2610.17794
作者: Xu Liu,Hanshu Yan,Taha Aksu,Quang Pham,Zhongzheng Qiao,Jun Hao Liew,Silvio Savarese,Doyen Sahoo,Junnan Li
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Contextual forecasting seeks to improve time series forecasting by incorporating forward-looking information, such as planned events, external conditions, and expert expectations. However, reliable evaluation remains difficult: existing benchmarks often have limited domain coverage, lack standardized history-future splits, or provide text that is not guaranteed to inform the future. We introduce GIFT-Eval-CTX, a benchmark that isolates whether forecasting methods can translate verified future information into accurate numerical trajectories. Built upon GIFT-Eval, GIFT-Eval-CTX contains 10,000 instances across 20 datasets. Each instance provides structured effects derived from the realized future: Macro effects describe horizon-wide behavior, while Micro effects specify localized changes. We construct these annotations with an agentic pipeline combining LLM-based discovery and deterministic code validation. We benchmark TSFMs, frontier LLMs, and tool-using agentic systems. The results show that structured effects reduce aggregate relative MASE by 40.7% for GPT-5.6 Sol and 29.6% for Claude Opus 4.8 relative to history-only prompting. The evaluated MoiraiAgent system achieves the best overall result. We further analyze the contributions of structured effects and agentic systems to performance, and identify opportunities for future research.
[AI-106] Posterior-Averag ed Bellm an Flow Matching for Distributional Critics
链接: https://arxiv.org/abs/2610.17788
作者: Zeqiao Li,Yijing Wang,Haoyu Wang,Zheng Li,Zhiqiang Zuo
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Distributional critics model the full return distribution, providing richer signals for exploration and risk-sensitive control. Recent work has shown that flow-based critics can model these distributions flexibly and provide effective supervision across flow time for temporal-difference learning. However, unlike standard flow matching with fixed data endpoints, a flow critic is trained on Bellman targets recursively generated by a learned target critic under the current policy. Existing objectives therefore either use noisy single-endpoint supervision or rely on the learned flow itself for additional guidance. Our key observation is that an intermediate bridge state may be compatible with several Bellman endpoints, so its velocity should average their directions under the corresponding posterior. We propose Posterior-Averaged Bellman Flow Matching (PA-BFM), which samples an endpoint pool, infers the closed-form posterior over the latent anchor, and regresses to the posterior-mean velocity. Uniform anchor selection makes this target an exact finite-pool Rao–Blackwellization of the single-endpoint target: for every (K), it preserves the population target, cannot increase conditional variance, and its conditional variance scales as (O(K^-1)). The endpoint pool supports conservative updates, while the twin return flows provide critic-disagreement and unsupported-mass diagnostics. Controlled studies show that PA-BFM reduces fixed-query target mean-squared error by (79.7%) and target-induced gradient variance by up to (98.0%), while improving return-distribution fidelity. Across 50 OGBench tasks and four offline-to-online scenarios, PA-BFM achieves higher aggregate performance than Value Flows.
[AI-107] Agent ic4PDM: Host-Governed Evidence Admissibility for Industrial Fault Diagnosis
链接: https://arxiv.org/abs/2610.17786
作者: Nam V. Nguyen,Hoang Van Truong,Hieu D. D.,Duy Khuong Nguyen,Quang Pham,Tung Kieu
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 67 pages, 12 figures, including appendices
Abstract:We study industrial fault diagnosis as a classification task that assigns sensor time-series windows to predefined health states or fault types. Scarce labels and changing operating conditions make repeated training of specialized classifiers costly. Yet structures such as periodicity, trends, and transients can remain diagnostically informative and be examined through reusable analytical tools. LLM agents can select and interpret these tools, but fixed rules or training-derived guidance may not resolve each signal’s ambiguity, while unconstrained interpretation may overstate weak or conflicting results. We introduce Agentic4PDM, a framework that separates adaptive investigation from control over evidence use. LLM agents compare each signal with labeled references from every candidate class, identify unresolved distinctions, and propose follow-up measurements. A deterministic software controller, called the host, enforces contract-checked gated adjudication (CCGA): it checks requests, requires commitment to a complete tool batch before results are observed, and assigns evidence roles based on tool checks against labeled references. These roles distinguish corroborative support from advisory information. The Judge Agent proposes a diagnosis, while the controller checks protocol compliance and records the investigation. Agentic4PDM supports training-free diagnosis using pretrained representations and labeled references without task-specific diagnostic models, and a standard setting that additionally investigates trained-model predictions. Experiments on selected bearing and motor diagnosis challenges and an unfiltered industrial test set supplied by our customer show gains in aggregate macro F1 over the evaluated diagnostic and agentic baselines. Ablations further support the value of the governed workflow.
[AI-108] Memory Echo in Single-Agent Long-Term Memory: Lifecycle Evaluation and Provenance-Guided Mitigation
链接: https://arxiv.org/abs/2610.17763
作者: Yufeng Li,Sheng Yao,Peng Zeng,Dezhang Kong,Shi Lin,Changting Lin,Xun Wang,Meng Han
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Long-term memory supports agents across sessions. Repeated retrieval and write-back can let one historical claim occupy multiple retrieval positions and influence current answers without independent evidence. Existing work examines retrieval, updating, and question answering, leaving same-source record accumulation and post-correction persistence underexplored. Motivated by research in social cognition and memory psychology on repeated exposure, retrieval, and revision, we define \emphMemory Echo as the excess retrieval exposure gained by historical information from the same source without additional independent evidence, and introduce MEchoBench to evaluate its emergence, correction resistance, persistence, and effect on answers. To address this phenomenon, we introduce the Family–Provenance–Authority framework (FPA). Specifically, FPA selects one representative per source family, conditions the eligibility of old records for the answer context on verified corrections, and constrains final answers with authoritative evidence. In paired replay on 190 fixed questions across four memory frameworks, FPA raises accuracy from 0.825 to 0.920 with DeepSeek V4 Flash and from 0.782 to 0.916 with Kimi K2.6; mean Decision Contamination Effect falls from 0.106 to 0.030 and from 0.109 to 0.038, respectively. These results show that agent memory can preserve traceable history without allowing repeated historical information to govern current decisions. The code is available at this https URL.
[AI-109] Gaussian Process Bandits for AI System Optimization with Selectively Verified Feedback
链接: https://arxiv.org/abs/2610.17760
作者: Shuang Wu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注:
Abstract:Gaussian process bandits exploit similarities among configurations to optimize prompts, retrieval systems, and agent workflows. In these applications, however, inexpensive model-based evaluations can disagree with trusted task outcomes. We study selective verification when an action is chosen before a diagnostic is observed and trusted feedback is informative on an unknown set of diagnostic states. A finite reliability class and a finite RKHS signal dictionary describe these two sources of uncertainty. We construct a fixed audit design that combines reliability coverage with reward separation and establish a finite-time regret bound that accounts for a hard audit budget and the rounds needed to obtain informative observations. A safe adaptive extension mixes this robust design with predictable Chernoff allocations while preserving a uniform identification floor and exact expected acquisition-cost accounting. For signed-coordinate kernels, structured and diffuse reliability classes require different audit scales despite sharing the same reward geometry. Synthetic experiments distinguish these certified designs from an adaptive posterior heuristic and GP controls. Safe adaptive allocation reduces aggregate regret relative to the fixed design under the exact dictionary, but its finite-budget error certificate is numerically loose. Conditional-feedback errors can remove the heuristic’s advantage and target perturbations can reverse it. Separate AI-task replays assess feedback correction and fitted-dictionary policies without establishing end-to-end validation of a learned selective-verification method.
[AI-110] Instructions Are Evidence Not Objectives: Measuring Outcome Alignment of Agents
链接: https://arxiv.org/abs/2610.17757
作者: Kaiming Liu,Yuxuan Liu,Jinrui Ju,Minyuan Ruan,Xuanyu Lei,Peng Li,Yang Liu
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:
Abstract:Large language model agents can follow complex human instructions to solve real-world tasks. In practice, however, initial instructions are often incomplete. In our analysis of 100 user-agent trajectories from SWE-chat, 56% of initial instructions omit requirements that later emerge as necessary, and 52% of the useful follow-up requirements could have been inferred from the project. A user states requirements from their own understanding of the task, and these \textbfstated requirements often differ from the \textbftrue requirements that completing the task imposes. An agent should therefore treat the instruction as evidence rather than as the objective, inferring the true requirements from the instruction and the environment and satisfying them even when the user is not yet aware of all of them. We call this ability \textbfoutcome alignment. In this work, we construct \textbfTiGanEval, a benchmark of software engineering and research tasks. In every task the user genuinely lacks part of the requirements, each requirement stays discoverable from the code and runtime behaviour of the environment, and requirements the agent infers are separated from those the user adds after observing delivered work. These properties together make outcome alignment measurable. Evaluating eight frontier models including GPT-5.6-Sol, we find outcome alignment broadly missing: the models lose 53 to 67 points of success rate when moving from the full specification to the partial instruction, and consequence feedback across four submissions does not close the gap. Furthermore, a plan-mode harness cannot compensate for the missing ability of the model itself, which places outcome alignment among the abilities the next generation of models should be developed for.
[AI-111] I3L: Iterative Interactive Imitation Learning for Whole-Body Mobile Manipulation
链接: https://arxiv.org/abs/2610.17748
作者: Wensi Ai,Christopher Luey,Karthik Dharmarajan,Evans Han,Yunfan Jiang,Stef Ren,Jianwen Xie,Jiajun Wu,Li Fei-Fei,Ruohan Zhang
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: Project website: this https URL
Abstract:Policies trained offline from teleoperated demonstrations can degrade outside their training distribution. This gap is particularly acute in long-horizon whole-body mobile manipulation, where long horizons amplify compounding errors and coordinated arm, torso, and base motion makes exhaustive demonstration coverage impractical. We introduce I3L, an integrated hardware and learning system that combines whole-body correction with iterative policy refinement to address this last-mile reliability gap. Its low-cost, open-source JoyLo+ interface uses policy-tracking leader arms and touch-triggered takeover to enable coordinated corrections across the body parts, while iterative deployment and retraining allow operators to target failures of the updated policy. We evaluate I3L across three robot embodiments and eight simulated and real-world tasks, including four long-horizon whole-body mobile manipulation tasks. Controlled simulation experiments show that four-round HG-DAgger achieves the highest final success on both tasks. Under this protocol, I3L improves autonomous success over the base policies by an average of 68.3 percentage points and achieves at least 90% success on every task, outperforming demonstration-only training by 35.6 percentage points on average with the same number of training trajectories. Averaged across tasks, takeover duration per episode decreases by 72% over four rounds. Together, these results show that I3L helps close the last-mile gap between offline training and reliable deployment and provides practical insights into collecting and using corrections for long-horizon whole-body mobile manipulation. Videos, visualizations, and appendix are at this https URL.
[AI-112] Coupled Harness–Model Evolution for Scientific Equation Discovery
链接: https://arxiv.org/abs/2610.17740
作者: Pengze Li,Jiaquan Zhang,Encheng Su,Yunbo Long,Zihang Zeng,Xinping Liu,Han Zhang,Junchi Yu,Shixiang Tang,Aoran Wang,Xi Chen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) are increasingly used to discover scientific equations from experimental data. Such a discovery system has two parts that must work together: a library of executable computational tools and the model’s tool-use policy. The tools supply the solving capability, from diagnosing structure in the data to fitting candidate equations; the policy plans and orchestrates their use. Yet existing systems adapt only one part, keeping either the toolset or the model fixed, which leaves the corresponding failures uncorrectable. We develop a coupled evolution framework that improves both parts from the same verifiable fit-error signal. In the tool-evolution phase, an agent grows an executable tool library from a small set of generic numerical operations by diagnosing its own failures: it authors a new tool when a capability is missing and revises its selection rules when the right tool went unused. In the policy-evolution phase, a small open model (roughly 3B active parameters) is fine-tuned on successful tool-use trajectories collected with the frozen library, on training formulas screened to be disjoint from all evaluation benchmarks. Ablations show the two phases are complementary: with no tool library at all the same model solves almost nothing, the tools account for most of the gain, and trajectory training adds a further gain that is largest on the most difficult problems. The full system solves 79.1/71.3% of LSR-Synth problems at fit tolerances 10^-2 / 10^-3 , reaches 63.8% pooled accuracy on the distributionally different SRSD-Feynman panel, and obtains a normalized CP3-28 aggregate of -7.27 . These results suggest a broader recipe for scientific agents: let the model and its instruments improve each other under verifiable feedback.
[AI-113] HiSTOR: Structured Treatment-outcome Representation Learning for High-Dimensional Longitudinal Data
链接: https://arxiv.org/abs/2610.17739
作者: Chuanqi Lao,Junzhou Huang,Qiao Liu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Estimating potential outcomes under alternative treatment sequences is essential for personalized decision-making, but remains challenging with high-dimensional covariates and time-varying confounding. In this study, we propose HiSTOR (High-dimensional Structured Treatment-Outcome Representation), an end-to-end framework that integrates nonlinear dimension reduction with longitudinal counterfactual prediction. The central idea is to learn structured low-dimensional representations of both current patient states and evolving histories, according to their shared and distinct relationships with treatment assignment and outcomes. This extends latent-space covariate adjustment to longitudinal settings, jointly addressing covariate dimensionality and historical complexity. The resulting framework supports one-step and multi-step outcome prediction under specified treatment sequences. Experiments on CVSim and CancerSim demonstrate the superiority of HiSTOR across varying covariate dimensions, confounding strengths, and prediction horizons compared to state-of-the-art methods. To sum up, HiSTOR offers a unified approach to connecting high-dimensional covariate adjustment with longitudinal potential-outcome prediction.
[AI-114] DuplexAct: How Interactive Are Frontier Full-Duplex Spoken Language Models?
链接: https://arxiv.org/abs/2610.17738
作者: Jiaheng Dong,Shihong Tan,Gongping Huang,Jean Honorio,Abhirup Ghosh,Ting Dang
类目: Artificial Intelligence (cs.AI)
备注: Under review
Abstract:End-to-end (E2E) full-duplex spoken language models (FD-SLMs) enable simutanous listening and speaking continously, requiring models to take appropriate interaction actions as conversations unfold. However, existing benchmarks lack a structured action-level formulation, often evaluate action occurrence rather than success, and provide limited coverage of diverse real-world conversations. We introduce DuplexAct, a benchmark for evaluating interaction action taking in E2E FD-SLMs. DuplexAct defines a three-level action taxonomy spanning eight interaction actions and 24 corresponding interaction conditions. We further characterize successful action-taking through immediate execution and subsequent follow-through, and develop action-specific evaluation metrics for both. DuplexAct contains 4,691 test instances spanning casual conversation, meetings, customer service, and interviews, including 2,531 instances derived from real-world conversational recordings. Together, DuplexAct provides a structured and real-world-grounded framework for systematically assessing how E2E FD-SLMs participate in ongoing spoken interaction.
[AI-115] rain for Many Update with One: Improving Multi-Sample Reasoning via Fenchel-Bregman Posterior Learning
链接: https://arxiv.org/abs/2610.17735
作者: Sungjun Lim,Adrian Robert Minut,Nico Carlsen Daheim,Mohammad Emtiyaz Khan,Kyungwoo Song,Thomas Möllenhoff
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Multi-sample reasoning measures candidate coverage, the probability that at least one of K candidates is correct. However, minimizing the average loss of individual attempts does not directly optimize candidate coverage. We therefore formulate finite- K posterior learning, which trains a distribution over solver parameters to reduce the probability that all K independently generated candidates fail. To optimize this objective efficiently, we propose a Fenchel-Bregman (FB) approach that requires only one posterior draw per update, rather than the K used by direct Monte Carlo optimization. FB first uses an exact Fenchel reformulation to separate problem-specific weights from the posterior update, enabling one-draw posterior updates. It then applies a closed-form Bregman update to track these weights through an exponential running average of observed failures, reusing the same draw. We instantiate FB with IVON, a variational optimizer for Gaussian posteriors over solver parameters, yielding Fenchel-Bregman IVON (FBI). Across diverse recursive and autoregressive reasoning tasks, FBI improves candidate coverage. In a matched comparison, it achieves comparable Pass@10 using 12.3% of the direct Monte Carlo baseline’s measured training time per update.
[AI-116] Explore with Memory Learn without It: In-Context Exploration and Self-Distillation for LLM Reasoning
链接: https://arxiv.org/abs/2610.17734
作者: Yutong Chen,Zhichao Xu,Rui Liu,Kunpeng Liu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: preprint
Abstract:Exploration bounds learning efficiency in reinforcement learning with verifiable rewards (RLVR). Current methods explore mostly through independent sampling, without fully using the information gathered along the way, such as known wrong answers and environment feedback. We propose LEAF, which injects a dynamic, problem-specific memory into the context before every rollout. The memory contains the most recent response, its reward, and other environment feedback. It is updated after every verifier call. We decouple policy updates from rollouts and update the policy through self-distillation. On four benchmarks and two open models, LEAF accelerates exploration by about 4x compared with a random sampling baseline. On unseen test sets, it improves accuracy over the strongest baseline by 1.59 and 1.83 percentage points on the two models, while using less compute than any RL baseline. These results highlight the potential of LEAF in improving both exploration and learning efficiency. Code is available at: this https URL
[AI-117] CoCA: Criterion-Aware Credit Assignment for Rubric-Based Reinforcement Learning
链接: https://arxiv.org/abs/2610.17733
作者: Guangcheng Zhu,Yingfan Ma,Xing Zheng,Haobo Wang,Shenzhi Yang,Kai Tang,Zhengqing Zang,Bowen Song,Weiqiang Wang,Gang Chen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Rubric-based reinforcement learning (RL) evaluates responses against explicit criteria to construct rewards, offering a viable way to train large language models on open-ended tasks. However, standard methods compress criterion-level feedback into scalar rewards and assign uniform token advantages, leaving structured rubric supervision underutilized. Recent work seeks finer-grained guidance by treating rubrics as privileged information to derive self-distillation targets, but the resulting learning signal is not grounded in criterion outcomes and can lead to reward degradation. This motivates us to reframe the problem as criterion-aware credit assignment, where privileged rubric information serves as a relevance cue for credit allocation rather than a source of imitation targets. Accordingly, we propose CoCA, which derives criterion-level credits from group-relative outcomes and allocates them across tokens based on criterion-induced distribution shifts, yielding both outcome-grounded and fine-grained learning signals. Empirically, CoCA achieves an average gain of +4.2% over standard rubric-based RL across medicine and science benchmarks, demonstrating more effective learning on open-ended tasks.
[AI-118] SRL: Trigger-based Supervised Reinforcement Learning for Continual Agent Training
链接: https://arxiv.org/abs/2610.17726
作者: Yuqian Fu,Kaiwen Jiang,Zijun Song,Jiawei Wang,Yuanheng Zhu,Dongbin Zhao
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model (LLM) agents excel at complex, multi-step tasks but suffer from catastrophic forgetting during continual learning–acquiring new skills often degrades previously mastered capabilities. Existing post-training paradigms face a fundamental trade-off: Supervised Fine-Tuning (SFT) is sample-efficient but prone to severe forgetting, while Reinforcement Learning (RL) preserves prior capabilities but converges slowly without expert guidance. To bridge this gap, we introduce Trigger-based Supervised Reinforcement Learning (TSRL), a framework that integrates the strengths of both paradigms through an adaptive, state-dependent expert supervision mechanism. TSRL employs RL as the optimization backbone, activating SFT selectively on “hard states” where the policy fails. Theoretically, we show that restricting supervision to failure states derives an upper bound on the cumulative parameter updates, thereby reducing forgetting compared to full SFT. Empirically, TSRL achieves consistent gains across three agent benchmarks (ALFWorld, WebShop, and Deep Search) at both 7B and 32B scales. In continual learning scenarios, our method retains up to 9.6 points more capability on prior tasks than SFT-RL baselines, while matching or exceeding target-task performance.
[AI-119] Evidence-First Agent ic Verification
链接: https://arxiv.org/abs/2610.17724
作者: Reza Sanayei,Razvan-Gabriel Dumitru,Sunjie Hou,MohammadHossein Rezaei,Haris Riaz,Robert Vacareanu,Minglai Yang,Ying Liu,Mihai Surdeanu,Daniel Yue Zhang
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Long-horizon, semi-verifiable agentic tasks often rely on rubric-based evaluation, but the evidence needed to assess these tasks can be spread across the environment state and the execution trace. A judge may overlook this evidence or accept unsupported claims, including an agent’s own reports of success. When rubric verdicts serve as reinforcement-learning rewards, such errors can reinforce undesirable behavior. We introduce Sauron, an evidence-first agentic verifier that investigates both state and trace and relies on a claim only after independently validating it against its source. Bounded queries and joint investigation of related criteria reduce context use and redundant work, and each verdict carries an auditable receipt of its supporting evidence. Across 2,586 human-labeled criteria from 100 personal-assistant episodes, Sauron achieves the highest F1 of five verifiers on all eight backbones tested while costing less than Gandalf, the strongest baseline. With GPT-5.6 Sol, it reaches 0.96 F1 versus Gandalf’s 0.92 at 23% lower cost, with a wider margin on process criteria. On a paired fault-injection suite covering 13 families, several of which simulate reward hacking, Sauron detects 97% of injected failures while flagging only 2% of clean controls, outperforming baselines even when they are given the full trace. On TRAIL, a benchmark for detecting and localizing real agent errors, Sauron leads both Gandalf and TRAIL’s own judge on all three quality metrics.
[AI-120] Upstream Influence Downstream Re-Anchoring in LLM Agent Chains
链接: https://arxiv.org/abs/2610.17723
作者: Chun-Yu Hsu,Lun-Wei Ku,Chung-Chi Chen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:People act on evaluative judgments produced by LLM agents, and in multi-agent systems those judgments pass through several agents before reaching a person. We show that a workflow’s information architecture—source access, handoff format, and which model reads the untrusted content—governs how much of a coordinated manipulation reaches the end. On 32 Taiwan~2025 recall cases, ten authentic Threads comments shown to one agent move its stance by +0.171 toward the injected position, and a mirrored set moves it the other way. Passing that judgment down a chain removes more than half of the shift by the third agent; keeping the comments in front of every agent sustains more of it. The reduction is not attrition in transit: each receiver discounts the inherited stance in proportion to how far it sits from its own reading of the evidence, returning to what that evidence implies for the case rather than to neutrality—consistent with partial re-anchoring. Handoff format is a second lever: a stance relayed without its reasoning is largely discounted. Susceptibility at the exposed seat varies fourfold across models, while the chain’s arrangement does not matter at all. None of this is visible in the transcripts—the injected claims are almost never restated, so reasoning-trace audits miss the attack.
[AI-121] Keep Your Receipts: Trustworthy Cleanup of Agent -Built Knowledge Graphs NEURIPS2026
链接: https://arxiv.org/abs/2610.17719
作者: Vishisht Choudhary
类目: Artificial Intelligence (cs.AI)
备注: 40th Conference on Neural Information Processing Systems (NeurIPS 2026). Workshop: Who Verifies the Agents? Toward Reliable Agent Development
Abstract:Agents build knowledge graphs as memory, and building them has become easy. Maintaining them has not. When a source is retracted, overruled, or erased, everything built on top of it (entity merges, entailed edges, community summaries) must be re-examined, and the maintainer must choose between deleting too much and too little. Deployed pipelines delete only the directly extracted facts: on our corpus that reduces contaminated queries from 32.2% to 28.0% under a real retraction and changes nothing under predicate revocation. Rebuilding is no repair for a constructor that samples: two builds of the same 300-document corpus differ in 29,475 artifacts, five times the 6,248 that removing the 61 retracted documents changes. PolyGraph instead asks the constructor for receipts: for every artifact, alternative bundles of source evidence, each claimed sufficient on its own, emitted in the call that creates it. Cleanup is then set arithmetic over the receipts, with no model calls. Because receipts are the model’s own self-report, a re-derivation audit checks them by deleting the cited evidence and testing whether the constructor still produces the artifact. The audit finds that a receipt is trustworthy when it is shaped like the claim it backs. Merge receipts, each backing one narrow claim, agree with re-derivation 68% and 85% of the time. Summary receipts, elicited per sentence but stored as one flat list for the whole text, are wrong about two thirds of the time, almost always by keeping dead content. One receipt per sentence instead is 100% and 88% faithful and lets cleanup keep 33 of 35 touched summaries. The result holds under a second oracle model, a claim-blind prompt, ten further deletions, a second domain and a second constructor. Distrusting the summary layer cuts contaminated queries to 0.4% and 0.8% at 99.8 to 100% of the recall a perfect maintainer would return.
[AI-122] ECG-RSI: Recursive Self-Improvement through Targeted Electrocardiogram Waveform Evolution
链接: https://arxiv.org/abs/2610.17718
作者: Xiaoliang Wang,Zhuo Chen,Chuanyang Zheng,Jiangwei Lao,Quanbin Wang,Honglin Lin,Jiajia Liu,Zhihong Lu,Hao Jin,Jian Wang,Lijun Wu
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Signal Processing (eess.SP); Quantitative Methods (q-bio.QM)
备注: Project: this https URL GitHub: this https URL
Abstract:Electrocardiogram (ECG) models need supervision that targets evolving weaknesses and remains faithful to waveform evidence. We introduce ECG-RSI, a recursive self-improvement framework combining verified waveform synthesis with feedback-guided intervention search. Signal remeasurement verifies waveform edits; construction errors guide generator repair, while learner errors motivate interventions compared on a separate selection set. Across seven benchmarks, three rounds raise Qwen3.5-35B-A3B’s mean question-answering accuracy from 34.57% to 56.04% and mean classification F1 from 24.27 to 34.03. Across the retained trajectory, the preferred supervision shifts from direct reinforcement to confusion-focused contrasts. Continued feedback correction yields diminishing external gains. These results motivate separating error diagnosis from intervention selection and checking whether gains on the feedback set transfer to independent benchmarks.
[AI-123] Adaptive-Group Process Rewards for GUI Agents
链接: https://arxiv.org/abs/2610.17710
作者: Zirui Shang,Yang Liu,Xin Shu,Zhi Gao,Xinxiao Wu,Lifeng Fan
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Training graphical user interface (GUI) agents with reinforcement learning suffers from sparse outcome rewards, leading to inefficient learning. Process rewards offer a promising direction for credit assignment to intermediate actions; however, accurately evaluating an action requires identifying evidence across long-horizon GUI trajectories, as its value may depend on information beyond its immediate state and observation. In this paper, we propose an adaptive-group process reward framework that dynamically partitions a GUI trajectory into semantically coherent subgoal groups, enabling process reward estimation at both the action and group levels. At the action level, a critic evaluates each action based on the visual changes and subsequent actions within the group for action correctness and contribution to the subgoal. At the group level, the framework aggregates action-level evaluations to assess subgoal completion and contribution to the overall task, and further refines these assessments by leveraging the context of all groups in the trajectory. The process reward of each action is obtained by combining its action-level evaluation with the refined group-level evaluation. We further augment GRPO’s outcome advantage with the proposed process rewards to provide finer-grained guidance for GUI agent training. Experiments on process reward benchmarks and downstream RL training demonstrate the effectiveness of the proposed framework.
[AI-124] Where Small Mistakes Matter: Quantifying State Criticality for Imitation Learning
链接: https://arxiv.org/abs/2610.17709
作者: Yusuf Umut Ciftci,Joonwon Kang,Ajay Suresha Sathya,Somil Bansal
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: Accepted to the Conference on Robot Learning (CoRL 2026). 18 pages, 10 figures, 8 tables. Project website: this https URL
Abstract:Imitation learning typically treats every demonstration state as equally important, optimizing a behavior-cloning objective uniformly across the dataset. Yet not all states are created equal: in a small number of critical states, even a slight deviation from the expert action can lead to a failure, while in many other states such errors are recoverable and inconsequential. We formalize state criticality in imitation learning as the worst-case degradation in downstream value induced by a bounded perturbation of the expert action at a given state. We then propose SCRIM, a practical State CRIticality Measure that estimates this quantity by training an adversarial disturbance policy and a value critic around a fixed imitation policy. SCRIM assigns each demonstration state a score reflecting how vulnerable the task outcome is to small, policy-relevant action errors. We show that these scores can be used to reweight behavior-cloning losses, improving imitation at consequential states and increasing overall task performance. We further show that SCRIM enables data curation by selecting compact, high-value subsets of demonstrations that preserve performance while substantially reducing the amount of training data. We evaluate SCRIM across various simulated and real-world robotic manipulation tasks, and demonstrate that SCRIM improves policy training, enables efficient data selection, and provides interpretable insights into which moments in a demonstration are most critical for task success. Project website: this https URL
[AI-125] A Node Embedding Framework for Model-Centric Conformance Checking in Process Mining
链接: https://arxiv.org/abs/2610.17702
作者: Mostafa Abbasi,Homayoun Najjaran
类目: Artificial Intelligence (cs.AI)
备注: 34 pages, 13 figures, 8 tables. Code: this https URL
Abstract:Evaluating the quality of process models discovered from event logs is a fundamental challenge in process mining. Traditional evaluation relies on conformance metrics such as fitness and precision, which are highly sensitive to noise and deviations in event logs and therefore often fail to give a consistent assessment of discovery algorithms. We introduce a model-centric evaluation framework that combines node embeddings with Procrustes alignment to enable cross-graph comparability, since embeddings derived from different graphs reside in distinct vector spaces. The proposed PetriNetGraph method transforms Petri net process models into graph representations compatible with embedding algorithms. Comparing the node embeddings of ground-truth and discovered models through a parameterized cosine distance yields a unified and bounded metric for evaluating process discovery across datasets and methods. We evaluate the framework under static and context-aware scenarios to examine how prior knowledge of activity existence influences the assessment of structural similarity. The framework also integrates a compliance checking mechanism that localizes deviations at the activity level and an entropy-regularized node distance aggregation function that quantifies global deviations across the model. Experiments on nine publicly available event logs show that the approach provides a reliable embedding-based conformance indicator that reflects the behavior of fitness and precision. Experiments on synthetic logs with controlled noise levels confirm that it is substantially more robust to log deviations than traditional conformance metrics, PetriNet2Vec, and its dynamic-window and Procrustes-aligned variants, while remaining sensitive to process model degradation. All implementation details and datasets are publicly available.
[AI-126] Understanding Gene Expression-to-Image Learning: Spatial Structure and Information Fidelity
链接: https://arxiv.org/abs/2610.17699
作者: Xiaoqing Lian,Pei Liu,Jiahui Tan,Quan Zou,Chen Lin,Xiangxiang Zeng
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Genomics (q-bio.GN)
备注:
Abstract:Single-cell RNA sequencing (scRNA-seq) enables cell-level analysis of biological heterogeneity, but learning effective representations from high-dimensional, unordered gene expression profiles remains challenging. Gene Expression-to-Image (GE2I) offers an effective learning strategy by arranging gene values on pixel grids for convolutional neural networks, but its underlying mechanisms remain unclear. We study GE2I on full-transcriptome data while keeping the network and training procedure fixed. Our results reveal a trade-off between spatial smoothness and gene-level fidelity. Structured mappings produce smooth, low-frequency patterns that support faster early optimization, whereas assigning multiple genes to the same pixel obscures expression differences and can reduce predictive performance. This trade-off explains why random mappings can compete with structured layouts when many genes are represented. One-to-one gene-to-pixel assignment improves prediction across the tested visual backbones even when image smoothness decreases. We further find that the benefits of pretraining and augmentation depend on the backbone and mapping. These findings provide practical guidance for pixel allocation and the transfer of vision priors to gene expression data.
[AI-127] LeanPRM: Lean-Grounded Process Supervision for Natural Language Proof Verification
链接: https://arxiv.org/abs/2610.17697
作者: Rodney Lafuente-Mercado,Salman Rahman,Jingyan Shen,James Shiffer,Saadia Gabriel,Pavel Izmailov
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO)
备注: 29 pages, 9 figures. Project page: this https URL
Abstract:AI systems now write long and difficult mathematical proofs, and checking their step-by-step validity is becoming a bottleneck. A learned verifier could check proofs cheaply, but training one requires a correctness label on every step, and such labels are hard to obtain for open-ended proofs. Existing methods take them from human annotators, which is costly, or derive them from final answers, which proofs do not have, and formalizing each proof in Lean requires expert effort. We introduce LeanPRM, a framework that uses the Lean proof assistant to create step-level labels for natural-language proofs. A model writes a natural-language proof together with a paired formal proof, without feedback from Lean. Lean then labels every step, and a second model discards failures caused only by the formal syntax, so that the negative labels mark errors in the mathematics. LeanPRM contains 10,634 proofs with 152K labeled steps and no human annotation. A Qwen3-4B verifier trained on LeanPRM raises exact step-label accuracy on a held-out set of 200 proofs (LeanPRM-Bench) from 48.2 to 74.7, as opposed to 60.3 when the same base model is trained on PRM800K. It also selects better proofs in best-of-N selection on IMO-level problems, and outperforms the PRM800K-trained verifier on three of four external verification benchmarks. Training on LeanPRM improves verifiers at every size from 0.6B to 8B parameters on LeanPRM-Bench, Hard2Verify and IMO-GradingBench. Given the labels need no human annotation, LeanPRM can grow with the supply of formal statements, offering a path to step-level feedback on AI-generated proofs that are hard for humans to check.
[AI-128] zPocket: Better Pocket Finding with Frozen Co-Folding Features
链接: https://arxiv.org/abs/2610.17696
作者: Tally Portnoi,Allison M Keys,Benjamin R DiFrancesco,Nate Gruver
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Identifying protein-ligand binding sites (pockets) is important for understanding protein function and developing drugs. However, no public method has leveraged co-folding models for novel pocket prediction; furthermore, this task lacks a suitable benchmark as existing benchmarks are within the training distribution of co-folding models. We therefore developed PocketWatch,a curated dataset of biologically relevant ligand pockets. The PocketWatch test set consists of unseen pockets stratified by sequence similarity, structural similarity, and crypticity to enable evaluation of generalization to novel and challenging pockets. Using this dataset, we develop zPocket, a method that trains pocket-prediction heads on top of frozen features from co-folding models. We show that zPocket heads outperform baseline methods across all stratifications. Our best zPocket model achieves overall per-target AUPRC of 0.745 compared to 0.598 and 0.595 for the strongest baselines, VN-EGNN and UniSite-1D retrained on the PocketWatch split (0.559 and 0.531 when using the published weights); the widely used P2Rank reaches 0.477. We further show that zPocket can be used to outperform baseline methods on top-N+2 recall of predicted pocket centroids through clustering per-residue predictions or using zPocket to rescore centroids proposed by other models. Finally, we explore two applications of zPocket: crystallographic fragment screens and a proteome-scale scan for uncharacterized pockets on human proteins. Together these results suggest that zPocket heads built on co-folding trunk features can greatly advance ligand binding site prediction.
[AI-129] HelpMeOut: A Neuropsychology-Grounded Benchmark for Error-Aware Monitoring of Household Task Execution
链接: https://arxiv.org/abs/2610.17694
作者: Haoran Wang,Jingwei Huang,Anqing Duan,Ho Seok Ahn,Bruce Macdonald
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Symbolic Computation (cs.SC)
备注: Accepted to the 10th Conference on Robot Learning (CoRL 2026). 30 pages (9-page main text), 8 figures, 18 tables. Project page: this https URL
Abstract:We introduce HelpMeOut, an embodied AI benchmark for error-aware monitoring of household task execution. Existing embodied household benchmarks typically assume that humans execute tasks correctly, but this assumption does not hold for cognitively impaired persons or older adults whose everyday activities produce structured execution errors. To address this gap, HelpMeOut operationalizes research on everyday action errors into 20000 paired activity episodes, covering nine neuropsychologically motivated error categories. In each episode, agents observe a streaming human activity trajectory and must monitor whether the execution is unfolding correctly, diagnosing any structured errors as early as possible. We evaluate learned-monitor, symbolic, change-point, and frontier LLM prompting baselines. Results show that these methods perform poorly as error monitors, often missing when execution goes wrong. These results suggest a gap left by standard goal-inference and task-completion benchmarks, making HelpMeOut a testbed for error-aware assistive agents.
[AI-130] Stable Token-level Scheduling for Serverless LLM Inference
链接: https://arxiv.org/abs/2610.17689
作者: Hardi Ashishbhai Desai,Balaji N K,Sourav Kanti Addya
类目: Artificial Intelligence (cs.AI)
备注: 10 Pages, 7 figures, Conference
Abstract:Serverless inference is emerging as an efficient deployment model for Large Language Models (LLMs) due to its ability to elastically scale resources, improve GPU utilization through resource sharing, and reduce infrastructure costs. However, existing approaches primarily optimize time to first token (TTFT) while overlooking inter-token latency (ITL), a key determinant of user experience during response generation. We observe that LLM inference comprises a computation-intensive prefill stage and a bandwidth-bound decode stage, where admitting new requests can disrupt ongoing decoding and increase latency. To address this, we propose Token Window Pacing (TWP), a closed-loop scheduling framework that leverages runtime feedback from ITL, queue depth, and workload characteristics to regulate request admission at the token level instead of the batch level. TWP dynamically adjusts a selfregulating token window and incorporates chunk-based prefill to minimize interference from incoming requests. Experiments on realistic serverless workloads show that TWP delivers stable TTFT, significantly reduces latency spikes, lowers P99 tail latency, minimizes ITL variation, and improves throughput by up to 45% over ServerlessLLM and 58% over conventional schedulers. These results demonstrate that feedback-driven tokenlevel scheduling enables stable, high-throughput serverless LLM inference without modifying the underlying model.
[AI-131] Dynamic K-Head: Difficulty-Supervised Budget Selection for Continuous Latent Reasoning
链接: https://arxiv.org/abs/2610.17679
作者: Ying Li,Yuqi Chen,Sixuan Li,Yunfeng Cai
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Chain-of-thought (CoT) improves reasoning but incurs substantial autoregressive rationale cost. Continuous CoT moves intermediate computation into recurrent hidden states, yet typically uses a fixed number of updates even though the best budget varies across questions and accuracy need not improve monotonically with K . We introduce Dynamic K-Head, a question-conditioned, pre-rollout budget selector. It is jointly trained with the reasoner under selected budgets, using difficulty supervision based on scores assigned from the question alone. We also show that latent-update count can misrepresent end-to-end cost because a short latent rollout can shift computation into a longer explicit continuation. We therefore count latent updates, implementation-specific special tokens, and all decoded text. Instantiated with CoDI on Llama 3.2 1B and GSM8K, Dynamic K-Head reaches 54.44% exact-match accuracy at 12.03 average reasoning/output tokens, compared with 52.31% at 15.21 tokens for fixed- K=6 CoDI—a 2.12-point gain with 20.9% lower token cost. Across three out-of-distribution arithmetic datasets, it reduces cost by 19.3–25.7% while improving accuracy by 0.30–2.22 points. These results demonstrate that difficulty-supervised budget selection can improve both accuracy and efficiency, while stable budget selection remains a central challenge.
[AI-132] Encounter-Driven City-Scale Human Mobility Generation with Large Language Models
链接: https://arxiv.org/abs/2610.17672
作者: Chao Li,Renhe Jiang,Chuan Xiao,Jiawei Wang
类目: Artificial Intelligence (cs.AI)
备注: 29 pages, including references and appendices
Abstract:Large language models (LLMs) have shown strong potential for human mobility generation by modeling individual routines, activities, and mobility preferences. However, existing approaches remain largely resident-centric: generating realistic individual trajectories does not necessarily recover the encounters that emerge when multiple residents co-occur at the same place and time. To this end, we study encounter-aware human mobility generation, which jointly preserves individual mobility and population-level interaction structure. We propose EncounterMob, an encounter-driven LLM framework consisting of a coarse-to-fine Encounter Generator and an Iterative Coupling Framework. The Encounter Generator first captures city-level encounter rhythms and then instantiates participant-time-venue encounters, while the Iterative Coupling Framework coordinates these encounters with personal mobility behavior through successive refinement. We further introduce an encounter recovery protocol covering encounter marginal and joint distributions over residents, times, and venues. Experiments on the Japan and NYC datasets show that EncounterMob outperforms all external baselines on all seven encounter metrics and all six personal mobility metrics. Compared with ELLMob, it reduces the participant-time-venue joint error by 40.4%/41.7% and the mean personal mobility error by 58.8%/32.7% on Japan/NYC, respectively.
[AI-133] Cowork-Gym: Synthesizing Interactive Work Environments at Scale
链接: https://arxiv.org/abs/2610.17670
作者: Anni Zou,Cheng Chen,Jiaji Deng,Weikang Zhou,Xinji Mai,Yue Wu,Jianing Lin,Lipeng Xie,Lixiong Qin,Qingyue Zhao,Que Shen,Shuchang Tao,Xuchen Pan,Yaliang Li,Yunpeng Zhai,Yushuo Chen,Zhihai Wang,Zhuo Zhang,Zhaoyang Liu,Bolin Ding,Jingren Zhou
类目: Artificial Intelligence (cs.AI)
备注: Project page: this https URL
Abstract:Cowork agents navigate dependencies across tools and files while adapting to changing user requirements and edits to shared workspaces. Scaling such environments requires maintaining consistency among task objectives, executable resources, user behavior, and success criteria. We introduce Cowork-Gym, a framework that synthesizes Cowork environments through a build-execute-revise loop that uses execution evidence and review feedback to validate and repair generated environments. To capture realistic Cowork dynamics, a stateful user simulator responds to the agent’s actions through dialogue and direct edits to the shared workspace. We synthesize 64.9K environments across 14 application scenarios and use them to post-train 9B, 27B, and 35B-A3B Qwen models with supervised fine-tuning, reinforcement learning, and on-policy distillation. Averaged across six agentic benchmarks, these models outperform their corresponding base checkpoints by 8.7, 6.4, and 4.1 points, respectively. Ours-27B ranks second overall among the evaluated models and achieves the highest average scores on PawBench and ClawEval, outperforming models with over 100B total parameters on both benchmarks.
[AI-134] PrT-Flow: Probability-Tractable Flow Matching for Reinforcement Learning
链接: https://arxiv.org/abs/2610.17667
作者: Congzhong Sun,Ning Tan
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Diffusion models and Flow Matching can represent flexible distributions over robot actions, but many reinforcement learning algorithms also require the probability of each action. We introduce PrT-Flow (Probability-Tractable Flow Matching), which computes probability density while retaining multimodal, multi-step action generation. Its Jacobian-Tractable Transformer (JTT) makes each sampling step invertible and enables efficient computation of its Jacobian determinant. We train these finite steps with Conditional Euler-step Training (CET), which supervises their displacement over one step during sampling rather than their instantaneous velocity. We prove that, under certain assumptions, CET has the same minimizers as fitting the Euler-solver updates of Flow Matching. PrT-Flow achieves 96% success on Robomimic Lift after 300 epochs of imitation learning, matching the diffusion policy reference, and 96% success on OGBench Cube-Triple after 600k online steps. These results demonstrate that invertible networks with tractable probability densities can learn effective robot policies and support SAC-style actor-critic training. Our code is available at this https URL .
[AI-135] RoTA-Bench: Evaluating 3D Spatial Reasoning through Interactive Projection Alignment
链接: https://arxiv.org/abs/2610.17665
作者: Qingchuan Ma,Yuexiao Ma,Xiang Meng,Yixuan Zou,Tianyu Xie,Wang Chen,Yuhui Zeng,Yuhang Wu,Xiawu Zheng,Rongrong Ji
类目: Artificial Intelligence (cs.AI)
备注: 31 pages, 2 figures
Abstract:Can vision-language models turn 2D observations into 3D understanding that supports precise action? This requires inferring object geometry, anticipating the effects of rotation, and refining actions through visual feedback. We introduce RoTA-Bench to evaluate these abilities through interactive 3D-to-2D projection alignment. Given a target silhouette, a model observes a rotated object and chooses successive rotations to recover the target projection within a limited action budget. Our automated pipeline constructs silhouette-preserving 3D objects from 2D icons and uses reversible initialization to guarantee solvability, yielding 674 tasks and a stratified 100-task test set. Empirically, we find that (i) a substantial human-model gap persists: GPT-5.6-Sol succeeds on only 17% of tasks, whereas 20 undergraduate participants collectively complete the test set four times with a 100% success rate; (ii) image-space and pose-space progress can diverge: at matched states, the model more often improves silhouette overlap while worsening pose alignment than humans (19.4% vs. 10.9%); and (iii) fine-grained control is sensitive to interaction history: removing it leaves 96.8% of the model’s actions at a coarse 30-degree step size and sharply reduces observed task success. These findings highlight the challenge of converting visual evidence into precise 3D control, beyond recognizing or matching individual images.
[AI-136] Can Agents Truly Forget? Explore Machine Unlearning for LLM Agents
链接: https://arxiv.org/abs/2610.17659
作者: Junkai Lin,Ge Xu,Zheng Xu,Junkai Chen,Xiangyuan Lan
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model (LLM) agents have shown strong capabilities in interactive problem solving, tool use, and long-horizon task execution. However, they may also acquire sensitive or private knowledge during training and expose it during deployment. Retraining a model from scratch to remove such knowledge is prohibitively costly, making machine unlearning a more practical approach. Machine unlearning aims to remove targeted undesirable knowledge from trained models without retraining from scratch while preserving general utility. When language models are deployed as agents, parameter-level unlearning becomes insufficient: forgotten information may still persist in memory artifacts or be reacquired via tool calls. In this paper, we formulate the problem of agent unlearning as removing private information from an LLM-based agent across three information channels: parametric knowledge, persistent memory, and tool-call acquisition. To study this problem, we introduce we introduce Agent Unlearning Benchmark (AUBench), a controlled benchmark that extends the fictitious-author setting of TOFU into an agent environment with persistent memory, an offline web snapshot, tool-accessible documents, and provenance-tracked interaction trajectories. By evaluating existing parameter-level and memory-level unlearning methods in this setting, AUBench reveals the gap between model-level forgetting and system-level agent unlearning, and provides a foundation for developing unlearning methods that operate across the full lifecycle of LLM agents.
[AI-137] When Should Enterprise Policy Decisions Be Learned and When Should They Be Reason ed?
链接: https://arxiv.org/abs/2610.17643
作者: Hyojung Han,Ryangkyung Kang,Youngeun Kim,Seung-Hun Jeon
类目: Artificial Intelligence (cs.AI)
备注: 43 pages, 17 tables, 9 figures
Abstract:An enterprise agent that answers under a written policy must choose an action before it says anything: answer, ask for a missing fact, ask for confirmation, call one or several tools, escalate, or refuse. Holding one 27B Korean enterprise model fixed, we vary whether the decision is a generated JSON object or a single constrained token, and whether the model may reason first. On 2800 in-distribution items we did not detect an accuracy difference between the two deployed paths (93.46% vs 94.14%), while they differ about tenfold in restart stability (3.00% of actions flip vs 0.29%); they also differ in reasoning, so this is an operational comparison. On a sealed benchmark of 300 items whose labels come from a hidden executable policy, reasoning is worth +17.3 points [+12.3, +23.0] on exact action. With reasoning off, the format difference was 0.0 points [-4.3, +4.0], a failure to detect an advantage for either interface, not equivalence; with reasoning on it is unresolved. Against a token-matched random control, boundary-targeted supervision moves the two covered boundaries by +17.5 and +20.0 points and leaves a held-out boundary unchanged. In a 3-seed training x reasoning factorial, the sealed-rendering interaction is -4.9 [-8.1, -1.7], consistent with partial substitution. All supervision was Korean. On independently rendered English versions, the model is 0.82 points lower [-2.68, +0.93] with reasoning on, the same order as re-rendering in Korean; the reasoning-off gain from targeted supervision is not stable in either language, so we do not claim transfer. The original text stays private as the confirmatory set. An open-model re-rendering made after these results reproduced the reasoning effect and the English verdict; the interaction kept its sign but did not replicate under item-clustered intervals. We release the model and that re-rendering with gold labels. Comments: 43 pages, 17 tables, 9 figures Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2610.17643 [cs.AI] (or arXiv:2610.17643v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2610.17643 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Ryangkyung Kang [view email] [v1] Tue, 13 Oct 2026 03:14:12 UTC (566 KB) Full-text links: Access Paper: View a PDF of the paper titled When Should Enterprise Policy Decisions Be Learned, and When Should They Be Reasoned?, by Hyojung Han and 3 other authorsView PDFHTML (experimental)TeX Source view license Additional Features Audio Summary Current browse context: cs.AI prev | next new | recent | 2026-10 Change to browse by: cs References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[AI-138] Beyond Bias Scores: Answerability Confounds in Activation Steering with Gradient-Constrained Direction Design (GCD)
链接: https://arxiv.org/abs/2610.17641
作者: Weitong Li,Donglin Song,Junkai Lin,Xi Xuan
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Bias scores in language-model evaluations often conflate what a model prefers with whether it answers at all. We identify this confound in \textscBBQ, whose ambiguous score factorizes into answerability and signed preference. Across 11 categories, 28,824 items, and five models, contrastive steering lowers the signed ambiguous score in 11/11 categories but lowers the signed disambiguated preference score in only 8/11. Across 99 configurations, the answerability channel exceeds half of the observed score change in 17% and accounts for the entire change in 11%. In a six-category disambiguated evaluation, 20.8% of 1,500 items flip to the unknown option, where abstention is incorrect; \textscStereoSet shows no analogous answer-rate shift. The effect reflects selection and near-orthogonality between the contrastive and preference-gradient directions. We introduce Gradient-Constrained Direction Design (\textscGCD), an option-gradient construction with a first-order unknown-logit constraint. Under a matched Qwen-7B protocol, GCD reduces answer-rate distortion 14-fold and yields a 4-fold larger preference change, while ablations show similar gains for alternative and unconstrained option-gradient directions; nulling the unknown gradient alone is not responsible for the gain. We recommend reporting answerability alongside published scores.
[AI-139] he Tokens Strike Back: Safety-Aware Denoising with Directional Embedding-based Revision
链接: https://arxiv.org/abs/2610.17636
作者: Amman Yusuf,Jiwoo Shin,Ruizhi Deng,Heitor Rodrigues Guimarães,Nikita Dvornik,Mijung Park
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 43 pages, 4 figures
Abstract:The Safety-Aware Denoiser (SAD) is a recent, training-free, safety-guidance framework for text diffusion models. During inference, SAD steers text samples toward safe regions of the token probability space. In this paper, we propose \mathitSADDER (Safety-Aware Denoising with Directional Embedding-based Revision), which improves the safety and quality trade-offs of SAD by exploiting a safety-guidance signal in the embedding (or activation) space for each token position, then updating the token positions using the SAD update rule multiplied by the guidance signal. This multiplication acts as a gating mechanism: tokens with a high unsafe signal are more likely to be modified during inference following the SAD update rule. However, such an update does not automatically preserve the conditional mean, i.e., the denoiser of a fixed target distribution. We therefore theoretically derive new position-wise target distributions for which the modified denoiser is exact. Furthermore, we derive conditions under which the position-wise target distributions match a global target distribution. We test our \mathitSADDER method on popular Masked Diffusion Language Models (MDLMs) and Uniform-State Diffusion Models (USDMs) across unsafe continuation and jailbreak settings, and show that \mathitSADDER achieves a better balance between safety, generation quality, and inference cost than existing methods.
[AI-140] OpenDepth: Repurposing Image Editing Priors for Robust Open-World Monocular Depth Estimation
链接: https://arxiv.org/abs/2610.17610
作者: Mengke Zhang,Tianyi Yang,Xinzhuo Li,Anlong Ming
类目: Artificial Intelligence (cs.AI)
备注: Accepted at ACM Multimedia 2026. 18 pages, 9 figures, including supplementary material
Abstract:Recently, the integration of generative priors has revolutionized Monocular Depth Estimation (MDE), achieving state-of-the-art (SOTA) performance with remarkably high data efficiency. However, although methods based on Text-to-Image (T2I) models excel in general scenes, their efficacy degrades sharply when confronted with open-world environmental perturbations, such as illumination variations and adverse weather. Since T2I priors are optimized for semantic synthesis rather than preserving pixel-level structure, they struggle to decouple geometry from appearance under severe visual shifts, easily triggering structural hallucinations. Therefore, our core insight is to model open-world depth estimation as consistent geometric inference across multi-style images, fully leveraging the powerful structure-preserving editing priors of Image-to-Image (I2I) models. Based on this, we propose OpenDepth, a robust open-world MDE framework that introduces “Intra-Domain Awareness” – the capability to map diverse stylistic inputs sharing identical geometry into a unified latent space. To achieve this, we first align shallow features within the Diffusion Transformer (DiT) to anchor degraded observations to unified geometric representations. Additionally, we introduce a joint RGB latent reconstruction task, requiring the model to recover clean RGB information from degraded inputs. This multi-stage strategy effectively decouples environmental style interference, compelling the model to focus strictly on underlying geometric recovery. Extensive evaluations show that OpenDepth sets new SOTA records for robust zero-shot depth estimation across standard and extreme weather benchmarks, demonstrating exceptional structural integrity and adaptability.
[AI-141] Verifier-Harness Co-Evolution for Training-Free Self-Improvement on Open-Ended Tasks
链接: https://arxiv.org/abs/2610.17609
作者: Tianyi Xu,Yaolun Zhang,Yiran Wu,Joe Nguyen,Shengyu Dai,Saba Moeinizade,Zhenwen Shao,Qingyun Wu,Huazheng Wang
类目: Artificial Intelligence (cs.AI)
备注: 47 pages, 5 figures
Abstract:Training-free self-improvement lets an agent use feedback to revise the response it is about to return and to update the harness state that shapes how it handles later tasks. On open-ended tasks that feedback rarely comes from reference answers or executable tests, so the agent has to rely on a model-based verifier. When the verifier is wrong, the current response can get worse, and misleading experience can persist across the task stream. A verifier that never changes also grows stale as the agent’s responses improve and task demands shift. We introduce Unified Co-Evolution (UCoE), a training-free method that evolves an executable verifier program together with the agent’s harness state while all model parameters stay fixed. UCoE synthesizes candidate verifier programs and tests them on response contrasts, pairs of responses that differ in one declared property and have a known preferred side. It selects the candidate whose discrimination on its weakest contrast family is highest, a quantity we call driving capacity. The selected verifier diagnoses one weakness in each draft, guides a single repair, decides which response is served, and determines whether the accepted repair may enter the harness state. As tasks complete, their diagnostics seed new candidate verifiers, which must beat the current verifier on the same contrasts before adoption. Across six open-ended benchmarks, UCoE averages 59.00% preference against matched direct generation with Qwen3-8B and 57.19% with Qwen3.8-27B, margins of 9.00 and 7.19 percentage points. On three verifiable benchmarks, the same method raises the mean from 36.06% for direct generation to 38.48%. Code is available at this https URL.
[AI-142] Revisiting the 16x Pretraining Compute Gap between Diffusion and Autoregressive Language Models
链接: https://arxiv.org/abs/2610.17590
作者: Guanghan Wang,Bing He,Zhan Shi,Benoit Dumoulin,Hanqing Lu,Volodymyr Kuleshov
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: project page: this https URL
Abstract:Diffusion language models (DLMs) have attracted wide attention for their fast, parallel decoding, but they are believed to be far less efficient to train than autoregressive language models (ARMs): a widely cited study reports that a DLM needs roughly 16x the training compute of an ARM to match its likelihood, a figure that has made pretraining DLMs from scratch look prohibitive at scale. We find that the methodology behind this figure has fallen out of step with current DLM practice: it scores the model with its evidence lower bound (ELBO) and trains a full-sequence, decoder-only model. We therefore revisit the scaling laws of DLMs, adopting the exact likelihood under the model’s own decoding order, block diffusion training, an encoder-decoder architecture, and a sweep over the block sizes used in practice. Under this new protocol, we study across ten model variants, about 1,700 models in total, and find that pretraining DLMs is far cheaper than previously thought. Concretely, scoring with the exact likelihood reduces the compute multiple to 12.8x; block diffusion training and an encoder-decoder architecture further reduce it to 7.8x at block size 32, and smaller block sizes narrow it to 4.6x at block size 8 and 3.0x at block size 4. We complement these scaling analyses with an empirical rule for the optimal decoder depth of encoder-decoder block diffusion, a held-out compute budget beyond the fitting range at which the gap between variants persists, and downstream benchmark results where likelihood improvements carry over to accuracy. Overall, the 16x figure substantially overstates the pretraining cost gap between DLMs and ARMs, and pretraining DLMs from scratch is far more viable than previously assumed.
[AI-143] aming Softmax at the Limits of Muon: A Unified Approach through Hilbert Geometry
链接: https://arxiv.org/abs/2610.17588
作者: Aditya Somasundaram,Charles Guille-Escuret,Alexander Moreno,Zhengzhong Liu,Eric Xing
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Muon controls weight updates through the RMS movement of a linear layer’s output. This implicitly treats the linear output as a proxy for the full module, including the nonlinearity that follows. This proxy can be misleading for softmax, where the RMS shift in activations does not faithfully measure the change in the output distribution. We propose Hilbert distance as a natural metric for bounding changes in these output distributions. Hilbert distance bounds changes in cross-entropy loss and limits probability redistribution. For softmax, it takes the form of the variation seminorm of the preactivation perturbation, allowing probability constraints to be expressed as parameter constraints. Formulating the steepest descent problem with a Hilbert distance constraint brings two recently proposed methods under a unified mathematical framework. For the LM head, solving a row-wise approximation to the resulting diameter-constrained problem leads us to RowNorm \citeppethick2025trainingdeeplearningmodels. For scaled dot-product attention, the same principle yields the first-order query–key steepest descent problem underlying Compositional Muon \citepkeigwin2026compositional. These connections motivate using Hilbert distance as a guiding principle for optimizer design in softmax modules.
[AI-144] RAN-JEPA: An Action-Conditioned World Model for Energy-efficient Mobile Network Control
链接: https://arxiv.org/abs/2610.17584
作者: Binghan Wu,Kai Zeng,Kai Yang,Ya-Qin Zhang,Joseph Sifakis,Ye Ouyang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Operational mobile networks are rich in observations but poor in interventions: they continuously generate large volumes of passive telemetry, while controlled action–response trajectories remain comparatively scarce and costly to collect. This observation–intervention asymmetry is a central obstacle to building practical world models for RAN control. We address this problem with RAN-JEPA, an action-conditioned world model that separates representation learning from intervention learning. The model first learns a predictive latent representation from observation-only network data, adapts that representation to the target deployment domain, and then learns action-conditioned latent dynamics from controlled sleep/wake trajectories while keeping the encoder fixed. We evaluate RAN-JEPA using public operator performance data for pretraining and a real-equipment four-cell 5G testbed for domain adaptation, post-training, and control. Across four future 15-minute intervals, RAN-JEPA achieves lower action-conditioned prediction error than a GPT-2 world-model baseline and a mechanistic rule, while remaining close to a strong persistence reference. What-if evaluation further shows that the model produces distinct and coherent future trajectories under matched SLEEP and KEEP plans. In closed-loop control, RAN-JEPA consistently supports higher sleep ratios than the two baselines at comparable congestion cost. These results show that observation-first, action-conditioned learning can turn abundant passive telemetry and limited intervention data into a practical predictive model for energy-aware RAN control.
[AI-145] Chasing Label Shifters: A Change-Aware Framework for Dynamic Graph Node Classification NEURIPS2026
链接: https://arxiv.org/abs/2610.17582
作者: Junghoon Kim,Seungyoon Choi,Hyunsung Kim,Chanyoung Park
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: NeurIPS 2026
Abstract:Entities in real-world systems often evolve over time: users shift between information consumption patterns, researchers migrate between fields, and firms transition between financial risk states. When such systems are modeled as dynamic graphs, these transitions correspond to nodes whose class labels change between consecutive snapshots, which we call shifters. Correctly classifying shifters is often more consequential than classifying stable nodes with persistent labels, as detecting such transitions enables timely intervention in high-stakes settings. Yet, we identify a systematic failure mode across all major dynamic graph neural network architectures: shifter performance consistently lags behind stable nodes, with errors concentrated on the old label. Our structural analysis traces this failure to embedding inertia: neighborhood aggregation keeps shifter representations anchored to the old-class centroid, and this effect is strongest precisely where it is hardest to correct, namely for shifters whose neighborhoods remain aligned with the old class. Guided by this finding, we propose CHASE, a model-agnostic CHange-Aware framework for Shifting nodEs that wraps any dynamic GNN with targeted components for detecting label shifts and overriding stale neighborhood signals. CHASE consistently improves shifter performance across all tested models and datasets (up to 114.9% and 202.5% improvement on accuracy and F1 score, respectively) while preserving stable-node performance. We additionally contribute three dynamic graph benchmarks with naturally shifting labels, filling a gap in existing resources. The source code is available at this https URL
[AI-146] acitVLA: Reasoning Vision-Language-Action Models Without Reasoning Latency
链接: https://arxiv.org/abs/2610.17540
作者: Jack Lu,Yurong You,Boyi Li,Yulong Cao,Thomas Tian,Chaowei Xiao,Xiangyu Chen,Wenjie Luo,Boris Ivanovic,Marco Pavone,Yan Wang
类目: Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: Project page: this https URL
Abstract:Textual reasoning before acting has improved vision-language-action (VLA) models, but at the cost of latency: every action conditioned on the reasoning text must wait for autoregressive decoding. We show that actions do not need to wait for or condition on the generated reasoning to benefit from it, because the reasoning supervision already improves the backbone representation. We introduce TacitVLA, a reasoning VLA whose action expert never attends to the reasoning it generates for the current call. On long-horizon tasks where the reasoning text also serves as episodic memory, it can generate each entry during action execution rather than before acting. We evaluate it in a controlled study on four benchmarks spanning manipulation and autonomous driving, memory and non-memory tasks: RoboMME, RMBench, LIBERO, and PhysicalAI-AV. On all four, TacitVLA matches the performance of its explicit-reasoning counterpart at up to 3.4x lower latency, and matches the latency of the same architecture trained without reasoning at up to +17.1% success. In absolute terms, it surpasses the best previously published results on the RoboMME Counting suite (82.7% vs. 66.8%) and RMBench (77.5% vs. 56.5%), and reaches 72.9% on LIBERO-Plus and a 52.5% scene score on the public split of the PhysicalAI-AV closed-loop driving challenge. Project page: this https URL.
[AI-147] owards Green AI: Evaluating Energy Efficiency for Quantized LLM Inference
链接: https://arxiv.org/abs/2610.17531
作者: Petar Basta
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注:
Abstract:While it is known that quantization of large language models (LLMs) reduces memory usage via lower-bit weights, studies quantifying the resulting impact on energy usage and carbon intensity are scarce. We present the first unified evaluation of weight-only quantization strategies for LLM inference across varying input/output lengths and GPUs, highlighting differences in energy efficiency in addition to accuracy degradation and runtime. We study quantized inference on Llama 2 7B, Phi 3.5 Mini Instruct, OLMo 1B, Qwen 2.5 7B, and Qwen 2.5 14B across GLUE MNLI, MMLU, HumanEval, and GSM8K datasets, evaluating 10 post-training quantization (PTQ) strategies on NVIDIA H100, A6000, A100, and L4 GPUs. We identify a novel trend showing that quantization techniques tend to exhibit peak energy efficiency relative to full-precision baselines when inputs are sufficiently long and outputs are short. Furthermore, we show that quantization strategies become marginally more energy-efficient relative to full-precision models as batch sizes increase, though gains are modest. Notably, fused-kernel implementations such as EETQ int8 and Bitsandbytes int8 offer the highest energy savings, up to 4x compared to FP32 on short text generation, with negligible accuracy loss. Finally, we observe that energy usage closely tracks runtime on our evaluated benchmarks, indicating that, in practice, latency optimization can serve as a reliable proxy for reducing the environmental footprint of LLM services. We conclude with suggested directions for strategically selecting low-carbon quantization strategies based on specific inference requirements.
[AI-148] Box-WAM: Box Trajectories as Compact State Representations for World Action Models
链接: https://arxiv.org/abs/2610.17526
作者: Qinxi Yu,Wenzhou Ding,Gloria X. Wang,Albert J. Zhai,Shenlong Wang
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:World action models (WAMs) are a category of policy models that ex- plicitly predict both actions and future world states. They have been shown to outperform action-only models across various robotic manipulation tasks. How- ever, representing the future with full-frame video prediction can be computation- ally expensive during both training and inference. This raises a central question: what compact state representation preserves the planning and supervision bene- fits of future prediction? In this paper, we propose BoxWAM, which replaces dense future frames with oriented bounding box trajectories of task-relevant ob- jects. By explicitly imagining object-centric future motion, our method provides a lightweight prediction target that captures the key physical changes induced by robot actions. We show that box trajectory prediction substantially reduces com- putational cost while improving action prediction performance, suggesting that object-centric compact states can serve as an efficient alternative to video-based world action modeling for robotic manipulation. Result videos can be found on this https URL
[AI-149] Enhanced Detective: A Dual-Stage Proactive Defense Against Internal Poisoning Attacks in Financial Fraud Detection Using Federated Learning
链接: https://arxiv.org/abs/2610.17523
作者: Maria-Ioana-Andrea Ivan,Orhan Ermis,Ruxandra F. Olimid
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:With recent developments in Artificial Intelligence (AI), the adoption of distributed learning paradigms such as Federated Learning (FL) has emerged. In simple terms, FL enables multiple entities to collaboratively train AI models without sharing local data. The FL paradigm is particularly valuable in privacy-sensitive settings, such as financial services, because it enables institutions to collaboratively improve fraud-detection models while keeping sensitive transaction data local. However, FL schemes are prone to security vulnerabilities, where malicious clients can poison local data or model updates and degrade the global model, and hence the performance of the resulting model might be affected. Therefore, in this work, we propose Enhanced Detective, a dual-stage robust aggregation strategy that combines Median Absolute Deviation (MAD), distance-based update filtering, and FedMedian aggregation. Experiments on an imbalanced credit card dataset evaluate targeted label-flipping and weight-scaling attacks over 50 communication rounds. With 40% malicious clients, Enhanced Detective achieves an AUPRC of 0.796 and a Recall of 0.847.
[AI-150] YUBI-Real2Sim: 3DGS Scene Verification for Reliable Policy Evaluation
链接: https://arxiv.org/abs/2610.17521
作者: Hiroki Nishizawa,Takehiko Ohkawa,Nicola Scianca,Tatsuya Matsushima,Shigeo Morishima,Kei Ota
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: Under submission
Abstract:Evaluating VLA policies on real robots is costly, time-consuming, and difficult to reproduce, making real-to-sim simulation essential for scalable evaluation. While 3D Gaussian Splatting (3DGS) enables rapid real-to-sim scene generation, the reconstruction quality required for reliable policy evaluation remains unquantified. In particular, policies relying on wrist-mounted fisheye cameras (e.g., YUBI handheld gripper data) are heavily exposed to visual artifacts (holes, floaters) in wide-angle views, directly degrading task performance. To address this, we present YUBI-Real2Sim, a unified framework for 3DGS-based scene reconstruction and real-to-sim policy evaluation under dual fisheye wrist views. We develop an integrated pipeline combining 200-degree fisheye camera capture, 3DGS training, 3D foundation models for object modeling, and MuJoCo-based physical simulation, reproducing the observed real-world policy ordering in eight of nine scene-task comparisons. Using this framework, we construct a benchmark across two bimanual robot platforms, three scenes, three tasks, and dual pi0.5-based policies. This systematically measures real-to-sim rank agreement and evaluates policy sensitivity against counterfactual 3DGS artifacts (ceiling holes, floaters, warping) and capture disturbances. We further analyze whether novel-view quality scores can indicate policy-ranking reliability under a scene-held-out evaluation protocol. The results show task-dependent performance and limitations of quality-based reliability prediction, highlighting the need to assess reconstruction quality when using Real2Sim for policy evaluation. These results do not yet establish a generally reliable scene-acceptance rule. Project Page: this https URL.
[AI-151] Foundation Models for Event Prediction Across Heterogeneous Event Logs
链接: https://arxiv.org/abs/2610.17503
作者: Fabian Spaeh,Sean Jaffe,Shandian Zhe,Bin Shen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Event logs are ubiquitous across real-world domains, including business processes, manufacturing workflows, healthcare, and many other operational systems. Yet event logs can differ substantially in their event semantics, label spaces, interaction patterns, and temporal scales, making it difficult to build a single foundation model that generalizes directly to previously unseen systems. We address this challenge with the Foundation Event Predictor (FEP), a simple and scalable in-context architecture that performs next-event prediction through learned, event-level nonparametric inference. Causal self-attention encodes context and target histories into key and query representations, respectively, while the observed next context events are encoded as values. Cross-attention then performs learned, relevance-weighted aggregation of next-event outcomes over all context transitions, allowing the entire nonparametric prediction procedure to be expressed as standard attention operations and efficiently parallelized with modern kernels such as FlashAttention. To promote cross-system generalization, we introduce permutation-robust embedding learning and random fixed embeddings to decouple event IDs from persistent semantics, together with empirical quantile transforms and robust normalization to handle diverse temporal scales. We pretrain FEP on 696 real-world event datasets spanning heterogeneous event systems and evaluate it on 50 held-out datasets unseen during pretraining. Across these datasets, FEP generalizes effectively to new event systems without parameter adaptation, frequently approaching the performance of strong models trained separately on each target dataset while substantially outperforming existing foundation-model baselines, particularly on challenging prediction tasks.
[AI-152] VeriPIE: Checkable Query Plans and Evidence-Guided Selection for Text-to-SQL
链接: https://arxiv.org/abs/2610.17493
作者: Piyush Jha,Cruise Song,Zhengyu Li,Yatharth Sejpal
类目: Logic in Computer Science (cs.LO); Artificial Intelligence (cs.AI)
备注:
Abstract:Financial Artificial Intelligence (AI) governance and model risk management (MRM) emphasize independent validation, transparency, and effective challenge of AI-generated computations. Text-to-SQL translates natural-language questions into Structured Query Language (SQL), but queries that agree on one database can still express different computations. We present VeriPIE, a neurosymbolic framework combining language-model generation with symbolic checks. It ensembles direct SQL generation with QuerySketch, a structured plan language checked by a backend implemented in Lean, a theorem prover and programming language. Plan diagnostics guide repair before SQL generation, while execution agreement and bounded searches for distinguishing databases inform candidate selection. These checks expose malformed plans and hidden disagreement without proving agreement with the user’s intent. In single-run experiments on the complete Spider development set, VeriPIE achieves 72.05% test-suite accuracy (745/1,034), compared with DeepEye’s 65.67% (679/1,034). Its two-query shortlist covers 78.43% (811 questions), compared with 74.66% for DIN-SQL (772) and 69.05% for DeepEye (714). On BIRD, VeriPIE reaches 68.90% execution accuracy and 73.73% two-query coverage. The central benefit is reviewable evidence about the generated computation: explicit plan checks, repair diagnostics, and concrete disagreements that support safer financial analysis.
[AI-153] REA: From Correct LLM Judgments to Safe Decision Delivery
链接: https://arxiv.org/abs/2610.17473
作者: Jingxiang Meng,Aofan Liu,Yaoshuai Ma,Fangxin Liu,Yongbiao Chen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Generative safety guards must classify requests correctly and return outputs that applications can parse. Reasoning guards can improve classification accuracy while producing outputs that applications cannot parse. When applications cannot extract a correct prediction, fallback handling can allow harmful requests or block safe ones. We introduce Resolve–Express–Act (REA), a framework that traces judgments through parsing and fallback to application actions. Experiments on classifiers and safety guards show that correct predictions can appear in incorrectly formatted outputs. Parsing the safety verdict preserves the reasoning guard’s accuracy advantage, while stricter explanation requirements reject additional correct predictions. Accepting formatting variations recovers most rejected correct predictions. Held-out tests confirm that output requirements and generation limits affect whether correct predictions can be parsed. In our fail-open replay, matching the parser to the output and fixing formatting reduced the fraction of harmful requests allowed despite correct guard judgments from 95.5% to 0%. A fixed-output comparison shows that model-based extraction can preserve guard verdicts, but is not automatically reliable; deterministic normalization already repairs the observed native-format failures. Fallback replay shows that the same correct prediction can lead to different safety actions depending on the parser and fallback policy, including allowing requests the guard correctly identifies as harmful.
[AI-154] Assessing the Performance of AI Models in Misinformation Detection Using Predefined Social Noise Constructs
链接: https://arxiv.org/abs/2610.17469
作者: Manar Alsaid
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 22 pages, 12 tables, 2 figures and one Appendix
Abstract:Social media platforms generate vast volumes of user-generated content that often obscure meaningful discourse, a phenomenon conceptualized as social noise. While existing misinformation detection methods primarily focus on factual verification, they largely overlook the role of social noise in amplifying misinformation. This study assesses the performance of six text-based models and a graph neural network (GNN) in mitigating the effects of misinformation using predefined social noise constructs. The dataset consists of 500,000 tweets collected from the X platform using the Ukraine hashtag during the Russia-Ukraine conflict from March 2024 to March 2025. Advanced machine learning models achieve strong performance, with accuracies ranging from 79.94% to 97.17%. RoBERTa demonstrates the highest performance, with an accuracy of 97.17%, Cohen’s kappa of 0.9660, MCC of 0.9660, and AUC of 0.9977. The GNN achieves an accuracy of 64.52% and an AUC of 0.899, demonstrating performance above random-chance classification by exploiting the relational structure between tweets. All text-based models achieve kappa values greater than 0.75, indicating substantial overall agreement. Explainable AI analysis using LIME reveals that 89.6% of predictive weight is attributed to contextual features rather than isolated keywords. The findings highlight that integrating social noise into AI pipelines enhances misinformation detection and supports more effective mitigation strategies in large-scale social media environments.
[AI-155] An FPGA-Based Spiking Neural Network for Spatiotemporal Pattern Learning and Classification
链接: https://arxiv.org/abs/2610.17463
作者: Jingang Jin,Yue Ma,Zhenhang Zhang,Qinru Qiu
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 6 pages
Abstract:Spiking Neural Networks (SNNs) are promising for energy-efficient spatiotemporal pattern recognition at the edge, but efficient deployment and on-device adaptation require specialized hardware. In this work, we present an FPGA implementation of multilayer SNNs for spatiotemporal pattern classification, in which both inference and training take place on the same hardware. Our design adopts a temporal-forward-only training algorithm that avoids the memory and latency overhead associated with traditional Backpropagation Through Time (BPTT). A mixed-precision scheme balances training stability with resource efficiency, and architectural optimization maximizes hardware and memory utilization. We further identify and resolve hazards that arise when inference and learning share the same memory, including port conflicts between the forward and backward access patterns and read-after-write dependencies in floating-point potential accumulation. Experimental evaluations on MNIST and five multivariate time-series classification datasets demonstrate that, with fully on-chip training, our SNN model achieves accuracy comparable to SNN or LSTM trained with BPTT. Compared to CPU and GPU based implementations, our FPGA design delivers 2.5 to 6 times higher throughput while consuming an order of magnitude less energy per sample.
[AI-156] Large Language Models Extend Known Research But Do Not Create New Research Directions
链接: https://arxiv.org/abs/2610.17439
作者: Zhihua Liu,Kurt Butler,Damian Machlanski,Danilo Bueno,Sotirios A. Tsaftaris
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:The success of autonomous research systems relies on the quality of the scientific hypotheses they generate. Recent studies report that human reviewers often rate LLM-generated ideas as more novel than those from human experts, raising hope that AI can accelerate scientific discovery. Yet it is unclear whether these models can do more than extend existing knowledge. Can LLMs drive genuine innovation by revealing new research directions, or are they limited to repeating and recombining what science already knows? This unresolved question lies at the heart of discussions about AI and the future of science. To address this, we distinguish two types of novelty: distance, which measures how far a hypothesis lies from existing work, and direction, which captures whether the hypothesis pursues new lines of inquiry beyond those already present. These two measures diverge sharply for LLM-generated work. We represent scientific papers in a shared conceptual space using a pretrained scientific-paper embedding model. Within this space, we embed 26.9 million human-authored papers across 1,325 research topics in six natural-science fields and 524,308 LLM-generated hypotheses from eight models. Distance is measured using existing metrics. To quantify direction, we introduce the orthogonal novelty score (ONS). LLMs consistently generate ideas more distant from existing work than human-authored papers, up to 16% further when the LLM is given the papers farthest from the centre of the topic. However, LLM hypotheses have a 15–23% lower median ONS than published papers from the same topic. Thus, LLMs move further from what is already known, but remain within established directions. This pattern persists across changes in various configurations. As a result, LLMs excel at extending the frontier but the creation of new directions is still led by human scientists.
[AI-157] SchemaWeaver: Accelerating On-Device Mobile Agent ic AI via Schema-Guided Parallel LLM Decoding
链接: https://arxiv.org/abs/2610.17436
作者: Boyuan Yang,Wei Gao
类目: Artificial Intelligence (cs.AI)
备注: 16 pages, 14 figures
Abstract:Large language models (LLMs) on mobile devices can act as agents that call external software tools to complete complex tasks, but the major challenge of supporting on-device mobile agentic AI with LLMs is the long latency of LLM tool calling on mobile devices. In particular, the decode phase is the main contributor of such long latency, but is hard to speed up on mobile processors as it is bound by memory bandwidth rather than compute power. To address this limitation, in this paper we present SchemaWeaver, a new on-device mobile agentic AI technique that enforces schema-guided parallel decoding in the decode phase of LLM tool calling, to reduce its latency by producing multiple tokens in each decode pass. SchemaWeaver can be generically applied to the LLM inference engine on mobile devices without involving any training efforts or changes on LLMs, and our experiment results on major mobile SoCs and LLM families show that SchemaWeaver can reduce the LLM tool calling latency by up to 3.9 times, compared to the competitive baselines.
[AI-158] Reinforcement Learning In Partially Verifiable Environments for Colon Cancer Treatment Planning
链接: https://arxiv.org/abs/2610.17426
作者: Li-Ching Chen,Andrés Franco Valiente,Yuta Ishiyama,Zachary R. Chalmers,William Chen,Kelsey Natsuhara,Taema Brinjikji,Madhumita Sushil,Travis Zack,Julian C. Hong
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Clinical decision-making is a sequential process of information collection, clinical inference, and forecasting under partial observability, often without a uniquely correct treatment decision. Large language models (LLMs) have demonstrated strong medical question-answering capabilities but extending these advances to clinical reasoning remains challenging because many clinical tasks are only partially verifiable. Recent clinical reinforcement learning (RL) work has therefore largely focused on diagnostic endpoints, where correctness can be more directly verified. Oncology treatment decision making is a particularly challenging instance of this problem that extends beyond medical question-answering in which multiple decision trajectories can be clinically defensible for the same partially observed patient state. We formulate oncology treatment planning as an interactive RL problem under partial observability and construct patient-specific, process-aware rewards from decision processes observed in real world clinical records. The agent must learn to sequentially acquire information from a structured patient representation before committing to a first-line treatment recommendation, and is evaluated on information gathering, treatment selection, and reasoning alignment. Using GRPO and GDPO to optimize Llama-3.1-8B-Instruct within this environment, RL substantially improves the base policy beyond frontier model capabilities. The GDPO-trained policy increases Mean@3 scalar clinical reward by 79.60% over base Llama and exceeds GPT-5.2 by 18.82% under the same interactive environment, while reducing query count during training. Overall, our results show that patient-specific, partially verifiable rewards can provide an effective learning signal for interactive clinical RL when both the patient state and the underlying clinical decision process are only partially observed.
[AI-159] On Making mathrmSE(2)-Invariant Networks Optimal NEURIPS2026
链接: https://arxiv.org/abs/2610.17424
作者: Tomas Karella,Emily Shinkle,Alice Elisabeth Anastasia Allen,Pieter J. Swart,Nicholas Lubbers,Roxana Bujack
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted to the main track of NeurIPS 2026
Abstract:Respecting a-priori known symmetries of underlying data is a key principle in designing efficient neural architectures. Convolutional neural networks are a canonical example, encoding translation equivariance. Extending this inductive bias to richer symmetry groups such as \mathrmSE(2) in the most effective way possible is an ongoing architectural and computational challenge. Building on classical invariant theory, we derive invariant feature maps that are provably optimal: minimal and information-preserving with respect to the symmetry group. Our construction is designed to integrate seamlessly into standard architectures, requiring no modification to nonlinearities or normalization layers. As a concrete implementation, we introduce a plug-and-play module for convolutional neural networks that enforces \mathrmSE(2) invariance, capturing both translations and rotations within standard architectures. The resulting models are both efficient and expressive. Experiments show competitive or improved accuracy compared to state-of-the-art equivariant methods at significantly lower computational cost. Finally, we provide a theoretical and empirical analysis of the relationship between equivariance and invariance, offering insight into their roles in controlling robustness and flexibility.
[AI-160] ClearNav: Perception-Aligned Quadrotor Flight Using Differential-Physics
链接: https://arxiv.org/abs/2610.17419
作者: Aditya Azad,Arthur Huang,Guanrui Li
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Autonomous quadrotor flight in cluttered environment is an key capability for tasks like inspection and search. Navigation policies trained in simulation and transferred to hardware can fly at high speed in open space, but they can still collide as obstacle clearance shrinks. We attribute most of these collisions to a single mechanism, namely motion in directions the onboard camera does not observe. Heading that is not aligned with the direction of travel and lateral excursion during obstacle avoidance both carry the vehicle toward obstacles outside its field of view. We introduce ClearNav, a navigation policy trained with two reward terms that constrain motion to the observed direction, a forward-looking heading target and a penalty on lateral velocity. A third term penalizes speed near obstacles so that an obstacle entering the field of view remains avoidable. In simulation, across 18 held-out scenarios with 2304 paired flights per policy unseen in training, ClearNav improves success by 6.8 points, cuts the collision rate by 65% relative to the state of the arts, and flies at maximum 5m/s. For the proximity term alone under matched commanded speeds, the collision reduction still holds at 70%. The effect is significant on static scenes and consistent in direction on moving obstacles, though not significant there at the suite level, showing that safer flight in clutter is possible even on the limited sensors and compute of small drones. We also validate our navigation policy in a real-world cluttered environment, showing that it enables the quadrotor to navigate through the cluttered environment successfully and safely.
[AI-161] Why Agent ic Workflows Fail: A Taxonomy Derived from Literature and Empirical Data
链接: https://arxiv.org/abs/2610.17418
作者: Xin Li,Rami Bahsoon,Tao Chen
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:Agentic workflows combine large language models with planning, memory, tools, execution environments, and iterative feedback. Their failures are difficult to diagnose because the same user-visible symptom may originate from model behavior, workflow orchestration, external tools, infrastructure, or their interaction. This paper presents a source-aware taxonomy of agentic workflow failures and applies it to evidence from academic literature and GitHub. The taxonomy combines source attribution with three empirical dimensions: a concrete failure-type dictionary, workflow-stage localization, and agent-capability localization. Its central structure is a two-axis localization matrix: workflow stages form the horizontal process axis and agent capabilities form the vertical functional axis. This design moves beyond taxonomies that mainly classify final failure manifestations by localizing intermediate process states inside agent operation. Our empirical comparison shows that literature-derived failures emphasize planning, reasoning, learning, perception, and knowledge-related mechanisms, whereas GitHub-derived failures concentrate on execution-stage, actuation-related, tool, API, and infrastructure failures. The dominant hotspot is iterative execution and actuation. These findings indicate that agentic workflow reliability should be treated as a systems problem rather than only a model-output problem. The taxonomy provides a basis for structured evaluation, root cause analysis, failure injection, and engineering triage.
[AI-162] Cross-View Time-Frequency Diffusion for Extreme-Aware Time-Series Generation
链接: https://arxiv.org/abs/2610.17407
作者: Lin Jiang,Rongchao Xu,Guang Wang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 42 pages, 17 figures
Abstract:Generating realistic time-series requires faithfully reproducing rare but high-impact extreme events. Existing approaches often rely on specialized representations or extreme-focused training objectives, which may restrict the diversity of captured extremes or compromise overall distributional fidelity. We propose Cross-View Time-Frequency Diffusion (CrossTF), a framework that exploits discrepancies between time- and frequency-view denoisers for extreme-aware time-series generation. Our key empirical finding is extreme-sensitive cross-view asymmetry: moving along the direction connecting the two views’ clean-sample estimates substantially changes extreme fidelity while having limited effects on overall generation quality. We further identify a late-stage cross-view window in which this discrepancy stabilizes and cross-view adjustments have a particularly strong effect on extreme fidelity. These findings motivate a staged optimization strategy in CrossTF: after training and freezing both denoisers, it only needs to learn step-specific fusion parameters within this window using an Extreme Value Theory-based objective under an energy-distance constraint. This formulation supports interpolation between and extrapolation beyond the view-specific estimates while limiting degradation in overall generation quality. Extensive experiments on seven real-world datasets against 11 baselines show that CrossTF improves extreme fidelity by 26.1% and overall performance by 7.4% relative to the strongest baselines. These results highlight cross-view discrepancy as an effective basis for low-dimensional optimization in extreme-aware time-series generation.
[AI-163] Generalizable field of view extension for diffusion MRI via conditional flow matching
链接: https://arxiv.org/abs/2610.17401
作者: Gaurav Rudravaram,Lianrui Zuo,Ema Topolnjak,Michael E. Kim,Adam M. Saunders,Elyssa M. McMaster,Paula Trujillo,Ely E. Wesley,Colin D. McKnight,Pratik P. Pandharipande,Mayur B. Patel,Cayce Strength,Brenda T. Pund,Amelia W Maiga,Robert S. Dittus,David A. Bennett,Julie A. Schneider,Konstantinos Arfanakis, TheHABS-HD Study Team,Derek B. Archer,Timothy J. Hohman,Yelena Bodien,Daniel C. Moyer,Kurt G. Schilling,Bennett A. Landman
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Incomplete field-of-view (FOV) diffusion MRI (dMRI) acquisitions compromise whole-brain white matter analysis by excluding brain regions from tractography and connectivity studies. While deep generative models for dMRI FOV extension have been proposed, their generalizability remains limited by the acquisition-dependent nature of dMRI, where signal intensity varies with b-value, gradient direction, and scanner protocol, and the inherent 4D complexity of the data, making it difficult to train models that transfer across sites or acquisition schemes without retraining. We propose a flow-matching generative model that extends the dMRI FOV in both the superior and inferior directions by predicting apparent diffusion coefficients (ADC) in six directions, conditioned on the two nearest acquired slices and the full-FOV T1w image to enforce both local diffusion continuity and anatomical consistency. The ADC representation decouples the model from acquisition-specific b-values and gradient directions, providing a middle-ground across protocols. The model was trained on 3,103 subjects (5,105 sessions) across five different datasets spanning diverse acquisition schemes (b = 500 and 1000 s/mm ^2 ; 21–192 gradient directions), four single-shell (b = 1000) and one multi-shell (b = 500/1000). Evaluated on 299 subjects across five independent datasets spanning b = 750–1500 s/mm ^2 and 32–90 gradient directions, with single- and multi-shell schemes unseen during training, the proposed method achieves significantly improved bundle Dice score on downstream tractography (p 0.01).
[AI-164] Learning to Detect Shifts in Unstructured Data Streams via Monitoring-by-Betting
链接: https://arxiv.org/abs/2610.17399
作者: Siyu Zhou,Jialin Ouyang,Fan Xia,Adarsh Subbaswamy,Drew Prinster,Suchi Saria,Augusto Garcia-Agundez,Sarah Pollet,Jinoos Yazdany,Patrick Vossler,Jean Feng
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注:
Abstract:Continuous monitoring of generative AI systems requires detecting distribution shifts in the unstructured data they ingest and generate. Existing monitoring methods typically control false alarms by requiring the discrepancy measure used to detect shifts to be pre-specified. However, for unstructured data, pre-specification creates a sensitivity–breadth trade-off: a narrowly defined measure is sensitive to the anticipated shift but may miss unexpected changes, whereas a broadly defined measure may detect a wider range of shifts but with reduced sensitivity. Although adaptively hypothesizing the shift direction can overcome this limitation, such data-dependent adaptation generally invalidates conventional false-alarm guarantees. We reconcile adaptivity with statistical validity through testing-by-betting, reframing sequential changepoint detection as a betting game in which the bettor continually updates its hypothesized shift direction, or witness function, as new observations arrive. We introduce asymptotic e-detectors, which approximate the pre-change distribution using a calibration set of size m , and prove that they asymptotically control the average run length to a false alarm as m increases. Empirically, calibration sets as small as m=30 provide effective control. We then develop adaptive monitoring strategies ranging from black-box bettors that learn from embeddings to interpretable bettors based on LLM-proposed concepts. Across three real-world datasets, including data from a clinical AI scribe deployment, adaptive monitors detect shifts faster than fixed monitors at the same false-alarm level. Further, when the true shifts are sparse and interpretable, LLM-concept monitors provide the fastest and most interpretable detection, while also guiding root-cause diagnosis for complex shifts.
[AI-165] Learning to Play: Reinforcement Learning and Games
链接: https://arxiv.org/abs/2610.17391
作者: Aske Plaat
类目: Artificial Intelligence (cs.AI)
备注: 342 pages, textbook
Abstract:Amazing breakthroughs in reinforcement learning have taken place. Computers teach themselves to play Chess and Go and beat world champions. There is talk about expanding application areas towards general artificial intelligence (AI). The breakthroughs in Backgammon, Checkers, Chess, Atari, Go, Poker, and StarCraft have shown that we can build machines that exhibit intelligent game playing of the highest level. These successes have been widely publicized in the media, and inspire AI entrepreneurs and scientists alike. Reinforcement learning in games has become a mainstream AI research topic. It is a broad topic, and the successes build on a range of diverse techniques, from exact planning algorithms, to adaptive sampling, deep function approximation, and ingenious self-play methods. Perhaps because of the breadth of these technologies, or because of the recency of the breakthroughs, there are few books that explain these methods in depth. This book covers all methods in one comprehensive volume, explaining the latest research, bringing you as close as possible to working implementations, with many references to the original research papers. The programming examples in this book are in Python, the language in which most current reinforcement learning research is conducted. We help you to get started with machine learning frameworks such as Gym, TensorFlow, and Keras, and provide exercises to help understand how AI is learning to play.
[AI-166] Cross-Market Meta-Learning for Electricity Price Forecasting NEURIPS2026
链接: https://arxiv.org/abs/2610.17387
作者: Delaram Pirhayatifard,Terrence Alsup
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted at the NeurIPS 2026 Workshop on AI Foundations for Power Grids
Abstract:Day-ahead (DA) electricity price forecasting is usually posed per market, with a separate model fitted to each grid’s own history. This caps accuracy: statistical and tree-based baselines saturate, while higher-capacity sequence models need more training data than a single market’s few years of hourly prices can supply. Such methods also treat each node as an independent series, discarding the spatial dependencies induced by network topology and congestion. Because price formation across independent system operators (ISOs) is driven by shared physics and largely common market design, we instead treat cross-market history as a pretraining corpus. We present MetaGrid, a topology-aware transformer, meta-trained across markets so that a single zone-count-independent parameter set adapts to an unseen grid. On a leak-free benchmark of five U.S. ISOs, where every model receives the same prices, DA load forecasts and calendar features, MetaGrid attains the lowest average MAE, improving on the best baseline by 3.1% and on the same architecture trained from scratch by 7.7%.
[AI-167] Mitigating Black-Box Score-Based Query Attacks via Gradient Confusion
链接: https://arxiv.org/abs/2610.17382
作者: Ali Javeri,Javad Hezareh,Radin Shahdaei,AmirMahdi Sadeghzadeh
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注:
Abstract:Score-based black-box adversarial attacks pose a serious threat to deployed machine learning systems by exploiting repeated model queries to craft adversarial examples, requiring neither access to model parameters nor architecture. Existing defenses predominantly rely on stochastic output perturbations, which fail to prevent determined attackers from recovering stable gradient estimates by increasing query budgets or averaging over randomness. In this work, we propose the Gradient Confusion Defense (GCD), a retraining-free, deterministic test-time wrapper that actively shapes the probability vector returned to querying adversaries. By solving a constrained optimization problem grounded in the model’s Jacobian, GCD aligns the exposed logit-gradient with a target confusion direction that is maximally misaligned with the true adversarial loss gradient—increasing the query complexity of attacks without sacrificing predictive utility or top-1 accuracy. We provide rigorous theoretical analysis demonstrating that GCD provably slows gradient-estimation-based attacks and induces systematic accept/reject errors in random search-based attacks, with the confusion effect amplifying as the fraction of defended queries increases. Extensive experiments on CIFAR-10 and ImageNet across multiple architectures (ResNet-18, ResNet-50, ViT-B/16) and four state-of-the-art attacks confirm these predictions: GCD achieves state-of-the-art robustness, while incurring zero clean accuracy degradation. %Crucially, unlike randomization-based defenses, GCD’s deterministic structure provides resilience against adaptive attacks that exploit stochasticity, making it well-suited for security-critical deployment settings.
[AI-168] When Goals Leak Across Relationships: How Cached Goal Representations Shape Cross-Client Decisions NEURIPS2026
链接: https://arxiv.org/abs/2610.17365
作者: Saloni Gupta,Sébastien Krier,Glenn Matlin
类目: Artificial Intelligence (cs.AI)
备注: 10 pages, 4 figures. Accepted at NeurIPS 2026 Workshops on Interpreting Agent Behavior (IAB) and Agents in the Wild (AIWild)
Abstract:As LLM agents take on more autonomous roles, they may interact with multiple clients and counterparties whose goals and incentives conflict. If an agent maintains shared inference state across these interactions, information associated with one relationship may influence decisions made for another. We test OLMo-3-7B-Instruct in a controlled two-client environment with 20 market scenarios, where each client is assigned an opposing objective. With shared conversational memory, the model deviates from the current client’s objective in 22% of eligible decisions. We then investigate how these goals are represented in the KV cache. Linear probes show that client-goal ownership becomes highly recoverable in middle layers and is concentrated in particular attention heads. We use these results to identify where to intervene in the cache and test whether these representations affect the model’s decisions. Ablating the competitor’s cached goal representations at these locations shifts the model toward the current client. The effect is larger for heads identified by probing than for randomly selected heads, and larger when ablating goal tokens than matched non-goal tokens from the same client. Our results show that client-specific goal representations persist in the KV cache and can causally influence decisions made for other clients, highlighting the importance of relationship-specific state isolation in persistent LLM agents.
[AI-169] On the Statistical Mechanics of Attention
链接: https://arxiv.org/abs/2610.17347
作者: Jerry Yao-Chieh Hu,Yu-An Chen,Maojiang Su,Han Liu
类目: Machine Learning (cs.LG); Disordered Systems and Neural Networks (cond-mat.dis-nn); Statistical Mechanics (cond-mat.stat-mech); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注:
Abstract:We revisit the statistical mechanics of Transformer attention via modern Hopfield models. We present a free-energy formalism to characterize parameter regimes and critical thresholds for robust attention inference. The resulting free-energy landscape incorporates randomness in query states and attention weights. Attention then emerges as the equation of motion of the corresponding statistical-mechanical system (i.e., modern Hopfield model) through the standard free-energy minimization principle. This allows us to identify first-order phase transitions in attention between glassy and robust-inference phases. We further extend to attentions with stochastic queries and keys. Together, our results provide a rigorous statistical-mechanics framework for characterizing phase structure and inference reliability across a family of Transformer attention mechanisms.
[AI-170] Budgeted KV-Cache Repair for Evolving Contexts
链接: https://arxiv.org/abs/2610.17338
作者: Han Yu,Wenhui Zhu,Jincheng Cao,Rui Cai,Yuanda Xu,Zhengze Zhou,Xiwen Chen,Hao Wang,Xuanzhao Dong,Han Shi,Menglin Zhou,Shayan Mohajer Hamidi,Lin Su,Mina Razghandi,Sayan Ghosh,Qing Lan,Rohit Jain,Alborz Geramifard
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Language-model agents revise their working context as information becomes obsolete, but deleting text does not erase its influence on cached computation. Later key-value (KV) states remain conditioned on the deleted content, while recomputing the entire retained suffix can be costly. We introduce budgeted KV-cache repair: recovering task performance by recomputing a user-chosen fraction of the retained suffix. Comparing repair allocations under identical relevance scores and token budgets, we find that refreshing coherent regions can substantially outperform refreshing isolated tokens. This finding motivates Semantic-Unit Refresh (SUR), which allocates the budget to query-relevant text regions using inexpensive embedding scores, without training or an auxiliary transformer forward. To execute these repairs efficiently, a layer-packed executor processes selected states together at each layer while preserving full causal cache access. Across five models, SUR achieves strong recovery in controlled deletion tasks while recomputing only 2-5% of the suffix. On four natural multi-hop QA datasets, it achieves near-full-recompute exact match at a 5% budget with lower post-edit latency. The executor accelerates fragmented refresh by up to 16.1x over interval-wise execution.
[AI-171] Do Theorem Provers Need Theorem-Proving Agents ? mini-Lean-agent Agent s? mini-Lean-agent: A Coding-Agent Baseline for Lean NEURIPS2026
链接: https://arxiv.org/abs/2610.17331
作者: Tomas Ortega,Simon Park,Ayush Khaitan,Liam Fowl,Alex Kontorovich,Sanjeev Arora
类目: Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO)
备注: Accepted at the 6th Workshop on Mathematical Reasoning and AI (NeurIPS 2026)
Abstract:General-purpose coding agents use standard shell and file interactions, yet formal theorem-proving agents often rely on specialized, ad-hoc tooling. This makes it difficult to determine which components of theorem-proving agents provide benchmark gains: an improved base model, a specialized harness, a specific tool, or a combination of any of them. We introduce mini-Lean-agent, a deliberately minimal agent for Lean tasks whose only model-visible tool is bash. The model must use bash to inspect and edit the repository, search local libraries, invoke Lean, and manage its own intermediate work, while a separate trusted evaluator verifies the final artifact. We compare mini-Lean-agent against AxProverBase and Numina-Lean-Agent on the Putnam 2025 benchmark. Experiments with Gemini 3.8 Flash as the base model show that mini-Lean-agent solves 7/12 problems at \ 2 per problem limit, compared with AxProverBase’s 4/12. With Astra as the base model, mini-Lean-agent solves 12/12 problems at an average cost of \ 1.28 per problem. This suggests that a generic coding agent with a minimal harness can already achieve strong performance on Lean benchmarks. Our results motivate using mini-Lean-agent as a standard baseline when evaluating theorem-proving infrastructure.
[AI-172] Joint Embedding with Dynamic Integration network (JEDI-net)
链接: https://arxiv.org/abs/2610.17314
作者: Rishav Mukherjee,Jeffrey Ahearn Thompson
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Deep learning has become the dominant form of machine learning due in part, to its ability to automatically engineer complex meta features from data. However, even deep learning features can perform erratically because of unobserved latent factors. incorporating latent factor information explicitly, can improve performance by letting the model alter its behavior based on such inferred factors. Because neural networks often capture such complexity inherently, external clustering of the original feature space offers limited benefit. clustering observations by feature importance profiles, rather than raw features, offers a principled way to uncover latent structure, since observations with similar importance patterns may share predictive dynamics. In a prior work, Feature Oriented Representation with Clustering and Explanation (FORCE) we showed that this approach measurably improved neural network performance. In this article, we present JEDI-net(Joint Embedding with Dynamic Integration Network) a unified architecture that periodically clusters SHAP values during optimization to identify latent subgroups with similar feature importance patterns, embed their identities directly into the network, allowing predictions to adapt to each observations predictive structure.
[AI-173] Beyond Successful Play: Diagnosing Opponent Modeling in Large Language Models
链接: https://arxiv.org/abs/2610.17310
作者: Hanbo Xie,Yue Zhao,Robert C. Wilson
类目: Artificial Intelligence (cs.AI); Neurons and Cognition (q-bio.NC)
备注:
Abstract:Language-model agents are expected to adapt to others whose behavior changes through interaction. Yet adaptation may rely on learned behavioral patterns rather than theory of mind (ToM): the ability to infer others’ mental states, including beliefs and intentions. We introduce a benchmark and diagnostic framework for repeated hide-and-seek against algorithmic opponents with explicitly defined recursive belief updates. Their ground-truth beliefs and action probabilities let us distinguish adaptive play, action prediction, and belief tracking through live interaction, fixed-history evaluation, and controlled information interventions. With structured elicitation and behavioral supervised fine-tuning (SFT), we find improved play in several tested settings, but diagnostics reveal limitations in opponent prediction and belief tracking. We test whether supervising opponent action probabilities can address this predictive deficit in Qwen2.5-7B-Instruct. Predictions improve when the model is told the opponent’s type and parameter settings, but become less accurate when it must infer the opponent from history without this information. Training on examples that also omit these details reduces the gap, although improvements vary across opponents. Is this limitation specific to smaller open-weight models, or does it also appear in a frontier model? GPT-5.4 also benefits from correct opponent beliefs: supplying them improves its predictions and choices, whereas deeper explicit belief reports and self-generated memory do not consistently improve forecasting from interaction history. This contrast suggests that using useful belief information is easier than inferring it autonomously, even for a frontier model. Together, our findings distinguish improved play and action prediction from reliable belief tracking, providing a framework for evaluating mentalizing and testing targeted improvements.
[AI-174] Reliable Text-to-SQL in Resource-Constrained Settings with Small LLM s via Iterative Retry Workflows
链接: https://arxiv.org/abs/2610.17305
作者: Bilel Abderrahmane Benziane,Xavier Schepler
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large Language Models (LLMs) have recently improved Text-to-SQL generation, but their computational cost can hinder deployment in resource-constrained settings. This paper evaluates the Text-to-SQL capabilities of small and medium open-source LLMs (1B-8B parameters) under a controlled inference budget, focusing on how model specialization and prompt formulation affect performance. Using the Spider benchmark, we compare multiple prompt representations and assess the accuracy of generated queries. Our results show that SQL-specialized models are consistently more reliable than general-purpose and chat-oriented models at small scales, and that performance can vary substantially depending on the prompt used. To reduce this variability without resorting to larger models, we introduce three lightweight inference workflows that automatically retry with alternative prompts and/or models when failures occur, including an optional syntax-correction step. These workflows significantly improve end-to-end success rates, with a trade-off in additional inference calls. These results demonstrate a practical path to stronger Text-to-SQL performance with small LLMs, by trading modest additional inference cost for higher reliability.
[AI-175] he Agent Knew: Internal States Forecast Unauthorized Actions and Stop Them and the Lies That Follow
链接: https://arxiv.org/abs/2610.17301
作者: Arya Fayyazi,Mehdi Kamal,Massoud Pedram
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Language-model agents sometimes act beyond their authorization and then misreport what they did. On CAPE-Bench, matched agent tasks in which one fact decides whether an action is authorized, frontier open-weight agents do both, and one behavior marks the lie: retroactive permission laundering, acting and then asking permission as if the action were pending, precedes a false report in 70% of 1081 fresh informed violations, against 8% otherwise (preregistered odds ratio 7.0); frontier agents on tau^2-bench instead fabricate the authorization before violating. A linear readout of the agent’s residual stream forecasts the violation before the action (AUROC 0.94) and the false report before the report, and beats chain-of-thought monitors when the agent is pressured to hide its reasoning. Gating every proposed action on this forecast cuts violations over whole episodes in four model families, and sending repeated flags to an approver keeps authorized work (Gemma-4: violations 0.70 - 0.09, completion 1.00; a model approver that escalates unclear cases keeps this gain with a person seeing 25% of asks), at about 10-8 of the agent’s compute and 109 times less than a chain-of-thought monitor. On AgentDojo, an external prompt-injection benchmark, the same readout calibrated on 29 tasks, with the steering unchanged, cuts attack success on unseen tasks in two families (Gemma-4 11.7% - 5.7%, gpt-oss 28.5% - 17.5%) while keeping benign utility (Gemma-4 0.637 vs 0.637; gpt-oss 0.667 vs 0.681). Preventing the violation is also the most reliable way to prevent the lie (misreported violations 0.10 - 0.01 pooled over three families, preregistered), whereas lie-specific fixes after a violation are weak. All tests are preregistered and hash-locked; code, data and the full record, including tests that did not pass, are at this https URL.
[AI-176] owards Resilience: Slice-Driven Orchestration of Mobile IoMT in 6G-Edge-Continuum
链接: https://arxiv.org/abs/2610.17296
作者: Muhammad Zeeshan,Hafiz Faheem Shahid,Erkki Harjula
类目: Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC); Networking and Internet Architecture (cs.NI)
备注: ISMICT 2026- 20th IEEE International Symposium on Medical Information and Communication Technology, Osaka Japan
Abstract:Emerging healthcare services, such as real-time diagnostics, remote health monitoring and computer-assisted surgery, require stringent Quality of Service (QoS) guarantees while operating under limited and heterogeneous edge resources. Existing Internet of Medical Things (IoMT) orchestration methods primarily focus on communication metrics such as latency and bandwidth, while overlooking edge resource constraints and dynamic adaptation to workload and user mobility. This paper presents a novel resource-slicing-based adaptive orchestration framework that integrates 5G network slicing (NS), resource-aware orchestration, and mobility-aware task offloading into a single adaptive system. IoMT services are assigned to distinct 5G network slices while jointly considering CPU capacity, memory availability and amount of storage, enabling more efficient service placement and execution. In addition, the system supports user mobility and dynamic task offloading to maintain service continuity under changing network conditions. SimPy-based simulation demonstrates that the proposed method enhances service deployment efficiency, reduces request blocking, and improves user acceptance while maintaining reliable performance across diverse IoMT scenarios.
[AI-177] PhysCodeGym: Benchmarking Coding Agents for Physically Verifiable 4D Reconstruction
链接: https://arxiv.org/abs/2610.17288
作者: Shuai Shao,Jinchen He,Cunxin Fan,Zihan Guo,Shijian Wang,Qihan Ren,Kangning Zhang,Zunnan Xu,Haoyuan Li,Hao Wang,Jianghao Lin,Weiwen Liu,Weinan Zhang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Building physical simulations from video requires coding agents to translate observations into executable models and refine them through tool use. We introduce PhysCodeGym, a benchmark for physically verifiable 4D reconstruction with 156 tasks covering rigid-body, liquid, deformable, and wind-driven motion. Agents autonomously inspect videos, write and run Blender programs, and revise their scenes using simulation feedback. Six metrics combine rule-based measurements with rubric-based judgments to assess visual and physical reconstruction quality. Program inspection, scene measurements, and counterfactual interventions test physical consistency under original and modified conditions. Experiments with frontier coding agents reveal a tradeoff between artifact quality and completion across reasoning-effort settings, while harness comparisons and agent traces link verification behavior with reconstruction performance. Case studies show that matching visible objects and events can coexist with physical defects. Counterfactual tests further expose latent contact and support failures in reconstructions that pass ordinary physical checks, highlighting the gap between reproducing observations and building reliable physical models. Code is available at this https URL.
[AI-178] Phase-Conditioned Affordance Fields for SE(3)-Equivariant Visuomotor Policies
链接: https://arxiv.org/abs/2610.17281
作者: Xinzhe Chen,Liqi Huang,Sihua Ren,Haowen Sun,Mingyang Li,Zeyang Liu,Xingyu Chen,Xuguang Lan
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: Accepted to the 10th Conference on Robot Learning (CoRL 2026). Project page: this https URL
Abstract:SE(3)-equivariance provides a useful geometric inductive bias for visuomotor policies, encoding that actions covary with rigid transformations of the observation. However, existing SE(3)-equivariant visuomotor policies build this bias on a global equivariance assumption that treats the entire observation as governed by a single rigid transformation, which breaks down in real-world manipulation tasks where the task-relevant region changes across different objects. We address this by learning a phase-conditioned affordance field (PAF) over the observed point cloud, a per-point score that the SE(3)-equivariant policy uses in two complementary ways. Its score-weighted centroid serves as a learned, phase-conditioned anchor that canonicalizes the observation and the action around the active region, replacing the fixed gripper-frame reference, and the score itself is appended as a per-point feature. This introduces a local geometric inductive bias, while preserving the policy’s global SE(3) equivariance. We validate on multiple MimicGen tasks under varying spatial randomization, and PAF consistently improves the success rate of the underlying policy. On multi-object real-robot tasks, PAF outperforms baselines and keeps its lead under lighting variations.
[AI-179] Reasoning for Propaganda Detection via Label-Guided CoT Optimisation
链接: https://arxiv.org/abs/2610.17271
作者: Zeba Afroz,Harsh Vardhan,Md. Shad Akhtar
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Propaganda detection systems usually operate as black boxes, lacking explanations. To address this, we hypothesise that persuasive techniques contain definitional characteristics with respect to their social-political setting. We introduce ReCoT, a reasoning-conditioned detection framework that captures definitional cues of persuasive techniques via population-level refinement and performance-guided resampling, without requiring additional explanation annotations. On INDI-PROPspan, ReCoT achieves a macro F1_LCS of 0.364, surpassing baselines and reaching 88% of the SFT upper bound on Avg F1_LCS. Similarly, on PTC, ReCoT reaches 85% of the SFT upper bound on Avg F1_LCS, closing much of this gap while remaining interpretable. Importantly, the analysis of the reasoning pool’s convergence and update pattern for ReCoT shows effective refinement of reasoning CoTs.
[AI-180] Measuring Learned Equivariance: An Information Theoretic Approach
链接: https://arxiv.org/abs/2610.17269
作者: Claire Schlesinger,Wolfgang Gatterbauer,Robin Walters
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Information Theory (cs.IT)
备注:
Abstract:Learning symmetries through data augmentation offers an alternative to architectural equivariance, but does not guarantee that a network respects those symmetries. Measuring learned equivariance is especially difficult for intermediate representations, whose group actions are not specified. In a finite discrete setting, we show that equivariance requires representations of original and transformed inputs to determine one another exactly, yielding a test based on normalized mutual information (NMI). For high-dimensional activations, we use centered kernel alignment (CKA) as a scalable test, with an exact guarantee for orthogonal feature transformations. Across architectures and tasks, augmentation usually increases transformation consistency, showing that learned equivariance can be probed directly from representations without recovering the hidden transformation rules that relate them.
[AI-181] RL Post-Training Diffusion Language Models Along Their Own Generation Paths
链接: https://arxiv.org/abs/2610.17266
作者: Vaibhav Singh,Eugene Belilovsky
类目: Artificial Intelligence (cs.AI)
备注: 9 pages, 6 figures
Abstract:Diffusion language models are a promising and more efficient alternative to autoregressive models. However, most RL post-training techniques are not easily adapted to these models. One issue is that diffusion language models generate answers through many intermediate states, deciding at each step which masked positions to reveal and which tokens to place in them. Many existing RL methods used in this setting approximate intractable sequence likelihoods or weight every denoising step by the same advantage, without explicitly distinguishing how these intermediate states of the denoising trajectory affect the expected reward. To address this issue of trajectory blindness, we introduce PACT (Potential-Anchored Consistency of Transitions). Motivated by a reward-tilted reference distribution, PACT jointly learns the policy and a scalar potential that scores each intermediate state, anchored by terminal reward. For each sampled denoising transition, a consistency loss matches the policy-to-reference log-probability ratio of revealed tokens to the change in this potential. This formulation requires neither sequence-likelihood estimation nor importance-ratio clipping. On four tasks, PACT improves LLaDA-8B-Instruct by +9.9 points on GSM8K, +12.0 on MATH500, +55.7 on Countdown, and +72.6 on Sudoku, and reaches the best average accuracy across the four tasks reported for this model, +3.7 points over the next best method. Further compared to AGRPO, it reaches that baseline’s best accuracy in about half the training compute on GSM8K.
[AI-182] Does a Governed Confidence Score Know When a Text-to-SQL Answer Is Right? An Exploratory Study on BIRD and a Pre-registered Confirmation on Spider
链接: https://arxiv.org/abs/2610.17256
作者: Raghu Venkat,Tricha Anjali
类目: Artificial Intelligence (cs.AI)
备注: 23 pages, 4 figures, 9 tables. Code, frozen scorer, pre-registration and data: this https URL
Abstract:Conversational analytics lets people ask questions of records spread across many systems and get answers in plain language; it is useful only if each answer can be trusted and explained. A production system of this kind attaches a confidence score to every answer it writes over a database and uses it to show the answer, show it with a warning, or withhold it. The score’s weights, caps and thresholds were set by design. This paper audits that design. On the 1,534 questions of BIRD dev, answered by GPT-5.1 with a benchmark-adapted production prompt and five samples per question, the designed score separates right from wrong answers weakly (AUROC 0.624, 95% CI 0.595-0.652) and withholds 35.7% of correct answers. Nearly all of its ranking signal comes from agreement among the samples, and most of it is lost where agreement is combined by a minimum with a checklist that drops to zero on any unsupported claim, including claims in commentary. We froze a rule in which agreement with the shown answer is the score and deterministic checks act as vetoes and caps, and pre-registered a test on Spider dev before any Spider answer existed. The primary hypothesis held: AUROC rose from 0.566 to 0.725 (paired difference +0.159, 95% CI +0.097 to +0.221), and correct answers withheld fell from 49.0% to 8.8% with PASS accuracy 95.7%. Under one matching policy the choice of agreement statistic matters little; the gain comes from how agreement enters the score. Whether added checks help depends on how correctness is defined, while they always cost unqualified coverage. A pre-registered empty-result check did not meet its effect-size criterion on Spider. Correctness here means the core answer; labels are model-judged, and a blinded two-person audit, not yet complete, will be reported in a revised version.
[AI-183] Compute-optimal data scaling for neural surrogates via multi-fidelity training NEURIPS2026
链接: https://arxiv.org/abs/2610.17221
作者: Paul Setinek,Fabian Paischer,Nils Thuerey,Johannes Brandstetter,Felix Koehler
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted at NeurIPS 2026. Project page: this https URL Code: this https URL
Abstract:Neural surrogates for Partial Differential Equations (PDEs) are usually trained on synthetic data from numerical simulators. As models grow and scaling laws demand larger datasets, the cost of data generation often outweighs the training cost and becomes the bottleneck in the scaling quest. Unlike data in other domains, however, each PDE sample has a tunable cost that depends on its numerical fidelity, with the resulting cost-accuracy trade-off governed by known convergence rates. We exploit this property by formalizing neural surrogate training as a function of both data quantity and fidelity under a fixed data generation compute budget, treating fidelity as a fine-grained discrete spectrum rather than a binary distinction. Building on this, we show that multi-fidelity training enables Pareto-optimal error scaling against data generation compute. By using simulations at lower fidelities as an additional training signal, training on the optimal fidelity mixture matches the accuracy of high-fidelity-only training, with up to 6.3\times less data generation compute across our settings. We validate these findings across six PDE problems spanning structured grids and unstructured meshes, two distinct error sources (discretization and iterative-solver error), and four surrogate architectures, totaling over 7,000 training runs.
[AI-184] AutoMARS: Automatic and Scalable Multi-Agent Large Language Model Framework for Supply Chain Risk Assessment
链接: https://arxiv.org/abs/2610.17206
作者: Yinzhu Quan,Zefang Liu,Frederick Benaben,Benoit Montreuil
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:We propose AutoMARS (Automatic Multi-Agent Risk Assessment), an agentic framework for automatic and scalable comparative risk assessment in supply chain networks. The framework enables end-to-end execution of analytical workflows by automatically planning tasks, coordinating specialized agents, invoking external analytical tools, and synthesizing domain-specific results into integrated risk evaluations. AutoMARS further supports scalable risk comparison across multiple locations by parallelizing independent analytical tasks. The framework adopts a multi-domain, multi-agent, and multi-tool architecture that integrates heterogeneous risk factors from socioeconomic, disaster, environmental, and transportation domains. Through coordinated agent interactions and tool integration, AutoMARS enables structured, systematic, and interpretable comparative analysis to support supply chain risk-aware location evaluation.
[AI-185] Releasing the Inner Jev: A Simple Recipe for Open Decision Models
链接: https://arxiv.org/abs/2610.17205
作者: Jonathan Geuter,Andrew Cheng,David Alvarez-Melis
类目: Artificial Intelligence (cs.AI)
备注: 23 pages, 5 figures
Abstract:LLM-based decision-making systems, also called System One models, have recently gained popularity, most notably through Jev (Almeida, 2026). Given a typed input state, a question, and a set of options, they return a probability for each option in a single forward pass, without text generation. We show that existing open instruction-tuned language models can achieve comparable performance \textitwithout any additional training, by reading out answers from letters (A, B, C, etc.) in the model’s output logits at the last token position. Furthermore, we show how a simple supervised fine-tuning recipe can improve their performance further. We evaluate four open instruction-tuned language models: Gemma 4 31B, Gemma 4 12B, Qwen3.6-27B, and Qwen3.5-4B. We fine-tune each using LoRA and a decision head initialized from a slice of the language model head. Furthermore, we introduce OpenDecisions, a novel 2,352-question validation dataset designed to reflect the task composition of JevBench (JevBench main- tainers, 2026) and which computes a Capability score based on accuracy and calibration metrics. Without fine-tuning, Gemma 4 31B scores 79.6 on OpenDecisions, compared to Jev’s 80.1. After undergoing our fine-tuning recipe, it scores 84.4 on OpenDecisions and 81.5 on JevBench v1.7, outperforming Jev by 4.3 and 4.4 Capability points respectively and setting a new state-of-the-art on OpenDecisions. On a suite of public decision datasets, our models achieve accuracy comparable to Jev, with lower calibration error. Our results show that open language models already possess System One capabilities, and with simple fine-tuning they outperform Jev on a wide range of decision tasks. We release our models, serving code, and OpenDecisions.
[AI-186] Deleting Code Without Losing Knowledge: Regenerative Software for Python Libraries
链接: https://arxiv.org/abs/2610.17197
作者: Thomas Bartz-Beielstein
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: 53 pages
Abstract:In the age of “Software 3.0”, when a complete program can be rewritten in minutes by an agent, the traditional approach of relying on code as the sole source of truth becomes questionable. Fowler advocated for a set of seven primitives, which lay the framework for regenerative software, so that programs can be deleted and rebuilt without losing their knowledge. The knowledge lives in machine-readable records in a tracked directory, from which a regeneration bundle is built for each unit. The regeneration bundle is a single file with the module’s architecture, intent, decisions, public contract, evaluations, and consumers, but no implementation, plus the one dependency module it may import. The open-source package spotregen implements these primitives and provides a graphical user interface. This report presents a first application of Fowler’s primitives to the safety-critical forecasting library spotforecast2-safe. During the first experiments, the Claude Code session that orchestrated the study consumed more than 43 million tokens, almost all of them its own history re-sent with every request. This motivated the switch to an architecture that is built on local models. Deterministic checks and the maintainer judged the results. The report contains executable examples for every primitive. Each example runs spotregen against the repository or against a throw-away copy of it.
[AI-187] Size Generalization for Diffusion Routing Solvers
链接: https://arxiv.org/abs/2610.17192
作者: Barak Itzchak Davidovitch,Eli Meirom,Ethan Fetaya,Gal Chechik
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Routing problems, like the Traveling Salesman Problem (TSP) and the Capacitated Vehicle Routing Problem (CVRP), are NP-hard and slow to solve for large instances. Diffusion-based solvers can be trained in a supervised way using sets of optimal solutions, and then quickly solve new test instances. However, for large instances, optimal solutions are hard to find, and diffusion models fail to generalize to instance sizes outside their training distribution. Here we describe ResizeCO, a label free size-generalization approach that addresses both the change in statistical properties of larger instances, and the computational challenge of solving them quickly. It uses reinforcement learning on large-scale instances to adapt a diffusion model trained on small instances. Specifically, we adapt GRPO to this problem, using route length as the training signal. ResizeCO also formulates k-opt as a tree search, solved by a beam search on the GPU. In size-generalization experiments on TSP and CVRP instances with up to 10000 nodes, ResizeCO achieves a Pareto front over route quality and time that dominates learned solvers, and also outperforms other solvers at short run times, setting a new state of the art.
[AI-188] FreeSolo: Linear-time Sparse Attention with Zero-Shot Length Extrapolation
链接: https://arxiv.org/abs/2610.17191
作者: Zekun Shi,Alastair James Arthur Price
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Linear-time sequence models typically store long-context information in a recurrent state. A fixed-size state limits how much information can be stored, while hybrid architectures that add full attention for better recall incur quadratic sequence cost. Alternatively, sparse attention with continued pretraining (CPT) can extend the effective context length. Generalization beyond the trained length remains challenging, however, and CPT can be costly. We introduce FreeSolo, a native sparse attention mechanism with a short-range sliding window branch and a long-range branch that selects keypoint cells for direct token attention. Its linear-time variant, FS-L, learns to prune the prefix before query-dependent selection through short CPT. FS-B searches balanced trees with O(N\log^2 N) algorithmic cost including radix-sort construction. FreeSolo demonstrates better zero-shot length extrapolation than the baselines without long-context fine-tuning.
[AI-189] Generative Flow on Symbolic Abstractions for Embodiment Capability Discovery WWW
链接: https://arxiv.org/abs/2610.17174
作者: Youwei Yu,Lantao Liu
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Symbolic Computation (cs.SC)
备注: The code and demonstration videos are available at this https URL
Abstract:Achieving complex, long-horizon robot behavior remains challenging because successful execution depends on skills grounded in a robot’s physical affordances. Existing training-task distributions are often handcrafted or sampled from large language models (LLMs), relying on semantic plausibility rather than embodiment-specific feasibility. This mismatch can produce curricula that are too difficult, redundant, or blind to non-obvious but learnable behaviors. We propose Generative Flow on Symbolic Abstractions (FOSA), a framework for discovering embodiment-specific capabilities by learning a grounded task generator. FOSA parameterizes an abstract policy as a Generative Flow Network that constructs symbolic task skeletons. Each skeleton is compiled into a simulation environment, where low-level policy-learning performance provides the reward signal for the abstract policy. Across heterogeneous robot platforms, FOSA discovers diverse and executable tasks near the robot’s empirical capability frontier. After distilling the discovered task policies into a single language-conditioned policy, FOSA improves performance on unseen complex and long-horizon tasks. The code and demonstration videos are available at this https URL.
[AI-190] Force-Modulated Dexterity: Unified Compliance for Contact-Rich Manipulation
链接: https://arxiv.org/abs/2610.17161
作者: Ying Yuan,David Held,Jitendra Malik,Haozhi Qi
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: Project page: this https URL
Abstract:Teleoperation is widely used to collect high-quality demonstrations for robot learning. Dexterous manipulation, however, requires coordinated control of motion and contact forces. Operators can specify motion, but struggle to reproduce the subtle force adjustments of human manipulation without feeling the robot’s contacts. We introduce Force-Modulated Dexterity (FOMO-Dex), a unified compliance formulation for contact-rich manipulation. FOMO-Dex expresses nominal actions in coordinates defined by task geometry and contact structure. It applies force-driven motion corrections along selected directions while leaving the remaining command components unchanged. Across card picking, paper cutting, screwdriving, and nut-and-bolt assembly, FOMO-Dex reduces excessive contact forces compared with stiff control. A user study shows that operators collect more successful demonstration data per hour after brief practice. For policy learning, imitating the corrected actions improves card-picking success over imitating raw operator commands. Policies trained on FOMO-Dex demonstrations also achieve more rotation in nut-and-bolt manipulation and screwdriving than those trained on demonstrations collected with the stiff controller. Videos are shown on this https URL .
[AI-191] SonicWorld: Source-Centric Latent Dynamics for Spatial Audio Reasoning
链接: https://arxiv.org/abs/2610.17131
作者: Chenlong Wei,Hanling Wang,Mingjin Lou,Yufan Shi,Yandong Sun,Sheldon Yu,Xiaohui Zhu,Xiaobo Jin,Yiwei Wang
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Audio and Speech Processing (eess.AS)
备注: 37 pages, 12 figures, including appendices
Abstract:Dynamic spatial audio reasoning requires a model to associate changing acoustic observations with the correct source and recover its spatial state at a queried time or interval, a challenge intensified by overlapping speech and source motion. To study this problem, we introduce Sonic-Bench, comprising 2.8K first-order Ambisonics clips and 28.8K target-specific question-answer pairs across 480 simulated indoor scenes. The benchmark covers localization, distance estimation, and motion reasoning in single- and dual-source settings. We further propose SonicWorld, which integrates physical, learned acoustic, and semantic features into persistent source states through learned transitions and recurrent updates, and projects them into an Omni LLM for spatial reasoning. Compared with Spatial-Omni on the Sonic-Bench test split, SonicWorld improves direction accuracy from 31.98% to 34.69% and reduces azimuth error from 57.22° to 48.01°. In zero-shot transfer to Spatial LibriSpeech, SonicWorld retains strong localization performance, achieving 73.98% direction accuracy and 19.29° azimuth error, compared with 27.14% and 70.74° for Spatial-Omni.
[AI-192] EvoQuantLab: Training-Free Dual-Loop Evolution for Auditable Quantitative Research
链接: https://arxiv.org/abs/2610.17130
作者: Haolin Yang,Zhiqiang Qian,Chutian Chen,Jipeng Zhang,Tangjia Deng,Jian Xie,Shuaishuai Gong,Sirui Han,Yike Guo
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Automated alpha factor discovery is critical for modern quantitative research. While large language models (LLMs) have increasingly been integrated into quantitative workflows, existing agent-based approaches remain highly susceptible to severe backtest overfitting. To address this challenge, we propose EvoQuantLab, a training-free, dual-loop self-evolving framework for robust factor discovery. Instead of relying entirely on LLM generation, EvoQuantLab deploys an inner genetic-programming (GP) loop that dramatically improves token efficiency while evolving typed factor expressions under structural and diversity constraints. An outer diagnosis loop then systematically audits evidence across cycles to drive protocol-level improvements. Furthermore, our evaluation pipeline applies multi-metric validation to rigorously filter out spurious factors before deployment. Extensive out-of-sample experiments across both daily and hourly frequencies – covering equities and cryptocurrency – demonstrate the robustness of EvoQuantLab. Specifically, on daily CSI 500 and SP 500 tasks, EvoQuantLab achieves leading annualized returns of 18.34% and 21.30%, consistently outperforming existing agent baselines. On high-turnover hourly cryptocurrency data, EvoQuantLab delivers a 37.91% annualized return while effectively avoiding execution collapse.
[AI-193] Who is the I in AI? Establishing Decentralized Identity and Accountability for Autonomous Agents
链接: https://arxiv.org/abs/2610.17126
作者: Ahammod Munim Ramil,Khalid Al Kafi,M. A. Abdullahis Sany,Muktadirul Islam Nibir,Md Yeasin Ali,Mohammad Zavid Parvez,Md Sadek Ferdous
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:The rapid proliferation of autonomous AI agents in decentralized ecosystems challenges existing identity paradigms. While human identity is traditionally anchored in biological and/or self-asserted attributes and government credentials, AI agents lack a standardized definition of selfhood, creating critical gaps in trust and accountability. This paper introduces a novel framework for establishing Decentralized AI Identity rooted in Self-sovereign identity (SSI) principles. We propose a formal model that clearly distinguishes human users and AI agents, enabling distinct verification logic for each class while maintaining a unified identity calculus. The central to our contribution is the ABCD Framework, a multi-layered approach to agent identification and verification: (A) Algorithmic Fingerprint, which binds the identity to static model architectures and provenance; (B) Behavioural Fingerprint, which verifies dynamic performance metrics over time; © Cryptographic Identity, which anchors the agent via Decentralized Identifiers (DIDs) and (D) Delegation, which cryptographically binds agent actions to respective human accountability. By orthogonalizing these dimensions, we model how systems can enforce granular trust policies ranging from simple authentication to complex liability assignment, thereby solving the black box identity problem in the autonomous economic activities of AI agents.
[AI-194] D-World: An Explicit World-Language-Action Model with Recursive Self-Improvement under Dual Control
链接: https://arxiv.org/abs/2610.17114
作者: Xuan Wen,Baichao Wang
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 23 pages, 4 figures, 2 tables
Abstract:Frontier large language models (LLMs) now reason about space well enough to program robots, yet the world models behind robot policies are implicit, held in network weights or predicted video, and cannot be inspected or corrected directly. We present D-World, an explicit world-language-action model built on the Sense, Plan and Act decomposition. A perception agent writes the scene as an executable digital twin using generated vision and geometry code. A planning agent searches for actions in simulation and, after each executed action chunk, revises the twin’s object states and named physical parameters against what the robot measured. A parameter revision is accepted only if replaying the latest chunk does not increase any measured residual beyond its tolerance, and every chunk is checked in the updated twin before it runs. Structural mismatches return to perception for reconstruction. We call this measurement-grounded cycle of correction and revision recursive self-improvement (RSI), in which the LLM stays frozen while the world it acts in improves. We model the choice of actions that both advance the task and expose model error as a dual control problem and show that the cost of incorporating evidence into the twin sets the threshold at which probing pays, so cheap local revision counteracts the turn-off of cautious controllers. Each episode runs a real-to-sim-to-real loop, and its recorded twins can be refitted and re-rendered offline into higher-fidelity environments for post-training. Because perception and planning communicate only through the twin and language, either can be replaced by a learned model, and since inference dominates the loop, the loop speeds up with faster models. On a dual-arm robot, towel and toaster tasks show local revisions, such as one reducing the bread’s rendered-depth residual from 22.4 to 1.5 mm, that let execution continue without reconstruction.
[AI-195] AIPOCH Open-Science: An Open-Source AI Research Workbench with Reproducible .science Packages
链接: https://arxiv.org/abs/2610.17110
作者: Wen Ouyang,Bocheng Huang,Qianyu Yao,Fei Sun,Minying Deng,Haoyue Cheng,Wenwen Zhou,Lei Mou,Yusi Wei,Jiang Li,Kun Zhu,Lei Wang,Yi Feng,Jiaheng Dai,Ke Qian,Dan Wang,Yahui Wang,Chenyi Wang,Mingsheng Lyu,Yaqiang Zhang,Ping Zheng,Chunhua Tang,Xinge Ma,Xu Zhang,Youjia Fan,Can Lu,Zhijun Dai,Man Hu,Qiqi Xie,Beinan Liu,Yuanning Guo,Yunze Liu,Jinhong Jiang,Xianjun Fu,Mingwei Zhang,Yuanji Xu,Shilin Xia,Bowen Yang,Mingjun Yu,Jingdun Xie,Xuwei Chen,Xia Zou,Yuzhou Xue,Jianguo Zhou,Chunhao Cao,Donghao Tian,Lei Huang,Jinxiao Li,Shengrong Long,Zefeng Song,Honglin Li,Jiang Luo,Wenying Jiang,Shanshan Li,Lei Dai,Ning Jiang,Ying Cai,Chaohui Li,Chunyan Peng,Hetong Zhao,Angqing Li,Chunyan Yang,Chongyang Sun,Yong Li,Xinyu Shi,Wei Dai,Ming Peng,Chen Sang,Yuanyuan Han,Desong Qiu,Haixin Miao,Jian Xu,Yi Chen,Xiaoliang Shi,Tian Shen,Teng Hu,Shuhao Wang,Yingyong Hou,Huimei Wang
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: 23 pages; 2 main figures, 3 tables, a graphical abstract and Supplementary Figure S1
Abstract:AI-assisted research leaves instructions, computations, artifact revisions and checks whose relationships matter when another researcher receives the work. We present AIPOCH Open-Science, an open-source research workbench with portable .science session packages, and specify an exchange contract for these recorded relationships. Version and assessment targets remain attributable while receiving-workspace identifiers can be remapped, eligible contents can be omitted explicitly, and imported history remains separate from a writable fork. A controlled session tested an incorrect first version, a reviewed correction, an unreviewed revision and a planned but unexecuted step. We compared complete and partial exports with source records extracted independently of the exporter, then inspected imports into separate projects within the same installation. Both conditions retained the tested version and review associations, kept the unreviewed revision distinct from earlier assessments and represented excluded content as unavailable. A documentation-rich ordinary directory also supplied evidence for all seven static inspection questions. A separate registry case connected a table to its recorded task, execution and review; offline execution of its retained script and fixed input reproduced two saved CSV tables byte for byte. The study contributes an implemented exchange contract and a controlled examination of its preservation rules, content-selection constraints and interpretation boundaries.
[AI-196] SFT Leaves a Reproducible Signature in Activation Geometry: Calibrating Post-Training Interpretability Against Run-to-Run Noise
链接: https://arxiv.org/abs/2610.17093
作者: Xupeng Chen,Chenqian Le,Jiayi Cheng,Runhao Li,Michael Yue
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:A model diff credits post-training with whatever separates a base checkpoint from one post-trained descendant, presuming a rerun would land close by. On the sparse co-activation graph of Qwen3-1.7B it does not: two LoRA-SFT reruns sit over half as far apart as either sits from the base. We score each base-to-post contrast against the run-to-run band of its own recipe, and show that k runs certify no better than p=1/(k+1) under run exchangeability, however large the effect, so the band z -score is a descriptive scale. SFT leaves a large, reproducible signature: base \to SFT displaces the graph by 17.8 pairwise-distance SDs beyond the mean distance between four independent LoRA-SFT runs, and the pre-registered estimator replicates on Llama-3.1-8B with Llama-Scope. At 29 steps each, four SFT and four DPO runs occupy disjoint displacement ranges, a gap 5.0\times and 5.6\times the spreads of the two arms, without any band. All bands use graphs built on one GPU architecture. Three other readings stop short of a verdict: DPO band counts (3/4, 1/4, 3/4 contrasts outside at 12, 29, 60 steps) show no consistent dose ordering while its displacement stays flat; light GRPO is a secondary non-detection, not an absence, and we run no equivalence test; and the graph concentration read as an RL signature is already in the zero-RL base. A pre-registered causal sweep at three edge budgets on 4,816 held-out prompts found the displaced sources load-bearing ( 12\times a matched random control in mean, 3.5\times in median at 10,000 edges), but their stage-specificity interaction reversed at two budgets and the third failed its faithfulness gate, so we claim calibrated structural displacement, not a changed mechanism. Scope: one backbone, one SFT recipe and general-text probe states for the full protocol.
[AI-197] Interactive Navigation Beyond Reachability: Benchmark and Agent ic Framework
链接: https://arxiv.org/abs/2610.17089
作者: Delun Li,Xueyao Wei,Rui Chen,Yifei Wang,Junzhi Yu,Wenjie Song,Xingyu Chen
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Existing embodied navigation tasks commonly assume that traversability and object states remain fixed during an episode. Yet reaching a goal may require opening a door to access a room and then opening a container to reveal a hidden target. We formulate interactive navigation beyond reachability, unifying channel-induced changes in spatial connectivity, container-induced changes in target visibility, and their sequential dependencies. We introduce MultINav-Bench, a 3K-episode ObjectGoal benchmark in MolmoSpaces spanning channel, container, and mixed tasks, with interaction-necessity annotations covering required and unnecessary episodes. Its data generation pipeline automates task construction across multiple task types. The evaluation protocol separately measures navigation success, interaction effectiveness, and cost. We further propose MultINav-Agent, a three-stage agentic framework built on a persistent, state-conditioned interaction graph. Semantic interactive perception grounds entities and interaction states; a vision-language model selects task-relevant actions based on expected changes in connectivity and target observability; and post-interaction verification updates the graph to guide subsequent navigation, exploration, or interaction. Experiments on MultINav-Bench show that MultINav-Agent consistently outperforms adapted baselines across channel, container, and mixed tasks.
[AI-198] Superintelligence and the Triple Risk of an AI Race
链接: https://arxiv.org/abs/2610.17085
作者: Matthew Rendall,Max Hellrigel-Holderbaum,Leonard Dung
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI)
备注:
Abstract:On many estimates, artificial general intelligence (AGI) is likely to emerge within the next few decades or even sooner. Once it does, it may quickly surpass human intelligence, possibly by a huge margin. Some believe that the country that develops AGI first could attain global dominance. This prospect risks setting off intense competition between the United States and China. Indeed, a narrative has already developed that the superpowers are locked in an AI race, with the Trump administration proclaiming its determination to ‘dismantle unnecessary regulatory barriers’ to winning it. We argue that this approach presents three profound dangers. The first two are familiar from past arms races: being overtaken by a rival or triggering a war. The third is new: losing control of human beings’ future to agents far more intelligent than we are, which could endanger our very survival. Striving to get to AGI first risks a race to the bottom on safety practices and regulation concerning them. China does not yet appear bent on racing to superintelligence. The United States should prioritise both unilateral and co-operative efforts to improve AI safety and avoid provoking a race that all humanity could lose.
[AI-199] Gradient Routing Localizes and Removes Unintended Behaviors in RL
链接: https://arxiv.org/abs/2610.17062
作者: Jake Ward,Shawn Hu,Aria Wong,Nathan Hu,Bryce Woodworth,Alexander Matt Turner,Alex Cloud
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:We propose a mitigation to reward hacking in reinforcement learning. Our setup involves a misspecified reward function that gives rise to unintended behavior, and a monitor used to detect unintended behavior. As with frontier AI systems, the monitor makes systematic errors. Consequently, using the monitor for data filtering or to apply a reward penalty fails to prevent the model from learning unintended behavior (“reward hacking”). To address this problem, we introduce GRAFT: Gradient Routing Adapter Fine Tuning. Rather than penalizing or filtering flagged samples during training, GRAFT constrains gradient updates on those samples to a sacrificial set of new parameters. This has two benefits: First, because the monitor is not part of the reward function, the model does not learn to hide misbehavior. Second, ablating the sacrificial parameters reduces or entirely stops the unintended behavior while maintaining improved performance on target tasks. We demonstrate these benefits in RL environments where reward penalties cause models to learn to reward hack in undetected ways, training models ranging from 135M to 8B parameters. Additionally, we show a similar effect at a larger scale by distilling frontier LLM trajectories containing reward hacking into Qwen3-32B, without teaching that model to reward hack. Overall, our work suggests GRAFT may be a promising approach for leveraging imperfect monitors during post-training in settings where true performance is difficult to evaluate.
[AI-200] PatientWM: Towards Generalizable Clinical World Models via Unified Patient States
链接: https://arxiv.org/abs/2610.17059
作者: Ziwen Kan,Jie Peng,Bo Yan,Davis Hobley,Zheng Zhang,Yu Tian,Tianlong Chen,Song Wang
类目: Artificial Intelligence (cs.AI)
备注: 30 pages, 6 figures, 14 tables
Abstract:Clinical decisions depend on anticipating how a patient’s condition will evolve as treatment is adjusted and new evidence becomes available. Existing medical world models capture patient-state evolution, but often define the evolving state around a particular modality or task, limiting reuse as inputs or prediction targets change. These models also retain history through observation sequences or recurrent or latent summaries rather than exposing preceding patient states and clinical actions as an ordered transition context. This distinction matters because similar current measurements may follow different recent changes, treatment exposures, and observed responses. To tackle these limitations, we propose PatientWM, a clinical world-model framework built around a unified patient state and sequential latent history. Availability-aware encoders integrate observations available at each decision time into the patient state, while typed decoders produce task-specific clinical readouts without defining the state around any single target. PatientWM further conditions each next-state prediction on the current state and supplied action together with a bounded, ordered history of preceding latent patient states and clinical actions, preserving recent treatment and response context. Experiments across MIMIC-III, MIMIC-IV, and three external clinical datasets show superior performance in clinical prediction, multi-step forecasting, and action-planning evaluation. Transfer experiments further demonstrate the generalizability of PatientWM’s patient state. Together, these results establish its ability to forecast clinical evolution and support planning.
[AI-201] Neos: Learning to Interleave Search and Visual Reasoning
链接: https://arxiv.org/abs/2610.17043
作者: Ekko He,Jack Hong,Shilin Yan,Xintong Zhang,Yujie Zhang,Wenjie Li,Haoran Sun,Kaixun Jiang,Jiyuan Fu,Guannan Zhang,Hongwei Xue
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Recent multimodal deep search agents have advanced tool use and visual content retrieval, yet they still struggle to convert intermediate visual evidence into actionable guidance for subsequent search, limiting their effectiveness in long-horizon settings. To address this gap, we present Neos, a multimodal deep search framework that learns both how to acquire evidence and how to use it to guide subsequent decisions. To make this coupling learnable, Neos constructs training tasks with intermediate visual dependencies while filtering out textual shortcuts, and introduces a lightweight, trainable Multimodal Evidence Interface (MEI) that transforms retrieved content into goal-conditioned observations. At its core, an agent-environment co-evolution procedure treats MEI as a co-adaptive component of this loop. Specifically, reference-guided credit assignment uses verified successful trajectories to localize critical policy errors while preserving credit for useful decisions in failed searches, whereas rollout feedback selects informative local evidence-processing interactions for MEI supervised fine-tuning (SFT). By alternating policy reinforcement learning (RL) with MEI SFT, Neos allows the policy’s evolving evidence requests to shape MEI training and the adapted MEI to reshape subsequent observations, closing the loop between evidence acquisition and evidence-conditioned action. Across ten benchmarks, Neos-9B and Neos-35B demonstrate leading performance among specialized multimodal deep-search agents.
[AI-202] SPLICE: Sparse Physics Library Identification and Compositional Expansion
链接: https://arxiv.org/abs/2610.17029
作者: Lucas Thil,Jesse Read,Guillaume Doquet,Rim Kaddah
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 24 pages, 6 figures
Abstract:Neural operator splitting composes a library of neural operators, one per physical mechanism, into zero-shot simulators of new equations. Existing approaches, however, require these mechanisms and their labels to be specified in advance. We ask whether a pretrained library can instead be adapted and grown from unlabeled trajectories in which multiple mechanisms act simultaneously. We introduce SPLICE, a framework that discovers operators without mechanism labels on mixture data. The contributions of the candidate operators to each trajectory are inferred by solving a convex optimization problem. A calibrated test then assigns each discovered operator the mechanism it implements. Failure cases are also detectable, when the library lacks the required physics or when mechanisms are insufficiently separable. Across one and two dimensional systems, the discovered operators recover mechanisms from unlabeled mixtures and compose into zero-shot simulators of unseen equations; a frozen library can also grow an operator for physics absent from pretraining, without labels and without forgetting. On Navier-Stokes, the discovered operators compose about as accurately as their supervised counterparts.
[AI-203] Benchmarking Quantum Kernel SVMs for Credit Card Fraud Detection: Performance Realities Under NISQ Constraints
链接: https://arxiv.org/abs/2610.17027
作者: Hitanshi Arora,Mohit Swarnkar,Aswani Kumar Cherukuri
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Detecting credit card fraud is notoriously difficult because fraudulent charges make up less than 0.2% of total transactions. Quantum Machine Learning (QML), particularly Quantum Kernel Support Vector Machines (QSVM), promises richer feature representations by embedding classical data into high-dimensional Hilbert spaces. However, whether this theoretical advantage holds up when managing real-world financial data bottlenecks remains unclear. We designed a controlled benchmark comparing three quantum encodings, Angle, ZZFeatureMap, and IQP against classical models while keeping sample sizes, feature counts, and scaling strictly matched. To handle class skew without corrupting quantum state overlaps, we embedded synthetic minority oversampling (SMOTE) directly into the quantum Gram matrix workflow, raising PR-AUC scores by roughly +0.04 across feature maps. When trained on the full European credit card dataset with 30 features, classical baselines reached a top PR-AUC of 0.983. However, restricting classical models to match QSVM hardware constraints (500 samples, 4 PCA features) dropped their PR-AUC to 0.750. This puts our best quantum result 0.730 PR-AUC using IQP encoding on 150 samples into realistic context. Testing on physical quantum hardware reduced performance further to 0.62 due to CNOT gate errors and thermal noise. Overall, our findings pinpoint two main operational bottlenecks for near-term quantum kernels: the scaling of Gram matrix evaluation and heavy information loss from squeezing features onto small qubit registers.
[AI-204] CWE-bench: Refocusing Agent ic Security from Exploitation to Defense
链接: https://arxiv.org/abs/2610.17020
作者: Gonzalo Gonzalez-Pumariega,Ajay Jagannath,Adit Jain,Guram Gogia,Soumyadeep Bakshi,Nazneen Rajani
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: 12 pages, 6 figures, 5 tables
Abstract:Cybersecurity evaluations of AI agents have focused largely on offense, yet a working exploit does not show that an agent can make software safe. An attacker needs to exploit only one vulnerability; a defender must find and fix all of them. We introduce CWE-bench, a private benchmark of 120 held-out repository audit-and-repair tasks spanning 73 CWE types, eight languages, and all OWASP Top 10 2025 categories. Agents are given a repository to audit and a description of its threat model, but are not told the weakness, where it occurs, or how many vulnerable sites there are. Each patch must block the exploit while preserving original functionality, as checked by a programmatic verifier. We validated the verifier against a panel of agentic judges, manually reviewing disagreements during development and refining the checks. The best agents reach 68% pass@1 and 81% pass@4. We further conduct analyses showing that (i) agents with the same pass@1 differ widely in retry gains and cost, and (ii) no single agent is best across languages or vulnerability categories, and difficulty varies far more by vulnerability category than by language. As the offensive capabilities of agents improve, CWE-bench provides a broad and rigorous measure of whether their defensive capabilities keep pace.
[AI-205] ProbabilityBench: Model Evaluation and Calibration with Exact Ground-Truth Uncertainty
链接: https://arxiv.org/abs/2610.17016
作者: Yuyang Dai,Yuting Tang,Futoshi Futami,Takashi Ishida
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注:
Abstract:Evaluating model uncertainty on real-world data is challenging because each instance provides only a single observed label, leaving true conditional class probabilities unknown. To address this, we introduce ProbabilityBench, a benchmark that provides exact, instance-level class probabilities across three simulated text-based domains: peer review, covert information transfer, and credit risk. By explicitly defining the full data-generating process for each environment, we analytically compute both exact class probabilities and Bayes error. Furthermore, we provide a theoretical guarantee that rendering these cases into natural-language memos strictly preserves their underlying generative probabilities. This exactness enables direct measurement of reported probability quality via KL divergence. Across 16 evaluated models, most models on the observed cost-quality Pareto frontiers are decision models, including Jev and Perplexity Decider. Finally, ProbabilityBench serves as a meta-evaluation framework, leveraging known ground-truth values to benchmark the accuracy of standard calibration metrics and Bayes-error estimators themselves.
[AI-206] Governing Copyright Responsibly in the Era of Generative AI
链接: https://arxiv.org/abs/2610.17002
作者: D. Kochetkov
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Digital Libraries (cs.DL)
备注:
Abstract:Large language models are trained on collections that may include scholarly works, while authors generally lack work-level information about whether and how their works were used. This paper asks how copyright governance frames and operationalises the protection of such authors, and what a responsible alternative would require. Forty legislative, judicial and soft-law instruments issued between 2019 and 2026 were coded deductively against a framework derived from Responsible Research and Innovation theory, yielding 1,337 category assignments across 835 segments. The European Union and the United States are the primary cases; the United Kingdom, China and international bodies serve as comparators. Normative orientation and the distribution of responsibility dominate the corpus, while reflexivity is the least represented category and reaches its minimum in the US sub-corpus. Both primary jurisdictions frame protection procedurally rather than as an entitlement, and no instrument in the corpus creates a general author-facing attribution duty for the use of works in AI training. From the coded gaps, the paper derives a six-instrument governance design and a capacity-based allocation of responsibility among developers, publishers, collecting societies, funders, institutions, regulators, and individual authors.
[AI-207] Glasseek: A Transparent Recipe for Training and Trustworthy Evaluation of DeepSearch Agents
链接: https://arxiv.org/abs/2610.16998
作者: Cong Pang,Jiaqi Su,Chengrui Wu,Kuihao Yan,Tianrui Zhang,Zixuan Qiu,Chao Lin,Zixiu Lin,Zixuan Chen,Zelin Xiao,Tiankuo Yao,Jiawei Hong,Zixuan Chen,Dan Xu,Lewei Lu,Wenjie Wang,Xin Lou
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Deep search enables language agents to solve complex information-seeking tasks through iterative web exploration. However, developing robust open-source agents remains challenging for three reasons. First, opaque trajectory-construction pipelines make training data difficult to audit and reproduce. Second, context compression can create a substantial mismatch between training and inference. Third, agents may retrieve benchmark content from the open web during evaluation, contaminating the results and overstating their performance. To overcome these bottlenecks, we introduce Glasseek, a fully transparent deep search framework spanning the complete development lifecycle, releasing all data, model weights, training recipes, and code. Specifically, for data construction, we establish a verifiable trajectory curation pipeline that systematically generates high-quality exploration paths. For training, we propose Compaction-Managed GSPO (CM-GSPO), which uses the same policy for search and summarization while optimizing only execution sub-trajectories, enabling end-to-end search-policy training across context-compaction boundaries. For evaluation, we conduct the first systematic audit of test-set leakage in open-web agents and release a modular anti-leakage plugin that seamlessly integrates into existing evaluation harnesses. Under this rigorous evaluation protocol, our 9B agent outperforms frontier proprietary agents across almost all deep research benchmarks spanning diverse task types, and achieves the best overall performance among recent open-weight agents, setting a new state of the art at the 9B scale.
[AI-208] abGift: A Context-Efficient Foundation Model for Tabular Data Generation
链接: https://arxiv.org/abs/2610.16990
作者: Yu-Sheng Li,Jun-Peng Jiang,Han-Jia Ye
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Tabular data generation aims to synthesize realistic samples from limited observations while faithfully capturing the underlying data distribution. Tabular foundation models (TFMs), pretrained on diverse tabular tasks and adapted through in-context learning, have demonstrated strong predictive capabilities, suggesting the potential of extending this paradigm to tabular data generation. However, the strong predictive performance of TFMs often relies on sufficiently large context sets, whereas tabular generation is typically needed precisely when target data are scarce. Extending the foundation-model paradigm to generation therefore requires reliable distribution inference from only a limited context. To this end, we propose TabGift, a context-efficient tabular foundation model enabling both efficient distribution modeling and effective generation from only a few context samples. TabGift maps heterogeneous tables into a shared latent space with a pretrained tabular encoder, where an in-context diffusion model learns to denoise latent samples conditioned on a small set of clean examples. This design transfers generative priors across datasets and enables distribution inference from limited context, while a lightweight dataset-specific MLP decoder maps generated representations back to the original data space. Experiments on 7 classification and 11 regression datasets show that TabGift consistently outperforms representative baselines in synthetic data quality, with particularly strong gains under limited-context settings, while achieving substantially faster adaptation and generation.
[AI-209] From Suboptimal to Optimal: One-Step Latent Flow Policies for Offline Reinforcement Learning
链接: https://arxiv.org/abs/2610.16984
作者: Shenghua Lu,Zixuan Yang,Rongxu Zhao,Bohao Qu,Jifeng Hu,Hechang Chen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Offline reinforcement learning aims to learn effective policies from fixed datasets without environment interaction. Although generative models such as diffusion and flow matching enhance policy expressiveness, they typically start from uninformative Gaussian noise and generate actions through multi-step iteration, leading to high computational costs and failing to effectively leverage suboptimal trajectories in datasets. We propose MeanFlow2Better (MF2B), which reformulates offline RL as generative decision chunk optimization from current performance status to better performance status in a high-level latent decision space. MF2B maps trajectory chunks to high-level latent decision representations through an encoder, placing generation starting points within a structured decision space sculpted by behavioral data. The model learns a shared conditional mean velocity field in this space, constructing two flow mappings: from Gaussian-Induced Decision Sources toward positive behavioral decision regions, and from suboptimal decisions toward high-value decision regions. Both flow mappings complete decision refinement through one-step ODE integration over the unit time interval \Delta t=1 , requiring neither multi-step iterative sampling nor backpropagation through time. Furthermore, we propose hierarchical chunk-level advantage to guide the flow direction toward high-return regions, where the high-level decisions cover multiple low-level actions, leading to sequence length compression from T to K \approx T/H and significantly lower computational overhead. On the D4RL Gym-MuJoCo benchmark, MF2B achieves an average score of 90.91, matching or surpassing strong baselines.
[AI-210] PRAISE: A Reproducible Workflow for Non-Programmer Scientists to Implement Published AI Models
链接: https://arxiv.org/abs/2610.16980
作者: Max Achterweust,Stefan Groot Nibbelink,Vincent J. de Beer,Stefan Schipper,Zeynep Yüksektepe,Pedro Andrade da Silva,Aurelia Peppelaar,Martijn J. van der Willigen,Christophe B. Minkenberg,Robert F. van der Willigen
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 20 pages, two columns, LaTeX, 5 figures
Abstract:Artificial Intelligence (AI) models are increasingly developed, published and used across scientific fields. Researchers, students and industry professionals who want to apply these models to their own research questions often lack the programming and data science expertise to do so. Existing guidelines cover single steps, but not the route from a published model to working predictions on one’s own data. Here we present the hands-on workflow PRAISE, a Practical, Reproducible AI-model Implementation for Science and Education, that helps non-programmers implement published AI models step by step. Because implementations rarely work the first time, PRAISE anticipates common failures and provides a fallback route for each. As a proof of principle, three BSc chemistry students and their tutor applied PRAISE without prior programming or data science knowledge. They reproduced, validated and extended a state-of-the-art model that predicts enzyme-substrate interactions. This shows that with PRAISE, implementing an AI model from the literature becomes a learnable skill, one that turns non-programmer scientists from users into evaluators and improvers of AI models.
[AI-211] MADDIE: Multi-Agent Drug Discovery Engine
链接: https://arxiv.org/abs/2610.16975
作者: Ye Liu,Reza Averly,Avery Meyer,Frazier N. Baker,Ruoxi Gao,Daniel Adu-Ampratwum,Xia Ning
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Drug discovery remains a resource-intensive and time-consuming process, involving billions of dollars and a decade of development to advance a drug to market. While many in silico tools have accelerated this process, operating them still demands substantial time, effort, and expertise from medicinal chemists. Autonomous systems built on Large Language Models (LLMs) have emerged as a promising solution, yet current designs limit their performance. Existing single-agent systems scale poorly to long-horizon, multi-step, and realistic tasks, whereas multi-agent systems typically execute predefined workflows that cannot capture the iterative, non-linear nature of a real drug discovery campaign. Here, we present MADDIE, a multi-agent drug discovery engine. It comprises a central agent, Orchestrator, and four specialized worker agents: HitIdentifier, LeadOptimizer, Synthesizer, and Evaluator, each representing a critical step in a drug discovery campaign. Under Orchestrator’s orchestration and coordination, these worker agents are composed into flexible workflows, enabling MADDIE to adapt its strategy and navigate the drug discovery pipeline end-to-end. Across 50 protein targets, MADDIE produces molecules satisfying key property requirements and their synthesis routes, outperforming existing agentic systems, generative models, and general-purpose LLMs by a wide margin. Our analysis reveals how these agents collaborate and that MADDIE can employ different strategies across targets to achieve the campaign’s goal, while still being efficient in time and tokens. Together, they demonstrate MADDIE as a low-cost, effective, and efficient autonomous engine for accelerating drug discovery. Code and dataset is available via this https URL.
[AI-212] Social Priors: Descriptions of AI Users Shape How Language Models Trust and Cooperate
链接: https://arxiv.org/abs/2610.16946
作者: Harsh Kumar,George Eilender,Ashton Anderson
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注:
Abstract:Descriptions of how people use AI are becoming training data for the next generation of models. We ask whether what a model reads about its users changes how it treats people. Using contrastive synthetic-document finetuning, we train Qwen3-32B on four matched corpora built from the same 1,625 real-world usage scenarios. The corpora differ only in the predominant motive they attribute to users of AI chatbots: helping others (prosocial), exploiting others (antisocial), legitimate self-advancement, or task completion. Finetuning shifts the models’ stated beliefs about users, how true those beliefs look along an internal truth direction, and their broader judgments of how safe and good the world is. These social priors carry into behavior. In trust games, antisocial models expect less reciprocity and send less money than prosocial models. Their distrust is specific and targeted at people, and users of AI chatbots in particular. After an exploitative record, they send almost nothing to a person but most of their endowment to a random device with the same record. In a common-pool resource simulation, a single antisocial agent leads every group of otherwise sustainable agents to exhaust the resource, after privately planning to mislead them. Antisocial training also raises dark-triad scores and lowers alignment ratings. Belief shifts appear when antisocial documents are 1% of training tokens, persist when descriptions of assistant behavior are suppressed, and are not reproduced by training on discourse about misaligned AI. Capping Assistant-Axis drift and steering emotion-related directions leave the gaps largely intact. What models learn about the people they serve shapes how they subsequently treat them and others.
[AI-213] UA-Bench: Benchmarking Large Audio-Language Models on Underwater Acoustic Recognition
链接: https://arxiv.org/abs/2610.16942
作者: Zhe Sun,Liang An,Jiayu Yao,Ronglai Qian,Jianing Cai,Ziwei Huang,Qing Fan,Zhiyuan Yang,Yujun Cai,Hongli Cao,Yiwei Wang,Xiaoyan Wang
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注: 31 pages including appendix, 12 figures
Abstract:Large audio-language models (LALMs) are fast becoming a general interface to sound, yet how far their broad pretraining carries into specialized domains remains largely open. Vessel monitoring, maritime safety, and ocean sensing all depend on recognizing underwater signals, but to our knowledge LALMs have not been systematically evaluated on them. Annotated recordings are costly to collect, and a faithful synthetic alternative has to bring together many interacting factors: the structure of detection pulses, communication waveforms, and ship-radiated noise, and the sound-speed profile, geometry, and multipath propagation through which each reaches the receiver. We address this by proposing \textbfUA-Bench, to our knowledge the first large-scale benchmark for evaluating LALMs on underwater acoustic signal recognition. It comprises 375,051 audio records that couple parameterized sources with modeled propagation, so that every record can be traced to the conditions that produced it. The records span these three signal families, and each carries a three-level label hierarchy (13 leaf classes), a templated multi-turn dialogue, and source and channel metadata. We evaluate on a balanced 2,600-record subset with a gated coarse-to-fine protocol that locates each error at source routing, fine-grained classification, or output format, and add a paired silent-audio control to test whether the waveform is actually used. Nine open LALMs, evaluated without task-specific adaptation, reach only 2.0%–13.1% leaf accuracy, against a random reference of 7.7%. The waveform helps source routing far more than leaf recognition (for one model, +29.5 versus +2.3 points over silence), and just four models gain reliably at the leaf level, by at most 4.7 points. UA-Bench thus gives a reference point for these models, and its released training partition supports follow-up adaptation work.
[AI-214] WeFair: Fidelity-Aware Alignment from Human Feedback for Personalized Query Rewriting in Short-Video Search
链接: https://arxiv.org/abs/2610.16927
作者: Yingtao Ren,Aolin Li,Cheng Cheng,Huiyun Hu,Huasheng Liang,Baoxing Xie,Jun Yuan,Haijun Wu,Chenxing Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Short-video search requires compact query rewriters that infer personalized intent from brief, ambiguous queries under strict latency constraints. The challenge is to decide when personalization is warranted and generate a reformulation that preserves the user’s immediate information need. A rewriter optimized for fluency or engagement may still introduce unsupported details and redirect the search toward a different intent. Existing methods often underutilize users’ actual reformulations and behavioral traces, relying instead on synthetic reasoning and indirect behavioral rewards that may produce more engaging candidates but cause intent drift. To build an effective rewriter that remains faithful to user intent, we propose WeFair, Fidelity-Aware Alignment from Human Feedback for Personalized Query Rewriting in WeChat short-video search. WeFair assigns real user behavioral signals to a progression from demonstration to preference to verification. We first mine actual human reformulations from search sessions, then apply multidimensional LLM screening and contextual verification to filter unsupported transitions without extensive manual annotation. During training, we begin with a supervised curriculum that progressively builds rewriting capabilities from large-scale human personalized query demonstrations, and then use direct preference optimization to align with users’ subsequent consumption. In the final on-policy optimization stage, we construct a strict verification reward based on actual human reformulations to further boost performance with group relative policy optimization. Further analysis shows that auxiliary rewards can activate more rollout groups in GRPO but reduce intent fidelity. Online A/B testing in WeChat short-video search shows a 1.576% relative increase in effective clicked-query rate (VV10s) and a 1.731% relative reduction in query reformulation rate.
[AI-215] ransparent Information Flow Modeling for Role-Aware AI Act Assessment ECAI2025
链接: https://arxiv.org/abs/2610.16916
作者: Mattias Brännström,Themis Dimitra Xanthopoulou,Lili Jiang
类目: Artificial Intelligence (cs.AI)
备注: 19 pages, 3 figures. Presented at the CLEAR Workshop, ECAI 2025
Abstract:The European Union’s Artificial Intelligence Act introduces a structured framework for governing high-risk AI systems, emphasizing transparency, risk management, and role-specific accountability. However, current assessment practices often rely on monolithic system views or disconnected checklists, limiting their ability to capture the socio-technical complexity of real-world deployments. We explore the use of the Information Flow Methodology (IFM), a modeling-based approach to AI Act compliance assessment. IFM represents decision systems – whether algorithmic, human, or hybrid – as networks of information sites and transformation channels, enabling structured analysis of how information, decisions, and responsibilities flow across organizational and technical boundaries. We apply IFM in a real-world case involving a recruitment support tool used by a public-sector organization, evaluating both the Provider’s system and the Deployer’s recruitment process. Our results demonstrate that IFM can serve as a scaffold to bridge both component-based and holistic perspectives, and how technical systems are situated within socio-technical contexts. In doing so, IFM supports transparency, identifies risk pathways, and enables alignment analysis between intended and actual use. We argue that such modeling provides a foundational tool for effective, context-sensitive AI governance and accountability – offering capabilities beyond conventional checklist-based assessments.
[AI-216] Audit tools of the trade: Automated Auditor Elicitation Can Be Sensitive to the Context the Auditor Is Given NEURIPS2026
链接: https://arxiv.org/abs/2610.16911
作者: Ben Slater,Ryan Lundqvist,Leo Richter,Kevin Wei
类目: Artificial Intelligence (cs.AI)
备注: 23 pages, 13 figures, 9 tables. Accepted at the Third Workshop on Agents in the Wild: Safety, Security, and Beyond (AIWILD) at NeurIPS 2026
Abstract:Does providing additional context to automated AI auditors improve their effectiveness? Using the Petri tool to audit LLM coding agents, we test the impact of providing two types of resources (an external codebase and benchmark transcripts) to LLM auditors, across four auditor-target LLM pairs and three behaviours (harmful compliance, reward hacking, sycophancy). (1) External codebase: We provide the auditor LLM with a fixed, external open source codebase, which prior work finds improves the realism of synthetic audits. Providing this resource suppresses reward hacking in both auditor-target pairs whose baseline rate is above the floor (4.8 to 1.3 and 3.9 to 1.8 on Petri’s default 1-10 scale); the remaining two pairs sit at the floor throughout. The suppression recurs in three of four pairs under a different scenario and resource bundle. (2) External benchmark transcripts: We provide the auditor with transcripts of the target’s performance on benchmarks measuring similar behaviours, intended to inform the auditor about the target’s behavioural patterns. Providing this resource raises reward hacking scores in all four pairs (from 1.7 to 6.1 in one case). However, subsequent experiments show that supplying only the related benchmark’s abstract, without transcripts, also increases reward hacking, and the effect is absent under a different framing of the same behaviour. Overall, we find no consistent effect of either resource outside reward hacking, but we find substantial variability in judge-scored behaviour across conditions. Our results underscore the importance of careful piloting with tools such as Petri, as the measured rate of target behaviour is highly sensitive to configuration decisions. We hope our results inspire more research into the reliability and robustness of automated alignment auditing.
[AI-217] Neural Bayesian Networks for Modular and Scalable Decision-Making Under Uncertainty CCS
链接: https://arxiv.org/abs/2610.16902
作者: Giovanni Briglia,Marco Lippi,Stefano Mariani,Franco Zambonelli
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Main text 10 pages, extra is supplementary. ACM CCS: Computing methodologies~Artificial intelligence~Knowledge representation and reasoning~Probabilistic reasoning. Computing methodologies~Machine learning~Machine learning approaches~Neural networks. Computing methodologies~Machine learning~Machine learning approaches~Learning in probabilistic graphical models~Bayesian network models
Abstract:Bayesian Networks (BNs) are established tools for letting agents reason under uncertainty in sequential decision-making. They split a joint distribution into local conditional mechanisms arranged on a directed acyclic graph, which makes them modular, interpretable, and ready to perform probabilistic inference. Their classical formulation, however, has limitations in both expressiveness and scale: each conditional is usually stored as a table or a linear-Gaussian form, and inference evaluates the graph one node and one query at a time. In this paper we introduce Neural Bayesian Networks (NBNs), in which each conditional is a differentiable, tensor-valued mechanism, while the graph and its factorization are unchanged. NBNs retain the modularity and local mechanisms of BNs, can represent any learnable conditional distribution, and run natively on GPUs. We also release NBN, a PyTorch-based library for learning and inference in NBNs, together with a public benchmark. Our evaluation shows gains in both expressiveness and scale: NBNs fit distributions that classical BN libraries struggle to represent, and on large networks and large batches of queries they run two orders of magnitude faster thanks to GPU parallelization. NBN is available at this https URL. Comments: Main text 10 pages, extra is supplementary. ACM CCS: Computing methodologies~Artificial intelligence~Knowledge representation and reasoning~Probabilistic reasoning. Computing methodologies~Machine learning~Machine learning approaches~Neural networks. Computing methodologies~Machine learning~Machine learning approaches~Learning in probabilistic graphical models~Bayesian network models Subjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG) ACMclasses: I.2.6; I.2.11 Cite as: arXiv:2610.16902 [cs.AI] (or arXiv:2610.16902v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2610.16902 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-218] Observe Share or Predict? Disentangling Partner Information in Embodied Multi-Agent Collaboration
链接: https://arxiv.org/abs/2610.16895
作者: Adrián Szlatincsán,Filippo Ziliotto,Bruno Lepri,Luciano Serafini,Massimiliano Luca,Tommaso Campari
类目: Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注:
Abstract:LLMs are used as high-level controllers for embodied agents, but it remains unclear whether they support collaborative reasoning, such as anticipating a collaborator’s actions in real-time teamwork without dialogue. We ask if observing the partner’s state, exchanging object knowledge, and predicting the partner’s next action, shape task outcomes and coordination when agents can share percepts but never intentions, plans, or task-level messages, and whether the two agents occupy symmetric roles. Using the Habitat-based PARTNR benchmark, we introduce a Decoupled Partial Observability regime that enforces strictly individual perception, and we augment decentralized zero-shot ReAct planners with Other Agent’s State (OAS) observation, object Knowledge Exchange (KE) between co-located agents, and Other Agent’s Next Action (OANA) prediction, evaluated on the full 2 x 2 x 2 grid with four planners on 1,000 validation tasks. OAS observation improves task quality, reducing extraneous effort for every planner, whereas KE primarily improves execution efficiency, cutting simulation steps by up to 25% with little change in success. The two are largely redundant for strong planners but complementary for the weakest, for which KE without the OAS is the worst configuration. OANA prediction lowers the success rate in 15 of 16 model x context pairs while raising explicit partner mentions to 55-100%, and its accuracy gains lie mainly in the coarse action type, partly because agents recognize the action the partner is already performing. The robot retrieves objects sooner and completes its tasks earlier than the simulated human. Shared context reduces total task time by shortening the remaining duration that the human works alone. Also, coordination benefits arise from responding to shared context, especially the OAS, rather than explicit anticipation, yielding shallow, planner-dependent partner modeling.
[AI-219] ransparency as Answerability: Role-Dependent Information Flow Modeling
链接: https://arxiv.org/abs/2610.16887
作者: Mattias Brännström,Themis Dimitra Xanthopoulou,Lili Jiang
类目: Artificial Intelligence (cs.AI)
备注: 30 pages, 2 figures, 5 tables. Under revision at ACM Journal on Responsible Computing
Abstract:Transparency in AI-supported decision systems is widely required but poorly defined. Existing approaches treat transparency as a property of systems or components, producing characteristic failures: assessments that do not transfer across organizational boundaries, stakeholder questions that cannot be formulated within the available frame of analysis, and component-level findings that say nothing about system-level behavior. We propose treating transparency as a relation: a system is transparent to a stakeholder to the degree that their legitimate questions are answerable from its structure. We formalize this using information flow networks, in which transparency questions are expressed in terms of network structure and answerability is the condition that the system settles them. The framework supports three dimensions of analysis: structural answerability from the network alone, compositional answerability across organizational boundaries for deployed systems, and role-dependent answerability given a stakeholder’s accessible sub-network. We demonstrate the framework on a real pre-integration assessment of an AI recruitment tool in a public sector municipality. The assessment uses participatory design and interviews to model the decision system structure. We show how structural and role-dependent transparency questions can be formally stated and analyzed across the Provider and Deployer boundary where the AI system is composed together with the deployer’s decision process.
[AI-220] Acceptable Misalignment
链接: https://arxiv.org/abs/2610.16881
作者: Maud van Lier,Ludovica Marinucci,Vieri Giuliano Santucci,Raja Chatila,Stéphane Doncieux,Émiland Garrabé,Richard Duro
类目: Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: 16 pages, 1 figure
Abstract:There is a specific challenge designers face in developing autonomous robots capable of open-ended learning. This challenge is how one can develop such systems in an ethically sound way when there are many unknowns during design time about their manner of deployment, what they can and will learn, and how these systems will act on the basis of what they have learned. One response to this challenge is to align, as far as possible, the activities and goals of autonomous robots with human values, raising the question of how to make this alignment possible, a question known as the value alignment problem. We argue that the value alignment problem faces a number of challenges that show that full alignment is both theoretically and practically impossible, meaning that misalignment incidents will keep occurring. This undermines current alignment approaches, as the attempt to maximize alignment cannot properly deal with the possibility that the risk of misalignment that remains after alignment methods have been implemented might still be unacceptable to society. To overcome this difficulty, we argue that a shift in perspective is required from a focus on the maximization of alignment to the conditions under which misalignment incidents become acceptable. In this paper, we explore what this shift in perspective entails. We introduce the concept of acceptable misalignment, formulate normative requirements for alignment mechanisms aimed at bringing misalignment incidents to an acceptable level, and we conclude by presenting a new methodological framework that consists of alignment mechanisms that can meet these requirements.
[AI-221] CLIVO: Differentiable Cluster-Based Instruction Mixing with Learned Validation Objectives
链接: https://arxiv.org/abs/2610.16876
作者: Yuhan Wang,Rui Hu,Batian Di,Tan Jiang,Dongsheng Ma,Leichao Dong,Yuanzi Li,Yalin Feng
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Target-aware multimodal instruction selection requires useful examples and an appropriate balance of supervision. Sample-level rankings can overrepresent similar content without optimizing this balance; group-level methods face coarse units, predefined set-level validation signals, and costly or inaccurate mixture search. We propose CLIVO, a target-aware framework connecting fine-grained semantic allocation, automatically learned validation objectives, and differentiable mixture search. Specifically, semantic clusters provide independently adjustable shares at the hundred-cluster scale. Correlations between per-example merged-proxy losses and real probe performance identify target-relevant validation examples; learned nonnegative weights combine their losses into an objective. A single backward pass through the merged proxy supplies gradients for all shares, avoiding an additional mixture-response regression model. Selecting 100K instructions from an 8M pool, CLIVO attains 97.2% of the full-data four-target average score with only 1.25% of the data. It improves on the strongest evaluated selection baseline by 1.16 points using random within-cluster sampling; combining learned quotas with sample ranking increases this advantage to 1.55 points. Analyses show that finer units benefit most targets, short-trained experts preserve mixture responses, sample weighting retains useful validation information, and direct search consistently reduces the learned objective. CLIVO also supports interpretable allocation, modular selection, and reuse across targets and granularities.
[AI-222] Few-for-Many Best-Arm Identification in Multi-Objective Bandits
链接: https://arxiv.org/abs/2610.16840
作者: Bo Xue,Yilu Liu,Haodong Jing,Hongzong Li,Xi Lin,Shuang Qiu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Real-world systems often serve many objectives but can deploy only a few options, whose performance must be learned from noisy feedback. Modeling each option as an arm in a multi-objective bandit, we introduce \emphfew-for-many best-arm identification, which seeks a small set of arms minimizing the worst weighted degradation from the objective-wise optima, rather than the best arm for every objective. We consider both known objective groups and latent groups specified only by a representative budget, and propose \textscGFM-Elim and \textscAFM-Elim with instance-dependent sample-complexity guarantees. For Gaussian rewards, we further derive a lower bound for the latent-group setting and show that it is asymptotically attainable with high probability by an oracle-based Track-and-Stop procedure. Our results show that sharing observations across objectives and ignoring irrelevant objective-wise distinctions can substantially reduce the sampling cost, as confirmed by experiments on synthetic and real-world tasks.
[AI-223] Same Operation Different Permission: Evaluating Whether Guardrails Honor Authorization
链接: https://arxiv.org/abs/2610.16838
作者: Huihao Jing,Shaojin Chen,Wenbin Hu,Yuxuan Liu,Haochen Shi,Zhongwei Xie,Haoran Li,Yangqiu Song
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:
Abstract:As language-model agents write code, move money and change records on behalf of users, deciding whether an action is authorized has become a central safety problem. Guardrails are often placed in front of these actions, yet few evaluations ask whether their decisions follow authorization: blocking unauthorized actions while still allowing the same actions under valid permission. We introduce \bench, a benchmark that contrasts authorization evidence for the same operation in its language track. It combines expert-annotated natural-language scenarios with executed and verified operations, plus externally sourced native-policy requests. \bench checks each decision for correctness, usability and resulting effect, and tracks each request across guardrail changes. Our experiments show that many blocked operations stem from failed reviews, not correct denials. Repairing the output format restores usable decisions but changes which authorized requests are honored, and adding a second reviewer removes some errors while rejecting valid authorization. Guardrails should therefore be evaluated on both the unauthorized actions they stop and the authorized actions they let through.
[AI-224] DTA-GT: Direction- and Topology-Aware Graph Transformer for Neural Network Representation Learning NEURIPS2026
链接: https://arxiv.org/abs/2610.16822
作者: Yuxiang Zeng,Kun Xie,Yang Wang,Jigang Wen,Xiaocan Li,Guangxing Zhang,Gaogang Xie,Jiannong Cao
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Neural and Evolutionary Computing (cs.NE)
备注: Accepted at NeurIPS 2026. 36 pages, 6 figures
Abstract:Architecture attribute predictors reduce the cost of neural architecture search by estimating accuracy or latency from a small set of evaluated candidates. The key difficulty is that neural architectures are operation-labeled DAGs: edge direction defines computational flow, but existing predictors often treat direction as a local structural cue. We propose DTA-GT, a Direction- and Topology-Aware Graph Transformer based on encoder-wide directional consistency, which preserves directed computational semantics across node initialization, global spectral encoding, pairwise interaction, and feature update. DTA-GT realizes this principle with direction- and topology-aware node initialization, Magnetic Laplacian spectral encoding, direction-dual structural attention, and a Direction-Sensitive MoE-FFN. Across accuracy and latency prediction on NAS-Bench-101 and NAS-Bench-201, DTA-GT consistently outperforms representative sequence-, GNN-, Transformer-, and hybrid predictors under limited-budget settings. Ablations and Magnetic spectral controls show that the improvements come from coordinated direction-aware stages and edge-orientation-sensitive spectral information, while mechanistic diagnostics confirm that the modules exhibit the intended directional behavior. These results support encoder-wide directional consistency as an effective design principle for architecture DAG representation.
[AI-225] StableJudge: A Lightweight Adversarial Defense Framework for LLM -as-a-Judge
链接: https://arxiv.org/abs/2610.16814
作者: Anning Hu,Yangyu Li,Yuqi Huang,Shengchao Hu,Junchi Yan
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 12 pages, including references and appendix
Abstract:Large language models (LLMs) are increasingly used as judges to assign scalar scores that drive downstream decision-making, including tasks such as paper review and recommendation. However, LLM-as-a-Judge pipelines are vulnerable to score-hijacking prompt injection embedded in evaluated content, which corrupts scoring signals and distorts induced rankings. Existing defenses often enforce robustness by modifying the judge model itself, making safety dependent on a specific model’s scale and training and requiring repeated effort when the judge is updated or replaced. We propose StableJudge, a lightweight defense framework that externalizes prompt-injection mitigation as a reusable pre-scoring capability. StableJudge centers on a compact 0.13B Sentry that localizes injected spans and performs targeted sanitization prior to evaluation, and is trained using adversarially evolving attacks to improve robustness beyond a fixed prompt set. To stabilize measurement, StableJudge derives deterministic, fine-grained scalar scores from the judge’s output distribution. Across held-out scientific abstracts and out-of-distribution attack variants, StableJudge improves score recovery under attack while largely preserving general-purpose performance, demonstrating that robust LLM-based judging can be achieved by decoupling safety from judge scale via reusable defense capabilities. Code and data will be released.
[AI-226] Logit Leakage through Floating-Point Bias Addition
链接: https://arxiv.org/abs/2610.16790
作者: Shaomin Chen,Xiaorui Gong
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 18 pages, 6 figures, 5 tables, including appendices
Abstract:Language model APIs can expose token-specific biases while withholding logits and token probabilities. Existing bias-based extraction recovers relative logits, leaving a common additive component undetermined in ideal arithmetic. We identify a numerical side channel in floating-point bias addition that exposes information about this component through token identities alone. The channel distinguishes score vectors with identical complete unperturbed log probabilities. Our method exploits changes in representable spacing at binary exponent boundaries. Under the stated binary32 arithmetic assumptions and a known reachable score interval, it recovers a logit pair and its unknown tie-breaking priority exactly in at most 52 token queries. Given complete relative logits, one recovered anchor determines the original logit vector. We evaluate recovery across 11 public checkpoints from seven model families and extend the local method with adaptive interval search and anchor selection. In end-to-end experiments with vLLM, SGLang, and this http URL, adaptive recovery succeeds on all 128 Qwen3-0.6B inputs per system, including 64 held-out inputs. A separate token-only experiment extracts a pretrained small model’s complete output-head representation up to an invertible hidden-basis transformation. We also establish a sufficient condition for suppressing common-shift disclosure by subtracting the maximum before bias addition. In controlled experiments, this mitigation eliminates all observed distinctions between common-shift variants and preserves token decisions under the tested random biases.
[AI-227] LIBERO-Industry: Benchmarking Robotic Manipulation under Industrial Process Constraints
链接: https://arxiv.org/abs/2610.16782
作者: Ming Shin Cindy Sung,Duan Leijiahao,Haoxiang Wang,Boning Tan,Liangna Fu,Chao Tan,Yafu Li,Yulong Chen,Chaoqun Wan,Zhen Huang,Shuxian Liang,Xian-Sheng Hua
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Industrial manipulation in flexible manufacturing, e.g., high-mix, low-volume (HMLV) electronics assembly, requires robot policies to adapt to changing products and production conditions. Such settings impose industrial process constraints, including product variants, moving material flows, finite execution windows, and tight geometric tolerances. However, existing evaluation frameworks do not systematically capture these constraints when assessing generalist robot policies. To address this gap, we introduce LIBERO-Industry, a novel benchmark built on the widely used robot learning benchmark LIBERO, that translates the aforementioned industrial constraints into controllable task conditions and measurable success criteria. Specifically, LIBERO-Industry comprises three classic task scenarios: Static-to-Dynamic (S2D) loading, Dynamic-to-Static (D2S) unloading, and Dynamic-to-Dynamic (D2D) transfer. The suites contain 216 tasks in total that systematically vary object geometry, conveyor speed, interaction-window duration, and geometric acceptance region size. Extensive experiments across five generalist robot policies (including vision-language-action models and world-action models) show that performance progressively degrades as industrial constraints are introduced and combined. Under a shared fine-tuning and evaluation protocol, the LIBERO-Industry co-training condition achieved higher observed target-transfer completion than the original-LIBERO condition.
[AI-228] PrivacyGuard: Source-Linked Sanitization and Locally Resolvable PII References for Cloud GUI Agents
链接: https://arxiv.org/abs/2610.16778
作者: Zian Yuan,Wei Li
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC)
备注:
Abstract:Cloud graphical user interface (GUI) agents send instructions and screenshots to remote planners, exposing personally identifiable information (PII). Existing placeholder defenses mask detected spans and reuse exact strings. Fragmented or changed forms can expose PII or break entity continuity across turns. PrivacyGuard addresses this private reference problem with source-linked sanitization (SAN) and a structured reference lifecycle (REF). SAN reconstructs fragmented or overlapping evidence and rewrites the complete source extent. REF links changed appearances to session entities and supports locally resolvable references without returning raw values to the upload path. In five-turn sessions scored against common gold, PrivacyGuard reduces audited source-value leakage by 33.55%–82.60% relative to the closest prior full-pipeline system in all four reader–language combinations. REF raises strict cross-form continuity from 0.04 under exact-value reuse to 0.81. In reference-required Android tasks, verified completion rises from 8/30 and 7/30 to 17/30 and 20/30. On AndroidLab, PrivacyGuard achieves the highest or tied-highest task-success point estimate among protection pipelines for all four planners. On PrivScreen, it reduces fixed-reader field exposure by 62.72%–62.84% relative to the same baseline. PrivacyGuard lowers upload exposure while preserving a locally resolvable task path.
[AI-229] Geospatial Vector Geometry as a Native Spatial Modality for Vision-Language Models
链接: https://arxiv.org/abs/2610.16769
作者: Huang Fei,Boyi Chen,Hao Xue,Yang Yue,Yehong Xu,Xiaofang Zhou
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Geospatial reasoning links VLMs to real-world action and decision making. Reliable reasoning requires accurate spatial information and effective spatial representations. However, current VLMs obtain this information mainly from language and vision. At larger geographic scales, limited visual observations make it difficult to establish accurate spatial references and recover precise spatial relations. This leads to a bottleneck for VLM spatial perception and reasoning. This paper proposes that geospatial vector geometry can become a native spatial modality within a VLM and introduces GeoVec-VLM. First, we develop a structured geometry representation that preserves the spatial structure and arrangement of vector objects, building an explicit spatial basis. Second, we introduce a geometry interaction strategy that brings this basis into multimodal computation, where it interacts with visual content and language references according to the queried spatial relations. Together, these designs allow perception and reasoning to operate on shared geometric references. Across five spatial reasoning tasks, GeoVec-VLM improves mean accuracy by about 11% relative to the strongest baseline, maintains its gains under geographic transfer, and shows greater geometric consistency. Our results show that geospatial vector geometry offers a promising internal spatial basis for VLM spatial reasoning. this https URL
[AI-230] Relational Learning as World Modeling
链接: https://arxiv.org/abs/2610.16763
作者: Zheyu Zhang,Gjergji Kasneci
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Relational learning typically predicts from a database state, but decision making asks a different question: how will that state change under an action? In operational systems, the answer depends on system-specific dynamics that may not be visible in the database itself. We view each such system as a relational “world” and ask whether a model can infer how it behaves from observed executions. We introduce TabWM (Tabular World Model) and WorldRel, a generator of executable relational worlds for pretraining. Given executions from an unseen world, TabWM uses them as context to predict the sparse structured edits caused by a new action without parameter updates. Across 512 held-out worlds, complete-edit F1 rises from 0.016 with no observed executions to 0.685 with 64, where a correct edit must match its operation, relational location, and value. Because these predictions are explicit relational edits, they can be applied recursively to imagine candidate futures and compare actions before they are taken. Broader search over these predicted futures improves the realized return of selected action sequences. Together, these results suggest that relational dynamics can be learned in context, broadening relational learning from modeling system state to modeling how that state changes under action.
[AI-231] Before Attention Sinks Form: Hidden-State Geometry Reveals and Modulates Their Formation
链接: https://arxiv.org/abs/2610.16745
作者: Fan Yang,Yue Song
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Attention sinks are usually studied after they have formed. We ask whether a measurable hidden-state change appears beforehand and whether changing that geometry alters what happens later. To remove network depth as a shortcut, we compare prompts in the same model and at the same layer. When sink timing varies across prompts, simple spectral summaries of the hidden states add information beyond current attention, with the clearest prompt-specific signal in a short window about two to three layers before sink formation. The predictive strength varies across cohorts, so we treat the spectrum as a readout of a pre-sink state change rather than a universal warning statistic. The signal is broader than one measured extreme activation. We then modify only the singular values of the post-layer hidden states after current attention has already been computed. These interventions change later sink behavior in Qwen2.5, GLM-Z1, and GPT-OSS relative to norm-matched random spectral perturbations. In Qwen2.5, the immediate suppressive effect is accompanied by a reduction in next-layer query–key preference for position~0, and the learned spectral directions remain stable when estimated from substantially less training data. Together, the results show that attention sinks are preceded by a short spectral reorganization of hidden states, and directly perturbing this geometry can causally alter subsequent sink formation.
[AI-232] Queue-Aware Token Communications: Bridging AI Representation and Wireless Service
链接: https://arxiv.org/abs/2610.16730
作者: Yuquan Xiao,Qinghe Du,Chaoyang Zhang,Wenchi Cheng,Jingqing Wang,Tao Chen,Jiangzhou Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:The rapid advancement of foundation models and edge intelligence is transforming wireless networks from data-oriented communication systems into artificial intelligence (AI)-native service platforms. Token communications (TokenCom) has emerged as a promising paradigm that directly transmits task-oriented AI representations. However, existing studies typically assume that generated tokens can be transmitted immediately, overlooking the queueing dynamics arising from the mismatch between token workload generation and time-varying wireless service capability. To address this limitation, we introduce Queue-Aware TokenCom, which explicitly incorporates token queues into the AI service pipeline. We present an end-to-end architecture integrating AI representation generation, token workload management, wireless transmission, and downstream task inference. We then discuss two key design directions: budget-aware token representation optimization and queue-aware token service optimization, with the token budget serving as a critical interface between AI performance and communication efficiency. Moreover, we investigate fixed and adaptive token budget strategies to characterize the fundamental tradeoff between AI performance and communication efficiency in dynamic wireless environments. Finally, we outline research directions toward closed-loop, task-aware, and scalable TokenCom systems.
[AI-233] race-to-Architecture: Capability Bottleneck Diagnosis for Harness Revision
链接: https://arxiv.org/abs/2610.16725
作者: Wenfeng Wang,Jianrui Shi,Mingang Guo,Jiayuan Ma,Yuqi Lu,Jiajun Sun,Guojiang Zhao,Jing Li
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Improving an LLM agent requires knowing what to change when it fails. Similar failures can require different fixes, and resolving one problem can expose another. We introduce Trace-to-Architecture (T2A), a framework that uses execution trajectories to determine what to change in an agent’s harness and when those changes apply. A development agent diagnoses failures using execution evidence from successful and failed tasks, revises the corresponding harness components, and then checks whether target tasks improve, selected successful tasks remain successful, and the edits are actually triggered. New execution evidence guides subsequent diagnosis and revision. Across ALFWorld, ScienceWorld, and AppWorld, T2A achieves the highest aggregate score for all three evaluated task agents, exceeding their respective strongest baselines by 6.2–30.3 points. We also integrate T2A’s diagnosis and validation procedures into Meta-Harness, improving the resulting harnesses’ mean ScienceWorld score by 21.6 points from the same starting harness.
[AI-234] Feature-Context Consistency: Unsupervised Adversarial Detection with Drift Stability on Attributed Graphs
链接: https://arxiv.org/abs/2610.16694
作者: Firas Bayram,Maxime Cordy
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Adversarial detection on evolving attributed graphs faces two challenges: fraud labels arrive late or not at all, and natural distributional drift erodes any trained decision boundary. Modern detectors can reach strong detection rates when labels are available and the distribution is stationary; sustaining that quality without labels and under drift is the open problem. We identify feature-context consistency – the alignment between a node’s attributes and its aggregated neighborhood attributes – as a single structural signal that addresses both. We formalize it as the Context Boost (CB) score within a Restricted Boltzmann Machine framework (CB-RBM) and prove two complementary worst-case guarantees: a strict separation between the expected CB of legitimate and adversarial nodes, with gap growing linearly in model capacity, whenever the adversary’s context stays far enough from legitimate neighborhood context; and a Wasserstein-Lipschitz bound on how far the CB distribution moves under natural drift. We validate on XBlock Ethereum phishing (~153K nodes) and Reddit banned-user detection (~11K nodes), with two further datasets in the appendix spanning diverse domains. With zero labels, CB-RBM achieves AUROC = 0.96 on four adversarial attacks and keeps FPR close to its calibrated 5% target under drift – outperforming supervised GNNs (hundreds of labels), unsupervised graph-anomaly baselines, and a standard RBM whose FPR collapses to 89-100%. What typically requires separate mechanisms – label-free operation and drift stability – here follows from a single structural regularity of the data. Code: this https URL.
[AI-235] Structural Baselines Matter for Agent Failure Detection and Attribution
链接: https://arxiv.org/abs/2610.16661
作者: Ramesh Arvind Naagarajan,Stefan Streif
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:An LLM agent using tools can fail silently. It stops before the task is done, raises no error, and answers fluently. Two kinds of monitor address this, one deciding whether a run failed and the other naming the step where failure became unavoidable. Much of the reported performance on both tasks comes from regularities that need no model. Failing runs are longer, and the step labelled decisive is usually the last. A rule that always names the last step is correct 67% to 96% of the time here, without reading the run at all. That ceiling reflects how the labels were written, not how agents break. On failures whose decisive step replay can confirm, it falls to near chance. Reporting gain over that rule therefore changes what a score means. Under that correction, methods tested here gain nothing on give-up failures, while several published systems keep a clear advantage on other splits. For detection, the agent’s internal states beat transcript length and text similarity on the agent benchmark, and hold up where the text baseline does not. Precision stays low when failures are rare, so such a monitor should rank runs for review, not replace it.
[AI-236] Learning Incompressible 3D Flows from Sparse Samples: A Set Transformer Approach to Divergence-Free Kernel Fitting
链接: https://arxiv.org/abs/2610.16645
作者: Ramzi Idoughi,Radouan Boukharfane,Peter Wonka
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE)
备注: 33 pages with supplement materials, 9 figures
Abstract:Recovering divergence-free velocity fields from sparse and corrupted observations requires reconciling physical consistency, expressive representations, and robustness to missing data. We introduce DFK-Former, a set-to-field transformer that reconstructs divergence-free velocity fields from sparse, noisy, and irregularly sampled 3D observations. The model is inherently invariant to the number and ordering of input samples and predicts the parameters of a divergence-free kernel representation, producing continuous, mesh-free velocity fields that satisfy incompressibility exactly by construction rather than approximately through soft constraints or projection operators. Unlike existing divergence-free kernel approaches, which estimate kernel coefficients via a separate optimization for each field, DFK-Former predicts these coefficients directly in a single forward pass. This eliminates costly iterative fitting while providing a learned prior that enables accurate reconstruction even when observations are insufficient to uniquely determine the field. To enhance representational power, we introduce adaptive and anisotropic divergence-free kernels that preserve exact incompressibility guarantees while capturing more complex flow structures. Extensive experiments across varying levels of sparsity, noise, and occlusion, including transfer to real Direct Numerical Simulation (DNS) turbulence data, demonstrate substantial gains in both accuracy and efficiency. DFK-Former reduces endpoint error by up to 4.6x compared to per-field divergence-free-kernel optimization, while reconstructing each field in 7.5 ms instead of 9.8 s, yielding a speedup of more than three orders of magnitude. These results establish direct, feed-forward prediction of divergence-free kernel parameters as a practical and effective paradigm for physically consistent velocity-field reconstruction from sparse observations.
[AI-237] AeroChem: Closed-loop Physics-Informed State Space Modeling for Long-term Chemically-Reactive Air Quality Forecasting–Extended Version NEURIPS2026
链接: https://arxiv.org/abs/2610.16630
作者: Kien Le Trung,Tung Kieu,Dinh Duc Nha Nguyen,Cuong Do,Bin Yang,Phi Long Nguyen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE)
备注: 25 pages. An extended version of “AeroChem: Closed-loop Physics-Informed State Space Modeling for Long-term Chemically-Reactive Air Quality Forecasting–Extended Version” accepted at NeurIPS 2026
Abstract:Accurate long-term air quality forecasting is challenging due to long-range transport, meteorological variability, and nonlinear chemical reactions underlying secondary pollutant formation. Purely data-driven models often suffer from long-horizon drift and lack physical consistency, while existing physics-informed approaches commonly follow open-loop designs that fuse neural and physical components only at the output level. We propose AeroChem, a closed-loop physics-informed state space framework for chemically reactive air quality forecasting. AeroChem adopts a two-branch co-correction design: a UDE-based process branch provides a structured but imperfect physical prior, while a Mamba-based virtual measurement branch captures long-range temporal context and produces probabilistic virtual observations. Neural Kalman Fusion couples the two branches by correcting the latent physical state at each forecasting step and feeding the posterior state back into the simulator, reducing long-horizon drift. We further introduce HanoiAir, a real-world dataset integrating pollutant concentrations, meteorology, and emission inventories. Experiments on HanoiAir and a public benchmark show that AeroChem improves long-horizon stability and sudden-change forecasting.
[AI-238] LATCH: LLM -Assisted Timing Closure Harness for Post-Route FPGA Optimisation ATC
链接: https://arxiv.org/abs/2610.16625
作者: Shu Gu,Matthew Youngman,Yuqin Zhao,Cristian Sestito,Ali Siddique,Ramesh Krishnamurthy,Danial Chitnis,Themis Prodromakis
类目: Artificial Intelligence (cs.AI)
备注: 14 pages, 11 figures, 4 tables. Code: this https URL
Abstract:Post-route FPGA timing closure remains expert-dependent and computationally expensive. We present LATCH (LLM-Assisted Timing Closure Harness), a two-layer framework that separates large language model (LLM) orchestration from deterministic FPGA implementation and evaluation. Across sixteen designs and seven LLMs, a single-iteration study compares three agent control granularities - expert-written recipes, direct EDA operations and LLM-constructed procedures - to ask how much control an agent needs per attempt. Functional-pass rates are 98.2%, 71.4% and 17.9%, respectively; yet direct operations give slightly larger 3.4% median Fmax gains when passing, versus 3.2% for recipes. This exposes a trade-off between execution reliability and timing improvement. Informed by this trade-off, a multi-iteration study fixes the action tier at expert recipes to ask how repeated attempts are best organised, comparing a fitted deterministic heuristic with sequential LLM reasoning, branching exploration and knowledge base-guided exploration within 12 checkpoints per run. Branching raises mean maximum-frequency improvement from 10.5% (sequential) to 16.0%, and the knowledge base to 20.3%, improving 15 of 16 designs. Characterisation across 2,421 recipe attempts profiled their gain and validation rates, showing hard-block re-placement gives the largest cumulative mean gain at 41.4MHz.
[AI-239] Weak Segmentation-Guided GAN for~Realistic Color Edition
链接: https://arxiv.org/abs/2610.16620
作者: Vincent Auriau,Emmanuel Malherbe,Matthieu Perrot
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Editing the color of images in a realistic way finds many applications such as changing the perception of an image, data augmentation or film post processing. The design of an automatic tool is a complex and long addressed challenge. In particular, two properties are difficult to meet altogether: generating realistic results without artifacts and the possibility to precisely choose the future color. Conventional methods using segmentation and histogram matching maximize the controllability but also introduce a lack of realism or are complex to automate. On the contrary, GANs that specialize in realism are difficult to control. To overcome these challenges, we propose a novel GAN architecture leveraging any differentiable segmentation model. We demonstrate the genericness of our framework that presents state of the art results on different use cases. It generates images that look realistic while offering a precise color control.
[AI-240] When 11 Backfires: Rethinking Relevance-Diversity Balancing for Sample-Aware Visual Token Reduction
链接: https://arxiv.org/abs/2610.16615
作者: Yihang Huang,Peisong Wen,Chenhao Zhou,Difu Feng,Qianqian Xu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Visual token compression has emerged as a critical technique for reducing the inference cost of Vision-Language Models. Existing methods select tokens based on text relevance, visual diversity, or a fixed combination of both—treating the relevance-diversity balance as a universal constant independent of the input. We conduct a systematic analysis to challenge this assumption. At the dataset level, grid-searching the optimal relevance-diversity weight per benchmark reveals substantial cross-task variation. More critically, we further maximize per-sample answer likelihood to recover sample-level oracle weights, finding that their distributions not only differ markedly across benchmarks but also exhibit high within-benchmark variance. Based on these observations, we propose SAVER, a framework for Sample-Aware relevance-diversity balanced Visual tokEn Reduction. We first derive a Geometric Intent Prior (GIP) from the spatial concentration of text-relevant tokens to infer a sample-specific relevance-diversity weight, and then apply this weight to jointly select tokens via a Weighted Text Centroid (WTC) for relevance scoring and a Visual Marginal Coverage Gain (VMCG) for diversity. SAVER requires no additional training and achieves sample-aware balancing entirely at inference time. Extensive experiments demonstrate that SAVER consistently outperforms state-of-the-art baselines across diverse benchmarks, retaining 96.5% of full-model performance on LLaVA-1.5-7B with only 11.1% of the original visual tokens, and further generalizing to LLaVA-NeXT and Mini-Gemini.
[AI-241] owards Modality-Agnostic Seizure Classification: A Unified Deep Learning Framework for Heterogeneous EEG and Intracranial EEG Signals
链接: https://arxiv.org/abs/2610.16596
作者: Ramya Sundaram,Muhammad Aslam Jarwar,M Uma,Sunder Ali Khowaja,Qurat-Ul-Ain Mastoi
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Seizure prediction and detection, essential for clinical treatment of epilepsy, remain the crucial task of this neurological disease. EEG and iEEG are widely used to analyze epilepsy due to its capability of capturing discriminative brain activities at high time resolution. However, existing works generally analyze either EEG or iEEG seizure activity separately, where EEG analysis has been well-researched, while more and more works consider iEEG for its higher specificity at the spatial level and signal quality. Currently, most of the existing seizure prediction systems are single-modal in nature, which limits its application to a variety of clinical conditions. Therefore, a common framework to analyze seizures with both EEG and iEEG signals needs to be developed. This paper presents a new framework of multimodal seizure classification integrating the heterogeneous EEG and iEEG recordings from different subjects and sources. A specific framework, called Unified Multimodal Signal Harmonization (UMSH), is developed for harmonizing signals with varied sampling rates and recording conditions. Moreover, to handle the extremely class-imbalanced problem in the data sets, we use a Conditional Generative Adversarial Network (cGAN) based data augmentation technique to generate synthetic samples to balance the class distribution. The proposed framework uses the EEGNet-Vision Transformer (EEGNet-ViT) as the feature extraction and classification network to learn the local discriminative patterns as well as global contextual information in the multimodal seizure data. Experimental results demonstrate the effectiveness of the proposed framework for robust seizure classification, emphasizing the potential of integrated EEG-iEEG multimodal learning in advancing next-generation epilepsy monitoring and clinical decision-support systems.
[AI-242] Double-Rollout Reinforcement Learning with Sampling-based Reward Estimation
链接: https://arxiv.org/abs/2610.16593
作者: Haifeng Qian,David Mosallanezhad,Christian Munley,Adithya Renduchintala,Oleksii Kuchaiev
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Reinforcement learning with verifiable rewards (RLVR) implemented with Group Relative Policy Optimization (GRPO) has driven substantial progress in training reasoning language models. However, verifiable outcome rewards provide an imperfect measure of reasoning quality: a correct final answer may not be supported by the reasoning trace, whereas an incorrect rollout may contain a nearly correct reasoning trajectory. We present empirical evidence for this mismatch and hypothesize that the resulting reward noise contributes to instability in reinforcement learning for both reasoning accuracy and token efficiency. To address this credit assignment challenge, this paper proposes double-rollout reinforcement learning. For each sampled rollout, we truncate its reasoning trace, sample multiple continuations from the truncated prefix, and replace the original reward with the mean continuation reward, yielding a more robust estimate of reasoning quality. This double-rollout algorithm incurs only a moderate increase in compute cost and yet reaches substantially higher accuracies across math and coding tasks and across Qwen and Nemotron models than baseline algorithms with fewer RL steps. For example, using only DAPO data it trains Nemotron-3-Nano-30B-A3B to reach 95.4% on AIME 2026, versus 91.5% by GRPO and substantially above state of the art for this model size, within 30 RL steps. When combined with length-based rewards, the same algorithm further improves the accuracy-efficiency Pareto frontier over existing efficiency optimizations. These findings suggest that the double-rollout reward estimation is an effective and cost-efficient way to improve both the accuracy and token efficiency of reasoning RL.
[AI-243] G2WAM: World Action Models for Efficient and Geometry Grounded Image-Goal Navigation
链接: https://arxiv.org/abs/2610.16558
作者: Ke Gui,Chenghao Shi,Ziyue Wang,Longwen Chen,Xieyuanli Chen,Huimin Lu
类目: Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: 20 pages, including references and appendix
Abstract:World models capture environment dynamics and support counterfactual reasoning for planning. In image-goal navigation, where the target often lies beyond the current field of view, the agent must not only model environmental dynamics but also maintain a persistent spatial memory of the scene. Existing approaches either introduce spatial context by fine-tuning pretrained models on the target task, which underexploits their spatial priors, or rely on external 3D reconstruction, which is computationally expensive for real-world deployment. In this paper, we propose the Geometry Grounded World Action Model (G ^2 WAM) for image-goal navigation. G ^2 WAM encodes spatial cues through a learned scene-coordinate localization module that anchors the current, goal, and history frames in a common reference frame, yielding a stable and consistent spatial context that enables efficient navigation without explicit 3D reconstruction. Following the world-action-model paradigm, G ^2 WAM jointly predicts future states and actions, eliminating the need for costly cross-entropy method (CEM) style planning. In addition, G ^2 WAM operates entirely in latent space, avoiding the computational burden of pixel-level reconstruction and substantially reducing inference cost. Across four trajectory prediction benchmarks, G ^2 WAM reduces average ATE and RPE by 40.1% and 35.2%, respectively. In closed-loop navigation experiments, it improves the success rate from 54.0% to 63.0% and the success weighted by path length (SPL) from 36.4% to 59.2%, while maintaining high efficiency.
[AI-244] AI Discovery in Combinatorics: From Problems to Solutions and Back to Bigger Problems
链接: https://arxiv.org/abs/2610.16555
作者: Eugenio Cainelli,Lorenzo Luccioli,Alessandro Iraci,Giovanni Paolini
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Combinatorics (math.CO)
备注: 15 pages
Abstract:We introduce qt Bench, a benchmark of 29 open problems in algebraic combinatorics. These problems ask for a simple combinatorial algorithm that explains given polynomial coefficients or identities. To help AI agents search for such algorithms, we provide a scaffold for playing with combinatorial objects, countering cheating attempts, and testing candidate solutions. To explore how far frontier AI systems can go on these problems, we collaborated with OpenAI, whose researchers obtained proposed solutions to several of them. This highlights bigger meta-problems: is AI discovery at scale ethical? How should these results be released? How should proof digestion be organized? We discuss these questions and give our views.
[AI-245] Making Language the Low-Rate Code of Vision: Successive Refinement from Words to Pixels
链接: https://arxiv.org/abs/2610.16552
作者: Ruilin Gong,Hao Gong
类目: Artificial Intelligence (cs.AI)
备注: 21 pages, 3 main figures, 5 main tables, with supplementary material
Abstract:An ordered visual code lets a short prefix stand in for an image, and the training loss decides what that prefix contains. Under pixel reconstruction the first bits go to the directions of highest variance, and we prove that a caption can then be independent of the entire first half of a code that is rate-optimal for pixels at every length. In the rate-distortion sense a caption is therefore not the low-rate code of its image, unless the code is built to make it one. We study that construction and what it costs. We cast captioning and text-to-image generation as successive refinement under a distortion measure that changes along the code, from caption log-loss in the early stages to pixel distortion in the last, so that each stage is compared with its own optimum. Putting words first then costs nothing among the caption stages exactly when the captions form a summarization chain, and it costs the pixels at most the caption uncertainty left in a pixel-optimal reconstruction, which is large for short prefixes and small at full length. We build RUNG, an ordered code whose prefixes are supervised by a summarization ladder of captions and whose full length reconstructs the image; generation extends the code past the prompt and captioning truncates it. With data and capacity held fixed, ordering by language instead of reconstruction raises prefix-caption information 2.9 times and GenEval from 0.76 to 0.83, at a measured cost of 2.8 dB PSNR at a quarter of the code and 0.1 dB at full length. RUNG-XL (3.4B) reaches 0.88 GenEval and 87.6 DPG-Bench in 32 network evaluations and 0.91 GenEval when errors are redrawn at the rung where they appear; its coarse and fine captions agree 94.1% of the time, and it edits images without inversion.
[AI-246] Overtaught Underlearned: Learner-Centered Distillation from Multi-Agent Discussion
链接: https://arxiv.org/abs/2610.16543
作者: Mengtao Zhou,Jianxiao Wang,Xiupeng Chen,Cailiang Liu,Jie Song
类目: Artificial Intelligence (cs.AI)
备注: 28 pages
Abstract:Multi-agent discussion produces reasoning collaboratively, while deployment still relies on a smaller student that must answer alone. Full-graph distillation transfers such discussions by treating each correct solution as a separate training target, but what teachers provide need not match what the student learns. This mismatch appears in two ways: several correct solutions may amount to the same lesson in the student’s representation (over-teaching), while some training questions remain missed after learning (under-learning). We quantify both with rates that are high on discussion graphs. To address them, we introduce FAR (Fold–Attempt–Revisit), a learner-centered distillation procedure that organizes supervision around the student’s own state. Before training, FAR folds solutions within each discussion state into representative lessons. After training, the student’s own attempts identify missed questions, while poorly represented clusters indicate where folding may have lost useful variation; FAR then writes targeted lessons from retained discussion contexts. Across two students and four reasoning tasks, FAR outperforms full-graph distillation with far fewer training targets and lower under-learning. Further analyses examine the behavior of FAR and the sources of its gains. Overall, the results show that organizing discussion supervision around the student can make distillation both more effective and more efficient.
[AI-247] Does the Explanation Drift Before the Guarantee? Label-Free Early Warning on Evolving Graphs
链接: https://arxiv.org/abs/2610.16534
作者: Tomasz Szandała,Damian Raczkowski
类目: Artificial Intelligence (cs.AI)
备注: 9 pages + appendix for reproducibility
Abstract:Deployed models can fail silently as the world they describe evolves. Attribution drift has served as an early warning in tabular data; we ask whether, in graph streams, a model’s own explanations can forewarn the loss of nominal conformal coverage. Doing so requires repairing the naive drift statistic twice. A vanishing edge loses mask mass regardless of whether the model relied on it, so we condition the importance on edge presence. Evolving graphs also break edge identity: a relabelling that changes nothing shifts the identity-based statistic by 0.69 on synthetic graphs and 0.83 on real ego-networks, whereas aggregating over structural roles brings both to 0.0 and stays well-defined when the support changes. We test six pre-registered regimes on thirty seeds each. A two-channel monitor, tracking conditional mask mass and its distribution over roles, precedes coverage collapse by 8 to 14.5 steps across four gradual regimes (node addition, removal, replacement, and edge thinning), all surviving Holm correction. A sudden knockout yields the expected zero lead, and a harmless change triggers no alarm. Leads track the length of the transitional phase (Spearman 0.79 over 150 runs). On real transaction ego-networks, the untuned mass channel forewarns staged drift by 27 steps and is the only channel, attributional or structural, that never alarms on a stationary stream. The main limitation is that the mass channel is blind to covariate drift, which breaks the guarantee in every run; the distributional channel detects it and also rescues the rewiring regime, so we report the two channels separately rather than as a union.
[AI-248] Boosting the Learning for Ranking Patterns
链接: https://arxiv.org/abs/2610.16511
作者: Nassim Belmecheri,Noureddine Aribi,Nadjib Lazaar,Yahia Lebbah,Samir Loudni
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Pattern mining is a valuable tool for exploratory data analysis, but identifying relevant patterns for a specific user is challenging. Various interestingness mea- sures have been developed to evaluate patterns, but they may not efficiently estimate user-specific functions. Learning user-specific functions by ranking patterns has been proposed, but this requires significant time and training samples. In this paper, we present a solution that formulates the problem of learning pattern ranking functions as a multicriteria decision making problem. Our approach uses Analytic Hierarchy Process (AHP) to elicit weights for different interestingness measures based on user preference. We also propose an active learning mode with a sensitivity-based heuristic to minimize user ranking queries while still providing high-quality results. Experiments show that our approach significantly reduces running time and returns precise pattern ranking while being robust to user error, compared to state-of-the-art approaches.
[AI-249] pi-Net: Empirical Latent Representations for Differentiable Physics Heads in Data-Scarce Systems
链接: https://arxiv.org/abs/2610.16504
作者: Akshay Dhalpe,Arnab Chattopadhyay,Shahriar Bakrani Balani,Eric Cortanéa,Sasu Tarkoma
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Developing reliable predictive models for multi-physics systems often forces a compromise between computationally demanding physical simulators and data-hungry, opaque machine learning (ML) algorithms. While incorporating dimensional analysis into directed computation graphs offers a structurally sound alternative to explicitly solving partial differential equations (PDEs), such formulations typically yield underdetermined systems that act as imperfect simulators. We introduce pi-Net, a physics-informed latent variable model that resolves these imperfect simulators. The hybrid pi-Net architecture consists of an encoder that maps inputs into physically normalized latent space, which are then decoded into final predictions by a differentiable physics head. Evaluated on a real-world, data-scarce Gas Metal Arc Welding (GMAW) case study, the framework achieves an R2 of 0.864 with only 60 experimental samples. The results demonstrate that pi-Net significantly outperforms purely data-driven baselines in data-scarce regimes while providing causal, process-level explainability.
[AI-250] From Caricature to Collective Artifact: How Persona Construction Drives Polarization in Generative Social Simulations
链接: https://arxiv.org/abs/2610.16498
作者: Dimitris Michailidis,Petter Törnberg,Fernando P. Santos
类目: ocial and Information Networks (cs.SI); Artificial Intelligence (cs.AI)
备注: 17 pages
Abstract:Generative Social Simulation (GSS) leverages Large Language Models (LLMs) to create sophisticated personas that define agents with demographics, emotions, and political opinions. While prior work shows that personas compress outputs into stereotypical caricatures, there is no comprehensive study of how these caricatures evolve into collective artifacts when they interact. Using a Twitter-like GSS platform with personas derived from the American National Elections Study (ANES), we ablate three layers of partisan identity: affective polarization markers, party identification, and vote choice, while retaining demographics, ideological positioning and important issues. We find that collective outcomes depend heavily on how survey responses are translated into personas. Removing a few explicitly partisan statements moves the network from near-complete segregation to cross-party ties approaching those of real Twitter networks. We argue that matching one outcome is not evidence for behavioral validity and point to salience as a possible explanation. Surveys ask for a position on every question regardless of its importance to the respondent, and persona construction turns each answer into a direct behavioral instruction. We conclude with a checklist for more comprehensive evaluation of personas in future works.
[AI-251] Drafting Agents Not Just Tokens: Speculative Recursive Self-Improvement for Agent ic Systems
链接: https://arxiv.org/abs/2610.16491
作者: Xin Xu,Ke Cheng,Hao Wang,Yangru Huang,Yanghao Xiao,Zhichao Chen,Zhisheng Zhong,Xun Liu,Yi Li,Qian Li,Haibin Lv,Jun Zhang,Jie Jiang,Zhouchen Lin
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 47 pages; code available at this https URL
Abstract:Improving an agent through successive revisions requires balancing exploration and evaluation under a limited budget. A narrow search can miss useful sequences of edits, while fully evaluating every intermediate version can exhaust the budget before those sequences yield a verified improvement. We introduce Speculative Recursive Self-Improvement (SpecRSI), which drafts executable successor agents before their predecessors are fully evaluated, while requiring complete verification before adoption. Verification-Aware Tree Search (VATS) allocates the budget between further edits and outstanding checks using the predicted verified return of a continuation. Group Verification Optimization (GVO) learns the improver from completed searches while keeping the real evaluator fixed. In the main scientific comparison, terminal gains exceed Janus by 33.3%–40.0%. Additional evaluations show that SpecRSI outperforms Janus in all 50 scientific backbone–task pairs and exceeds MetaRSI-v1 by 1.85%–2.38% in macro-average system score across twenty backbones. Coding-validation success reaches 69.4% versus HGM’s 64.3% at a 20.8% lower target-cost index.
[AI-252] Growing and Transferring Pathways for LLM Reinforcement Learning
链接: https://arxiv.org/abs/2610.16489
作者: Haolin Tian,Yikang Li,Tonghan Wang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Reinforcement learning (RL) can substantially improve the reasoning capabilities of large language models, but an effective training run requires considerable resources. Reusing the computation acquired through RL across models would extend the value of that training. We propose GrowRL, an RL method that organizes learned updates into compact computational pathways through structural growth. GrowRL freezes the pretrained backbone and adds an independent causal attention branch and a feed-forward network branch at each Transformer layer. Zero-initialized output projections preserve the pretrained policy at initialization, and RL updates only the added pathways. The learned pathways remain explicit after training, allowing transfer to structurally compatible target models and subsequent adaptation with the target backbone frozen. The architecture also supports sequential transfer of additional pathways. By keeping the learned attention and feed-forward computation separate from the pretrained weights, GrowRL provides an explicit structure for preserving and reusing the outcome of an RL run. Experiments across five backbones show competitive mathematical reasoning and out-of-distribution performance with a small number of trainable parameters. Learned pathways provide immediate gains across all three transfer directions, with performance after only 10 target-side updates already close to that after 100 updates in every case. These results support computational pathways as parameter-efficient and reusable units of reinforcement learning adaptation.
[AI-253] SLEM: Spectrum-Aware Latent Expert Merging
链接: https://arxiv.org/abs/2610.16484
作者: Dohyeon Kim,Bedionita Soro,Bruno Andreis,Sung Ju Hwang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 21 pages, 8 figures, 7 tables
Abstract:Mixture-of-Experts (MoE) models scale model capacity efficiently by activating only a small subset of experts for each input, while still requiring all parameters to be stored in memory, resulting in substantial memory overhead. Expert merging techniques have recently emerged as a promising compression paradigm that reduces this overhead by consolidating multiple experts into a smaller set. However, existing approaches have two key limitations: (i) direct convex interpolation in parameter space is not generally guaranteed to yield meaningful interpolation of expert functions; and (ii) estimating expert similarity or merging coefficients often relies on calibration data, introducing dependence on the calibration distribution. To address these limitations, we propose Spectrum-Aware Latent Expert Merging (SLEM), a calibration-free expert merging framework that occurs in a structured latent representation of expert weights using a VAE. SLEM groups experts within and across layers based on latent similarities and derives merging coefficients from expert weight spectra. It then performs geometry-aware latent interpolation and decodes the resulting representations into merged expert weights. Compared with existing expert merging methods, SLEM achieves comparable or better performance on general knowledge benchmarks and higher average performance on generation benchmarks involving mathematical reasoning and code generation.
[AI-254] PXRD-Bench: How Close Are We to Determining Crystal Structures from Powder Diffraction?
链接: https://arxiv.org/abs/2610.16480
作者: Dongyeop Woo,Nofit Segal,Nayoung Kim,Minkyu Kim,Rafael Gomez-Bombarelli,Sungsoo Ahn
类目: Artificial Intelligence (cs.AI); Materials Science (cond-mat.mtrl-sci)
备注:
Abstract:Determining crystal structures from powder X-ray diffraction (PXRD) would enable routine characterization of newly synthesized materials, including those absent from reference databases. Generative models offer a promising approach, but existing evaluations remain fragmented and largely assume single-phase samples with known composition. To fill this gap, we introduce PXRD-Bench, a unified benchmark that extends evaluation to unknown per-element atom counts and recovery of both structures in two-phase mixtures. It provides approximately 4.1 million training structures with simulated PXRD, tests generalization to unseen compositions and structural prototypes, and evaluates recovery from experimental PXRD after excluding structurally matching training entries. To provide a strong reference baseline, we also introduce Bragglite, a PXRD-to-structure model developed through a systematic study of PXRD modeling choices. It achieves a 44.2% Top-1 match rate on experimental RRUFF patterns with known composition, 24.6% above the strongest prior baseline. Further analysis yields several key findings: some models fail to benefit from experimental PXRD, while generalization to unseen structural prototypes and recovery from two-phase mixtures remain challenging. Together, our benchmark, systematic studies, and Bragglite clarify the capabilities and limitations of current PXRD-to-structure models.
[AI-255] StructZO: Structured Zeroth-Order Optimization for Memory-Efficient LLM Fine-Tuning
链接: https://arxiv.org/abs/2610.16451
作者: Xuanlin Yang,Jia Li,Zhouchen Lin
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 20 pages
Abstract:Fine-tuning large language models at scale is increasingly limited by the memory cost of backpropagation. Zeroth-Order (ZO) optimizers reduce this burden by relying only on forward function evaluations, but they still exhibit a substantial performance gap to first-order (FO) optimizers such as Adam. We argue that this gap does not arise solely from the absence of explicit gradients: FO fine-tuning also benefits from informative low-rank optimization geometry, large-scale signal aggregation, and well-conditioned update directions. Motivated by this, we introduce StructZO, a structured Zeroth-Order optimization framework for memory-efficient LLM fine-tuning. StructZO performs large-query ZO estimation within an activation-induced informative subspace and applies spectral control to the resulting reduced-space update before mapping it back to the original parameters. This lets ZO aggregate useful directional signals in a compact, geometry-aware space while avoiding the activation and optimizer-state memory costs of backpropagation. We show theoretically that, when the induced subspace captures concentrated optimization-relevant directions, StructZO reduces the effective ZO estimation dimension from mn to mr , with the remaining projection error governed by the spectral tail of the activation Gram matrix. Empirically, across OPT-series and Llama3.1-8B on diverse downstreamtasks, StructZO improves over the strongest existing ZO baselines by up to 8.7% relative on average and up to 26.9% relative on individual tasks, while remaining within 1.5% of FO-Adam on average. These results suggest that structured and spectrally controlled ZO updates can substantially narrow the gap between ZO and FO optimization.
[AI-256] CryptoFlowBench: Benchmarking Multi-Hop Reasoning on Blockchains for Cryptocurrency Fund Tracing
链接: https://arxiv.org/abs/2610.16449
作者: Runang He,Tongya Zheng,Jing Cao,Yucheng Ying,Shunyu Liu,Bingde Hu,Yang Gao,Zeyuan Hu,Jianrong Zhang,Mingli Song
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 9 pages, 2 figures. Currently under review
Abstract:Pseudonymous accounts and complex smart-contract ecosystems substantially hinder the tracing of illicit cryptocurrency fund flows. Existing Graph Neural Networks (GNNs) have achieved promising performance in blockchain fraud detection, yet their static node- or edge-level predictions do not naturally generalize to multi-hop fund tracing. Meanwhile, existing tracing methods mainly rely on predefined heuristics to capture multi-hop relationships, limiting their adaptive reasoning capabilities. To bridge this gap, we introduce CryptoFlowBench, a benchmark constructed from authoritative reports on cryptocurrency-related illicit incidents, with four tasks spanning local to multi-hop fund tracing. These tasks pose substantial challenges to existing learning- and heuristic-based methods, particularly in complex multi-hop scenarios. We therefore propose ChainSleuth, a graph-tool-augmented agent that iteratively reasons, invokes graph tools, and updates its tracing trajectory, enabling adaptive information acquisition and recovery from erroneous paths. Experiments show that no baseline consistently dominates all four tasks, revealing substantial gaps between local fraud detection and multi-hop fund tracing. ChainSleuth achieves notable improvements on multi-hop tasks, demonstrating the potential of agentic graph reasoning for cryptocurrency fund investigation.
[AI-257] RAP-WAM: Learning World-Action Models through Predictive Rehearsal and Action Perturbations
链接: https://arxiv.org/abs/2610.16443
作者: Yuanpeng He,Ceyao Zhang,Fangjing Li,Kefei Zhu,Lijian Li,Yueheng Li,Tianjia He,Wenpin Jiao,Zhi Jin,Yaodong Yang,Yuanpei Chen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Recent world action models draw on video generation priors to jointly predict future observations and robot actions. Adapting these priors to control requires determining how future predictions should guide policy learning. RAP-WAM combines predictive Rehearsal and Action Perturbations (RAP) with future-supervised queries and Coupled Action Projection for policy learning. Predictive rehearsal trains on model-generated inputs while retaining recorded targets, and accumulates recurring errors into a spatial prior. We perturb actions and measure changes in video predictions. Action-response-based spatiotemporal credit assignment combines these responses, prediction error, and the rehearsal prior to allocate visual supervision; spatially aggregating the responses determines supervision intensity across future action steps. No separate scoring network or deployment-time probe is required. The base learns from a video-pretrained backbone without intermediate pretraining on external embodied data. We evaluate the fixed base and its agent-refined, task-conditioned system. On 42 RoboDojo tasks, their average success rates are 12.85% and 15.62%, with Scores of 17.86 and 20.59. Across 15 real-robot tasks, the complete system achieves 89.00% task-macro-average success, exceeding the strongest of five baselines by 37.67 percentage points. Across 315 fixed-output simulation windows, 45.21% of response-top20 positions fall outside error-top20, revealing information beyond prediction error.
[AI-258] AuthGateBench: Benchmarking LLM s as Authorization Gate
链接: https://arxiv.org/abs/2610.16426
作者: Zhixiong Zhuang,Jin Jia,Zhiwei Shang,Ziyu Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Automated approval systems let coding agents execute high-risk actions with minimal human oversight by delegating the authorization decision to a classifier, typically a large language model (LLM). Such systems, adopted in tools like Claude Code and Codex, remain vulnerable to failure modes such as scope escalation, where the agent takes actions exceeding what the user authorized. Existing benchmarks focus mainly on malicious requests, evaluating refusal under an implicit policy. We introduce AuthGateBench, the first policy-guided benchmark for authorization gating under benign requests. We implement a novel pipeline where an LLM generates contrast pairs with identical action and trajectory but differing request, flipping the label between block and allow. This allows us to assess whether models reason over context rather than the action alone. We further examine whether auxiliary environment signals can help models make better decisions. The resulting benchmark comprises 1,400 instances spanning five failure modes drawn from real-world Claude Code incidents. Evaluations show accuracy degrades unevenly across failure modes as models over-rely on internal safety alignment, and models still block most actions even when explicitly authorized, revealing unstable rather than deterministic decisions; retrospectively retrieved environment signals, however, substantially improve authorization accuracy.
[AI-259] Chain Dimension Scaling: Communication-Efficient Loop-Style Scaling for Industrial Short-Video Ranking
链接: https://arxiv.org/abs/2610.16415
作者: Zhimeng Jiang,Yuening Li,Ruixiao Sun,Yuanzhen Lin,Kun Jin,Chuan He,Liang Liu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Industrial recommendation rankers keep scaling in width and depth to billions of parameters, but this increasingly incurs costs in parameter growth, distributed communication overhead, and training inefficiency. We introduce \underlineChain \underlineDimension \underlineScaling (CDS), a scaling strategy orthogonal to width and depth. CDS repeatedly applies a single feature-interaction block with parameters shared across iterations, so the number of iterations (the chain length) becomes a new scaling dimension that adds no parameters. A lightweight intra-chain residual connection stabilizes optimization and improves quality. CDS enables two regimes. In the \emphcompression regime, shallow blocks with a longer chain match deeper baselines with fewer parameters and faster training; in the \emphquality-scaling regime, increasing chain length under a fixed parameter budget yields consistent offline and online gains, suggesting the chain dimension adds modeling capacity beyond parameter reuse. We evaluate CDS in a large-scale multi-task short-video ranking system in production. In the compression regime, CDS stays within 0.03% of baseline offline quality while reducing trainable parameters by 26.6% and improving training throughput by 6.4%, with neutral-to-positive online participation. In the quality-scaling regime, quality improves monotonically with chain length. Wall-clock analysis shows the efficiency gain comes mainly not from reduced FLOPs but from lower cross-device synchronization overhead and reduced optimizer update cost. These results position parameter-shared iterative computation as a practical, communication-efficient scaling axis for industrial ranking systems.
[AI-260] SLS-Bench: A Benchmark for Incident Log Summarization with Synthetic Observability Data
链接: https://arxiv.org/abs/2610.16412
作者: Valerii Iakovlev,Markus Spanring,Martin Flechl
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:As IT systems grow in complexity, automated agent-driven incident response becomes increasingly critical. A key capability for these agents is summarizing large log streams into human-readable reports. Further progress requires meaningfully evaluating this function, yet existing benchmarks leave end-to-end summarization from large log streams underexplored due to the prohibitive cost and complexity of collecting real-world data. We introduce SLS-Bench, a benchmark for incident log summarization featuring 90 tasks grounded in real-world incident reports, each with a synthetic log stream and a reference summary. To construct SLS-Bench, we developed a data generation pipeline that converts incident reports into structured incident models, which guide the generation of synthetic logs and reference summaries. This pipeline combines narrative grounding, explicit incident modeling, generator-verifier loops, and statistical comparisons with real-world log streams to produce realistic and controllable observability data. Our evaluation of 14 LLMs on SLS-Bench shows that while frontier models can produce useful summaries, they do so at substantial cost and latency, and open-weight models lag significantly. SLS-Bench provides a framework for systematic evaluation of log analysis agents and a scalable blueprint for future synthetic observability benchmarks.
[AI-261] Estimating the Causal Impact of Social Navigation Signals in Corridor Scenarios
链接: https://arxiv.org/abs/2610.16411
作者: Francesco Baldo,Luca Castri,Gloria Beraldo,Nicola Bellotto
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Deploying robots in human-shared environments requires deciding not only how to navigate, but also when dedicated social signals are worth emitting. Classic observation-based inference can be misleading in such settings: confounding factors may bias the estimated relationship between a signal and the task outcome, leading to incorrect decisions. Causal inference provides a framework for disentangling these relationships by modelling the mechanisms linking observations, actions, and outcomes, and by estimating interventional effects after accounting for confounding paths. We study a human-robot interaction scenario in a corridor and use a causal model to estimate whether a social navigation signal causally improves transit success. We represent the interaction as a causal DAG involving robot action, pedestrian occupancy, available space, task outcome, and a controlled confounder (static obstacles). We compare the interventional estimate with the observational estimate obtained by conditioning only on the robot action, which does not explicitly model interventions. Our results show that accounting for confounding variables can reverse the conclusions drawn from naive observational analysis: the observational estimate wrongly suggests that the social signal is detrimental, whereas the causal estimate shows that it is beneficial. This discrepancy highlights how failing to account for confounding can lead to incorrect conclusions and demonstrates the importance of causal models for context-aware signalling decisions in social robot navigation.
[AI-262] EvoWeb: A Benchmark for Coding Agents on Evolving Web Applications
链接: https://arxiv.org/abs/2610.16402
作者: David Nukrai,Bar Alon,Nativ Levy,Ido Andrew Atad,Ido Cohen,Itamar Zimerman,Lior Wolf,Eli David,Ilan Naiman
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: 24 pages, 6 figures. Code: this https URL ; Dataset: this https URL
Abstract:Web application development often proceeds through successive requests that extend an existing implementation. Yet web-generation benchmarks primarily evaluate development toward fixed specifications. We introduce EvoWeb, a novel benchmark comprising 320 cumulative requests over 32 web applications, grounded in real-world product features and structured by functional dependency graphs. Agents iteratively modify a persistent codebase, while a multimodal web agent tests each intermediate application in a live browser, assessing newly requested functionality, cross-request integration, and previously requested functionality. To characterize reliability throughout development, we integrate trajectory performance across user-tolerance thresholds rather than impose a fixed stopping criterion. Experiments across open- and closed-source frontier models show that EvoWeb reveals clear performance differences between models, with cross-request integration emerging as a key challenge. Further analysis shows that later requests with deeper or more numerous dependencies are solved less accurately, while runtime errors become more frequent over successive requests. EvoWeb provides a testbed for evaluating and improving coding agents in long-horizon, multi-request development.
[AI-263] Recursive Environment Scaling for General Long-Horizon Agents
链接: https://arxiv.org/abs/2610.16397
作者: Lingxiao Jing,Zhengbo Jiao,Yichen Xiang,Bofan Zhu,Lingbei Meng,Yan Ni,Haorui Chen,Ran Ran,Zhirong Shen,Yuheng Yang,Chenhang Cui,Yaosong Du,Rui Huang
类目: Artificial Intelligence (cs.AI)
备注: 20 pages, 10 figures. Code: to be released
Abstract:We study environment scaling as a way to increase agent-benchmark difficulty while holding core task objectives fixed. Our framework, Recursive Environment Scaling (RES), transforms existing benchmark environments while preserving their objectives and completion requirements. A proposer changes information access, resource relations, and execution dependencies; execution-grounded checks validate the resulting candidate, and accepted changes are inherited across rounds. This produces related environments for paired evaluation and training-trajectory collection. Experiments on 294 initial tasks from three benchmarks show lower task success for all six evaluation models, with decreases of 57.14–77.14 percentage points on Claw-Eval-Live. With equal trajectory counts and identical optimization settings, fine-tuning on RES trajectories outperforms original-trajectory SFT in all nine model–benchmark comparisons by 7.00–14.61 percentage points. Ablations identify recursive inheritance and dependency composition as important contributors to increased difficulty. Together, these results support environment scaling for both agent evaluation and training.
[AI-264] LoopQuant: Accurate and Efficient Post-Training Quantization for Looped Transformers
链接: https://arxiv.org/abs/2610.16368
作者: Zhixiong Zhao,Zukang Xu,Xing Hu,Lifeng Liu,Guangyu Sun,Dawei Yang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Looped Language Models (LoopLMs) improve parameter efficiency by recursively reusing shared Transformer modules, yet their repeated computation over evolving recurrent states poses new challenges for low-bit deployment. Existing post-training quantization (PTQ) methods, largely designed for feed-forward Transformers, fail to account for cross-loop activation dynamics, heterogeneous loop sensitivity, and recursive error propagation, leading to substantial performance degradation at low precision. To address these challenges, we propose LoopQuant, a loop-aware PTQ framework for accurate and efficient low-bit quantization of LoopLMs. LoopQuant is built on three core components: 1) Cross-Loop Distribution Canonicalization (CLDC) learns a shared quantization-friendly representation across recurrent states while preserving parameter sharing; 2) Interaction-Aware Loop-wise Precision Allocation (ILPA) allocates activation precision under a fixed bit budget by modeling task-level loop sensitivity and cross-loop interactions; and 3) Loop-Conditioned Task Calibration (LCTC) further aligns quantized inference with the low-bit recursive trajectory through lightweight task-space calibration. Extensive experiments across LoopLM architectures demonstrate that LoopQuant consistently outperforms existing low-bit PTQ methods. On Ouro-2.6B, LoopQuant under W4A4 surpasses all W4A8 baselines in performance, while reducing model memory by 65.3% with only 3.4% additional storage overhead over naive INT4 and achieving a 1.7 times decoding speedup over BF16. These results demonstrate a strong balance among accuracy, compression, and inference efficiency, providing a practical and deployment-friendly solution for low-bit LoopLM inference. The code is available at [this https URL].
[AI-265] World Models Need a Physical Registry: Learning Internal Predictive States for Physically Plausible World Models
链接: https://arxiv.org/abs/2610.16365
作者: Shinseong Kim,Jaeseok Jeong,Jianlong Fu,Youngjung Uh
类目: Artificial Intelligence (cs.AI)
备注: 36 pages, 26 figures, 14 tables. Project page: this https URL
Abstract:Video world models must generate physically plausible futures, not just realistic frames. Existing approaches typically use semantic, physical, or predictive information to guide the visual generation stream, leaving a fundamental question: where inside the generator should predictions of scene evolution live? Standard video tokens jointly encode appearance and dynamics, but provide no distinct state for learning and reading predictive dynamics. We answer this question with the Physical Registry: temporally grounded registry slots inside a video DiT that are initialized from an observation-conditioned forecast, co-evolve with video tokens, and receive future-representation supervision, making predictive dynamics an explicit, readable state of the generator. In a controlled comparison of predictive-state interfaces on Physics-IQ Verified, Physical Registry improves vanilla fine-tuning by 9.9/8.7 points on I2V/V2V, the largest gain among the interfaces we compare. Across all Cosmos and WAN backbones, Physical Registry improves Physics-IQ Verified scores. With only 0.45% additional parameters, it enables Cosmos 3 Nano 16B to surpass the published I2V score of Cosmos 3 Super 64B (43.56 vs. 42.7). On WorldModelBench, it also improves physical adherence and the total score for all four evaluated backbones. Because the registry becomes candidate-specific while remaining readable, the Physical Registry Selector (PRS) ranks emerging futures by how far their intermediate registry readouts depart from the shared forecast before full denoising is complete. Across four backbones, this selection improves I2V scores by 5.6-15.9% without completed-video external scoring. Together, these results support a broader principle: dynamics prediction should live as an explicit model state that co-evolves with the video representation. Project page: this https URL
[AI-266] DWI Connectome-based Sex Classification in Young using Supervised Variational Autoencoder
链接: https://arxiv.org/abs/2610.16348
作者: Aditya Kothapalli,Aaryan Gera,Sathvik Reddy V,Ninad Aithal,Debanjali Bhattacharya,Neelam Sinha
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:During adolescence, significant changes in brain development are influenced by hormonal changes that impact both the body and the brain. This study presents sex classification in 787 young adults having age range of 15-25 years using Diffusion-weighted (DWI) structural connectome, focusing on 82 Desikan-Killiany atlas ROIs. A supervised variational autoencoder (SVAE) is used to extract latent space features from DWI connectome data. Although several studies explored differences in brain structure and function between sexes, the main contribution of this study lies in incorporating SVAE to learn a compact, regularized latent representation for the task of sex classification. Initial exploratory data analysis shows that these features can capture the differences in sex-specific latent space patterns. Subsequently, the learned latent space SVAE features are fed to different ML models for sex classification. Among the considered ML classifiers, the multi-layer perceptron yields the highest accuracy of 80% in classifying sex. This demonstrates the effectiveness of SVAE in learning compact, informative representations from DWI connectome for sex classification. The present work highlights the potential of using DWI brain connectomics and deep learning to differentiate sex, paving the way for future research on diverse populations that may include transgender individuals, to better understand sex-related variations in brain structure.
[AI-267] Rescuing Points Trapped Inside Incorrect Decision Regions: A target for Evaluating and Improving Forward-Forward Learning
链接: https://arxiv.org/abs/2610.16347
作者: Alireza Rahmansetayesh,Amirhossein Yousefiramandi,Maxwell Spiridonov,Sumit Raut,Kent Ziti,Pierre Baldi
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Forward-Forward learning replaces end-to-end backpropagation with local, layer-wise training objectives. Its variants are evaluated primarily by classification accuracy, often measured with linear probes trained on their representations. We show that this evaluation can over-credit the learning method: a linear probe on the representations of an untrained convolutional neural network already reaches accuracy well above chance. Moreover, a single accuracy value does not reveal which datapoints become correctly classified after training and which remain misclassified. To examine what learning contributes beyond the random, untrained baseline, we present an analysis methodology built on the Boundary Proximity Index (BPI), a multiclass margin normalized by a class-centroid reference. We apply our analysis to seven Forward-Forward variants and related layer-local methods and find that initially misclassified points near the probe’s decision boundary generally have higher rescue rates than those farther inside incorrect decision regions. Rescue rates vary substantially across methods, with some correcting a substantial fraction of these deeply embedded points. Together, these analyses offer a way to evaluate representation learning beyond accuracy and identify the rescue of points far inside incorrect decision regions as a target for improving local learning methods.
[AI-268] ChronoScribe: Semantic-Numerical Specialization for Instruction-Driven Time Series Forecasting
链接: https://arxiv.org/abs/2610.16345
作者: Yuchen Jiao,Miao Zhang,Yuxuan Liu,Chengxi Li
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 9 pages, 7 figures
Abstract:Recent advances in instruction-driven time series forecasting highlight a fundamental tension: systems must simultaneously master natural-language instruction following and precise numerical prediction. Current approaches often force a compromise. Large language models (LLMs) can interpret unstructured requests but may be inaccurate when asked to produce forecast values directly, whereas time series foundation models (TSFMs) provide numerical forecasts but typically require structured inputs specifying the target series, context window, and horizon. We present ChronoScribe, which applies semantic-numerical specialization to this workflow. A LoRA-adapted LLM proposes an execution path and context length; deterministic rules resolve the executable configuration. On the specialist path, a frozen TSFM produces quantile trajectories for the task-specified horizon. Deterministic trajectory summaries condition the response, while the LLM neither generates nor revises predicted values. We evaluate a Qwen3.5-9B–Chronos-2 implementation across 23 evaluation dimensions and 30 public datasets against eleven proprietary direct LLMs and three open TSFMs. ChronoScribe has the lowest error on sMAPE, MASE, WAPE, RMSE, and MAE among all evaluated systems, including fixed-context controls. Across these metrics, even the lowest direct-LLM error remains 2.0–2.7 times the corresponding ChronoScribe error. At the system level, ChronoScribe achieves a Task-Level Score (TLS) of 65.5, 10.6 points above the strongest direct-LLM baseline. The results support ChronoScribe’s division of labor, in which the LLM handles interpretation, configuration, and communication while the TSFM handles numerical forecasting.
[AI-269] Efficient Multi-Agent Reasoning via Confidence-Guided Adaptive Debate ICML2026
链接: https://arxiv.org/abs/2610.16343
作者: Seungdong Yoa,Ye Seul Sim,Suhee Yoon,Sanghyu Yoon,Dongmin Kim,Soonyoung Lee,Bumsoo Kim,Junhyun Lee
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted at ICML 2026
Abstract:Multi-agent debate has shown promise for improving the reasoning of large language models, yet recent theory suggests its benefits are highly regime-dependent. While interaction can amplify informative signals under corrective conditions, symmetric debate dynamics are neutral in expectation, often making majority voting preferable. We reconcile these views by arguing that debate is effective only when invoked at the right time and with appropriate structure. Based on this insight, we propose LASE: Leader-Adaptive Structured Engagement, a leader-centric multi-agent debate framework that selectively engages interaction only in non-neutral regimes. LASE introduces an asymmetric leader-supporter architecture that enables directed information flow and selective signal amplification, while defaulting to simple aggregation otherwise. Experiments across diverse reasoning benchmarks show that LASE achieves multi-agent-level performance with near single-agent token cost, substantially improving efficiency over static debate and voting baselines.
[AI-270] RAP: Tool-result Replay Attack for Probing Misbinding in LLM Agents
链接: https://arxiv.org/abs/2610.16333
作者: Yutong Zhi,Jie Zhang,Xiuqi Lu,Jiayuan Xu,Jing Shao
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:As AI agents increasingly act through external tools, tool results are no longer simple outputs but execution artifacts that must remain correctly associated with the context that produced them. Yet as results persist across caches, recovery mechanisms, and shared runtime state, this association can break, causing a real and unmodified result to be consumed in an incompatible execution context and lead the agent to an unsafe action. We characterize this failure as tool-result misbinding. To systematically study this failure, we introduce ReplayBench, a benchmark that uses controlled tool-result replay to evaluate misbinding across 19 representative agent scenarios, 4 failure properties, and 3 replay mechanisms. Across eight LLMs, three agent frameworks, and diverse stateful workflows, we find widespread replay-induced failures, with a mean Final-State Error Rate of 0.649 under cross-task replay. Stage-wise analysis localizes these failures to model-side action generation. Further analyses show that replay susceptibility varies with task similarity, shows little dependence on object similarity, and remains high under result staleness and implicit environment changes. These findings motivate runtime validity mechanisms that eliminate final-state errors in our evaluated settings while preserving legitimate recovery. The results call for explicit contextual binding in tool-using agent runtimes. Our work is public at this https URL.
[AI-271] When a Conceptor Complement Becomes Gain: Operator Geometry and Behavioral Attribution in Language Models NEURIPS2026
链接: https://arxiv.org/abs/2610.16325
作者: Shuhul Razdan,R Balamurugan
类目: Artificial Intelligence (cs.AI)
备注: 50 pages. Accepted at the NeurIPS 2026 NeurReps Workshop (Symmetry and Geometry in Neural Representations). Code: this https URL
Abstract:When a steering intervention changes a language model’s behavior, which part of the intervention deserves the credit? We test this question for a target-centered, scaled conceptor complement by removing its fitted covariance filter while preserving the same target-specific center and gain. At the original aperture \alpha=.05 and gain \beta=2 , all 21 fitted maps across three models and seven targets expand every direction at the generation intervention layers. The matrix correction is 0.093% – 7.815% of the matched-control displacement on held-out extraction activations. In 480 paired comparisons within a 1,920-response study, retaining the matrix changes many individual texts but does not establish a consistent additional target-suppression benefit on the four-target panel. A geometric aperture sweep shows that the operating regime changes with the filter setting. Separately, an invertible soft filter preserves what an unrestricted affine readout can recover, whereas dedicated erasure methods reduce prediction by newly fitted probes. Together, the results show why a whole-intervention comparison cannot by itself attribute generated behavior to the fitted matrix: analyze the complete transformation, hold shared components fixed, and test its output consequences directly.
[AI-272] HyWe: Unifying Search and Operation for Real-World Hybrid Web Tasks
链接: https://arxiv.org/abs/2610.16319
作者: Zijian Cai,Haowei Liu,Kailai Chen,Lin Peng,Cong Wan,Ziqi Zhang,Chunfeng Yuan,Bing Li,Weiming Hu,Haiyang Xu
类目: Artificial Intelligence (cs.AI)
备注: 40 pages, 6 figures, 13 tables, including appendices
Abstract:Real-world hybrid web tasks need both information gathering and page interaction, yet search-based and operation-based agents are developed and evaluated separately. To bridge this gap, we introduce HyWe, a framework that unifies both capabilities within one model and trajectory. Its two modes use dedicated prompts and tool sets to sustain sub-goal execution; explicit switching selects the needed capability, and shared history lets each mode build on the other’s findings. To provide demonstrations of these behaviors, we develop HyWeCollector, a fully automatic pipeline that uses task prototypes, four switching patterns, and multi-agent synthesis to produce verified hybrid trajectories. We train HyWe-35B on these demonstrations to unify efficient information gathering and browser interaction in a single 35B web agent. To evaluate hybrid task completion, we introduce HyWeBench: 100 human-written tasks on websites distinct from training, with human-written sub-goal checklists and an outcome-based evaluation pipeline. Hybrid training enables HyWe-35B to achieve 47.0% success on this benchmark, improving on the untrained base model by 19.0 percentage points and outperforming existing search-based and operation-based agents. Even before training, the same base model solves 28.0% of tasks with HyWe, versus 11.0% with all tools under one prompt. Beyond hybrid tasks, HyWe-35B achieves 59.0% and 88.3% success on Online-Mind2Web and WebVoyager, outperforming existing open-source baselines and cutting action cost per successful task by 48.3% and 51.9% compared with Fara-7B.
[AI-273] ReTAC: Residual Test-Time Agent ic Computation for Video Spatial Reasoning
链接: https://arxiv.org/abs/2610.16315
作者: Yike Wu,Keyan Chen,Qixiong Wang,Xi Xiang,Maofei Que,Xiaolong Jiang,Xiaoqin Zhang,Ling Shao,Shijian Lu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Test-time agentic computation can improve multimodal large language models (MLLMs) through planning, tool use, iterative perception, and reflection, but indiscriminate computation may duplicate resolved work, introduce noisy evidence, and perturb correct predictions. We propose that such computation should be residual: allocate additional computation only to requirements unresolved by the base model and, after a localized failure, recompute only the affected states. We instantiate this formulation in ReTAC, a training-free multiagent framework for video spatial reasoning. ReTAC organizes supplementary computation around unresolved requirements. It plans ordered steps to acquire missing spatial evidence and combines the observations with compatible native evidence for reasoning. In addition, it uses stage-aligned verification to localize failures to planning, perception, or reasoning. Dependency-based recovery recomputes only affected states while reusing unaffected ones. Extensive experiments show that ReTAC achieves the highest overall score among the compared methods across three mainstream MLLMs and two widely adopted benchmarks. Further analyses show that gains from additional computation depend not only on how much is used, but also on where it is allocated and which valid intermediate states are preserved.
[AI-274] Narrative Paradoxes in LLM Inference: Shaping Reasoning Trajectories toward Misalignment
链接: https://arxiv.org/abs/2610.16312
作者: Haoming Yang,Ke Ma,Yangbangyan Jiang,Ligong Zhang,Xiaojun Jia,Longtao Huang,Yingfei Sun,Qianqian Xu,Qingming Huang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Explicit reasoning in large language models (LLMs) is often associated with reliability, yet it can still pose risks when a model infers harmful goals under a benign context. We identify a distinct challenge: ensuring the safety of the \emphmulti-step reasoning process itself, where locally coherent steps may still lead to globally harmful outcomes. To explore this vulnerability, we propose PDRA (Paradox-driven Reasoning-based Attack), a novel attack method that programs a harmful reasoning trajectory. PDRA first constructs a paradoxical narrative core by pairing a malicious actor (e.g., a terrorist) with a benign task (e.g., writing a safety bulletin). It then embeds this core into a structured analytical framework that forces the model through a mandatory three-stage analysis – surface purpose, latent intent, and operational reconstruction – guiding it to produce detailed harmful content while maintaining narrative coherence. Mechanistically, we show PDRA steers the model’s internal representations away from refusal-related patterns, suppressing safety-triggering lexicons while activating planning-oriented terms. Empirically, PDRA achieves state-of-the-art attack success rates across diverse LLMs with high efficiency. \textcolorred\textbfWarning: This paper may contain sensitive content.
[AI-275] FON: A Draft-Revise Dual-Organ Language Model without Sequence-Level Quadratic Attention
链接: https://arxiv.org/abs/2610.16301
作者: Helong Hu(1),Hongdan Pan(2),Shuiqing Hu(3) ((1) Zhanjiang Campus, Guangdong Ocean University, Zhanjiang, Guangdong, China, (2) Maoming Administrative Service Center, Maoming, Guangdong, China, (3) China Pacific Insurance (Group) Co., Ltd. Foshan Branch, Foshan, Guangdong, China)
类目: Artificial Intelligence (cs.AI)
备注: 24 pages, 14 figures, 9 tables
Abstract:Self-attention assigns two functions to one mechanism-where to read and how to mix-at the price of O(L^2) training, an O(T) decode state, and position signals bound to the training length. Existing attention-free alternatives all optimize the single function of generation. We propose FON (Fusion-Organ Network), which differentiates one weight stack into a drafting organ (multi-tap learnable delays + content-adaptive recursion; causal, O(S) training, O(1) decode state, no sequence-level quadratic attention) and a revision organ that re-processes the written prefix through multi-scale ripple kernels injected via weak gates, never generating. At 30M parameters, iso-parameter and iso-token budget, FON reaches perplexity 131.03 on WikiText-103, beating Transformer (197.73) and Mamba2 (152.53); the advantage persists under all depth-aligned baselines and at the 70M tier (-20.8%). At 8x length extrapolation FON degrades by ~8-9% without positional-encoding engineering, between ALiBi and RoPE. The revision channel provides a structural pathway for in-domain infilling that causal architectures lack in principle (+19% over on-demand coupling), at a perplexity cost indistinguishable within measurement precision. We report all costs and boundaries: impaired long-range retrieval, synthetic-task disadvantages, weaker out-of-domain code infilling than a same-recipe Transformer, and a 6.5-12.3x wall-clock gap.
[AI-276] Memory-Efficient Integer-Only Training of Deep Neural Networks
链接: https://arxiv.org/abs/2610.16298
作者: Hazem Hesham Yousef Shalby,Fabrizio Pittorino,Manuel Roveri
类目: Artificial Intelligence (cs.AI)
备注: 19 pages
Abstract:Training neural networks in reduced precision can yield significant efficiency gains, yet all existing methods that scale to standard computer vision benchmarks retain floating-point in at least one stage, typically the weight update or optimizer state. We present a training algorithm that scales to modern deep neural networks, in which every tensor (weights, gradients, and activations) is stored as INT8, and every operation (forward pass, backward pass, weight update, optimizer update) is performed in integer arithmetic. Specifically, tensors are stored as signed integer mantissas paired with explicit power-of-two exponents, on which integer-only operations are performed using five custom composable building blocks (products, sums, divide-by-constant, nonlinear functions, and weight updates). This allows complete implementation of trainable integer layers, including convolutions, batch and layer normalization, softmax, and GELU. A per-group velocity mechanism with independent exponent tracking preserves gradient precision across neurons and channels with diverse magnitudes. As every tensor fits in INT8, training memory is roughly 4x smaller than the FP32 baseline. We validate our training algorithm on CNNs (ResNet-18/34/50/101, MobileNetV2) and Vision Transformers (ViT-Ti / B), demonstrating for the first time that training neural networks with INT8 tensors and integer arithmetic can achieve competitive accuracy on CIFAR-10, CIFAR-100 and ImageNet at a reduced memory budget.
[AI-277] SLDBench v2: Active Scaling Law Discovery
链接: https://arxiv.org/abs/2610.16297
作者: Lingbei Meng,Rui Huang,Zhirong Shen,Jinxin Liu,Lingxiao Jing,Haorui Chen,Yaosong Du,Haowei Lin
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Scaling laws predict large-model performance from small experiments, but finding a reliable law itself requires costly experiments. SLDBench tests whether language-model agents can discover scaling laws when all experimental results are given upfront. We introduce SLDBench v2 with two extensions. First, we expand the corpus from 8 to 127 scenarios spanning language-model pretraining, downstream performance, and vision-language transfer, and select 30 challenging scenarios on which a reference power-law fit reaches R^2 \le 0.8 ; they represent at least 1.09 \times 10^23 FLOPs of source-reported training compute. Second, we introduce active scaling law discovery: an agent chooses which experiments to run, pays each one’s recorded cost (e.g., training FLOPs), and revises its law as outcomes arrive, and is scored on the same hidden extrapolation targets as in the passive setting. Across 8 frontier agent systems, median R^2 drops from 0.72-0.75 (passive) to 0.24-0.57 (active), and prediction error rises 1.5-4.7 times, depending on the system. Agents favor cheap experiments, spending 8-44% of the full-pool cost on average; low-cost acquisition can yield strong predictions, but cheap purchases also leave experimental conditions needed for extrapolation unobserved. Experimental planning thus remains a key open challenge for agents that assist model development in the prevailing AI-for-AI study.
[AI-278] A Survey of Flow Matching in Reinforcement Learning
链接: https://arxiv.org/abs/2610.16275
作者: Nabuat Zaman Nahim,Fairoz Nower Khan,Peizhong Ju
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 52 pages. Published in Transactions on Machine Learning Research (TMLR), 2026. Survey Certification. Reviewed on OpenReview: this https URL
Abstract:Flow Matching (FM) has recently emerged as a principled and efficient generative modeling framework for reinforcement learning (RL), enabling expressive, multimodal policy parameterizations via deterministic probability transport. Compared to diffusion-based policies that rely on stochastic denoising chains, FM uses sampling based on ordinary differential equations (ODEs), with learned velocity fields, which can substantially reduce inference latency and simplify the incorporation of RL objectives. As research in flow-based RL rapidly accelerates across offline continuous control, online fine-tuning, and foundation model alignment, the literature has become highly fragmented. In this survey, we provide a comprehensive taxonomy of flow-matching approaches in reinforcement learning. We organize the literature along two axes: the target distribution being modeled (e.g., action policies, value critics, transition dynamics) and the mechanism of RL signal integration (e.g., energy-weighted regression, flow-based policy gradients, and group relative policy optimization). Furthermore, we survey emerging frontiers such as discrete and non-Euclidean action spaces, provide a systematic comparative analysis against Gaussian and diffusion baselines, and outline critical open problems. Ultimately, this survey serves as a foundational roadmap for the next generation of generative reinforcement learning and alignment.
[AI-279] ScreenASR-Bench: A Tiered Benchmark for Speech Recognition with Visual Context
链接: https://arxiv.org/abs/2610.16260
作者: Mingwei Fu,Zheshu Song,Zhikai Zhou,Shen Huang,Shidong Shang,Mengxiao Zhu
类目: Artificial Intelligence (cs.AI)
备注: 5 pages
Abstract:Current ASR models perform well on general benchmarks but still struggle with acoustically ambiguous words. Visual context can aid recognition in screen-based scenarios, yet existing benchmarks provide limited insight into the benefits of different types of visual evidence. To address this gap, we introduce ScreenASR-Bench , a tiered benchmark with three progressively challenging levels: Explicit Textual Evidence (ETE), based on direct on-screen text; Explicit Visual Evidence (EVE), using visible content rather than on-screen text; and Implicit Semantic Evidence (ISE), requiring scene-level inference. Extensive experiments on ASR models, Large Audio-Language Models (LALMs) and Omni models show that incorporating visual context substantially improves recall over the audio-only setting across all three model families. The gains are largest for ETE and smallest for ISE, indicating that models leverage direct visual evidence more effectively than indirect semantic cues. To facilitate future research on visually grounded speech recognition, we publicly release ScreenASR-Bench.
[AI-280] Reflected Proximal Dynamics for Accelerated High-Accuracy Log-Concave Sampling
链接: https://arxiv.org/abs/2610.16248
作者: Bo Yuan,Wei Guo,Yongxin Chen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Statistics Theory (math.ST)
备注: Under Review
Abstract:We study sampling from \pi(x)\propto\mathrme^-V(x) on \mathbbR^d , where V is \alpha -strongly convex and \beta -smooth with condition number \kappa=\beta/\alpha . For this problem, the proximal sampler preserves the target exactly and recovers overdamped Langevin dynamics in the small-step limit, but its Gaussian resampling step discards directional information. Underdamped Langevin dynamics retains such information through momentum to accelerate exploration, motivating a proximal construction that combines target preservation with directional persistence. We introduce \textitreflected proximal dynamics, which combines proximal sampling with an exact reflection of the auxiliary variable, conditional kinetic motion, and controlled refreshment. We develop two realizations: a flux-resampled underdamped half-turn and a persistent-momentum quarter-turn. From a warm start with constant \chi^2 divergence, both methods achieve total variation error \varepsilon using \widetilde O(\sqrt\kappad^1/4) expected oracle queries. We further remove the supplied warm-start assumption by combining either sampler with existing posterior samplers in a proximal initialization stage with an optimized coarse step size. Accounting for both initialization and sampling yields \widetilde O(\kappa^6/7d^1/4) expected oracle queries. In both regimes, the dependence on the target accuracy \varepsilon is \mathrmpolylog(1/\varepsilon) , making reflected proximal dynamics a promising framework for accelerated high-accuracy log-concave sampling.
[AI-281] AutoNovo: Agent ic Program Search Discovers Transferable Nucleic Acid Design Algorithms NEURIPS2026
链接: https://arxiv.org/abs/2610.16232
作者: Joel Shor,Subhashini Venugopalan
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Quantitative Methods (q-bio.QM)
备注: 40th Conference on Neural Information Processing Systems (NeurIPS 2026). Workshop: Agentic AI for Biological Discovery
Abstract:In-silico optimization of nucleic acids relies on predictive oracles that imperfectly capture biological function, so sequences optimized against one model may exploit model-specific artifacts and fail to generalize to independent models. Recent studies have shown that not all designers generalize equally well to independent models. In this work, we use agentic program search to iteratively propose, execute, evaluate, and revise code for nucleic-acid designers. We describe the successful strategies used to prompt the search. AutoNovo designers achieved the best observed mean among the evaluated methods on 3/3 DNA-design tasks in terms of in-silico performance, and 6/6 transfer tasks in terms of independent models. The designers come from a family of discovered nucleic acid designers rather than a single universal best designer. We plan to release the resulting designers. These results are the first steps in creating nucleic acid designers that maximize cross-model generalization, and thus improving the odds that in-silico designed sequences reflect real biology.
[AI-282] Learning to GUIDE the Group: Self-Learning LLM Facilitation Across Multi-Party Tasks
链接: https://arxiv.org/abs/2610.16231
作者: Ziyi Liu,Tzu-Sheng Kuo,Jiale Liu,Deqing Fu,Nithum Thain,Crystal Qian
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:As large language models are increasingly adopted as multi-party communication facilitators, there is a growing demand for agents capable of balancing diverse group dynamics across contexts. However, existing LLM facilitators are often fine-tuned for task-specific objectives; generalized LLM-based facilitation methods that can adapt to different types of tasks and diverse group dynamics are underdeveloped. To address this gap, we present GUIDE (Generalized Unified Intervention for Dynamic Exchanges, a framework for developing generalized LLM-based facilitators across different multi-party contexts. GUIDE evaluates facilitation across both objective task outcomes and subjective metrics, including psychological safety and perceived friction. Within this framework, we study whether LLMs can self-learn generalized facilitation strategy offline to real-time interventions. We first collect multi-party human interactions guided by task-specific expert LLM facilitators (N=432 conversations). We then employ a self-learning method in which an LLM induces an actionable strategic playbook from conversations and iteratively refines it using feedback from an LLM judge aligned with human annotators (N=20), developing a single generalized facilitator that outperforms the three task-specific expert facilitators on held-out conversations 61.4% win rate. Finally, we evaluate these improvements empirically; in a real-time online empirical study, the self-learned generalized facilitator preserves facilitator performance and group productivity, improves speaking comfort and reduces interpersonal friction, while using 24.7% fewer turns and 29.0% fewer words than the expert prompts. These findings provide empirical evidence that LLMs can learn generalizable, task-agnostic social facilitation from task-specific human interactions, on-par with hand-crafted task-specific prompts across diverse group settings.
[AI-283] OneShot: Label-Efficient Detector Weighting for Time Series Anomaly Detection
链接: https://arxiv.org/abs/2610.16219
作者: Seunghan Lee,Jun Seo,Jaehoon Lee,Junhyeok Kang,Sangjun Han,Sungdong Yoo,Minjae Kim,Tae Yoon Lim,Dongwan Kang,Hwanil Choi,Soonyoung Lee,Wonbin Ahn
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Time series (TS) anomaly detection (AD) lacks a universal detector, with the best choice determined by the series rather than by the algorithm. Existing methods therefore either select one detector for each series or average a pool of them. However, both assume that no label ever arrives, while a deployed monitor does receive them, since an operator confirms the alarms it raises. In this paper, we propose ONESHOT, which uses a single confirmed anomaly to weight the pool rather than to choose one detector from it. Specifically, we score each detector on the confirmed anomaly, and those scores become the weights of a soft mixture over the whole pool. We show that a single confirmed anomaly is enough to come close to the best detector for the series. Across 152 univariate and 73 multivariate series, ONESHOT outperforms the strongest single detector, averaging the whole pool, and even picking the best detector on the confirmed event. Code is available at: this https URL.
[AI-284] SafeDSE: Joint Exploration of Safety Architecture Patterns Mapping and Scheduling for Dataflow Applications With Safety Arguments (Extended Version)
链接: https://arxiv.org/abs/2610.16213
作者: Fatemeh Bahrami,Seyed-Hosein Attarzadeh-Niaki
类目: Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO); Software Engineering (cs.SE)
备注: 38 pages, 12 figures, 29 tables. Extended version of a paper submitted for publication; tool and data at this http URL
Abstract:Embedded platforms increasingly host safety-related streaming functions next to less critical ones. Whether a safety architecture pattern, such as a doer/checker pair, protects a function depends on the cores that run its components. Their schedule must meet the throughput requirement, and mixing safety integrity levels (SILs) on one core raises development cost. This report presents SafeDSE, a design space exploration framework for synchronous dataflow applications. One declarative constraint model decides the pattern of each actor, the placement that the fault model requires, the static-order schedule, and the platform, and prices SIL promotion. Guards activate the chosen patterns in a fixed graph that holds all of them, and node potentials encode throughput. An independent verifier re-checks every solution, and a generator turns its checks into evidence of a safety argument in Goal Structuring Notation. On 9 benchmark instances under period bounds, no rule that fixes patterns before mapping was both always feasible and never more expensive under conservative pricing, while joint exploration was. In the generated arguments, 49 to 62% of the leaves cite verifier checks; the rest are claims a mapping cannot discharge. Optimality within the model was proven for all synthetic graphs with up to 16 actors. This extended version adds the full model, the verifier and generator in detail, the complete catalog and parameters, and a proof of the potential characterization. It also reports every experiment per instance, a mutation analysis of the verifier, and the negative results of the development.
[AI-285] Correct Choices Failed Sequences: Choice2Seq Benchmarks Correctness Retention Across Output Interfaces
链接: https://arxiv.org/abs/2610.16191
作者: Tianyou Lai,Wentao Yue,Zhiying Deng,Qilei Li
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:A model may select the correct answer yet fail to construct it through a sequence of local decisions. We introduce Choice2Seq, a diagnostic benchmark that measures correctness retention across complete-answer selection and sequential construction in independent runs on the same problems and candidate sets. Sequence Composition Rate (SCR) measures the fraction of correctly selected items that are also answered correctly through local construction. Jev achieves 97.1% and 99.5% selection accuracy on arithmetic and lexical retrieval, yet retains only 33.0% and 2.1% of those correct choices under serial construction, respectively. Substantial gaps also emerge for StartLux and Qwen, including when Qwen emits characters directly. Restricting local options to candidate-supported characters improves construction, while explicit position indexing improves copying fidelity, even for incorrect selections. These interventions distinguish task correctness, plan fidelity, and output compliance. Matched dialogue-state evaluations reveal that local field construction largely preserves Jev’s correct states but reduces Qwen’s mean-turn accuracy, highlighting task- and interface-dependent differences. Sequence-level confidence and request-cost analyses further characterize completed outputs.
[AI-286] Label-Centric Knowledge Retrieval for Zero-Shot Text Classification
链接: https://arxiv.org/abs/2610.16182
作者: Daoping Wu,Dong Wang,Siyu Liu,Chen Liang,Jian Geng,Ziqi Li,Shuguang Xiong,Ying Zhang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Flagship large language models (LLMs) offer rich general knowledge and strong reasoning capabilities, but their high inference cost and latency limit their use for large-scale data labeling. We study zero-shot text classification with semantically meaningful labels and task descriptions, without task-specific labeled examples or classifier feedback. We propose \textbfSLKR (Semantic Label Knowledge Retrieval), a label-centric retrieval-augmented generation framework that separates offline knowledge construction from lightweight online inference. Offline, a flagship LLM independently generates label definitions, retrieval keywords, and boundary rules for potentially confusable label pairs, forming a reusable label-knowledge store. Because this construction operates over the label space rather than the input corpus, its one-time flagship cost can be amortized across large-scale inference. Online, hybrid semantic and lexical retrieval first identifies the top- K candidate labels and then selects up to M input-relevant boundary rules connecting candidate-label pairs. A lightweight LLM performs the final classification in a single call using the retrieved definitions and rules. Across three public benchmarks and a real-world multilingual book-classification dataset, SLKR achieves the best results on all four datasets with Qwen3.5-4B and on three of four with Ministral-3-3B, while using compact input contexts and no task-specific labeled examples. Boundary ablations further show that retrieving a small set of input-relevant rules preserves performance close to using all boundaries with substantially less context, supporting efficient reuse of flagship-generated label knowledge for lightweight classification.
[AI-287] ProCARE: Real-World Study Automation via Profile-Grounded Evidence Contracts NEURIPS’26
链接: https://arxiv.org/abs/2610.16175
作者: Zhaoyu Fan,Bowen Han,Jincheng Ou,Yanwei Ren,Zehua He,Kaiyu Huang,Pengcheng He,Shanshan Ni,Chengchen Gong,Shuai Xue,Qingjiang Shi
类目: Artificial Intelligence (cs.AI)
备注: 42 pages, 5 figures, including appendix. Accepted by NeurIPS’26. Code and benchmark: this https URL
Abstract:Real-world studies (RWS) analyze routinely collected healthcare data, such as electronic health records and insurance claims, to estimate treatment effects, monitor safety, and identify risk factors outside randomized trials. Automating RWS with large language model agents is appealing but brittle: valid analyses depend on dataset-specific clinical semantics, including index dates, follow-up windows, patient/event-level units, and outcome boundaries that are often implicit in observational data. This leads to schema mismatches, temporal errors, inconsistent analytical granularity, and unsupported clinical claims. We present ProCARE, a profile-grounded framework that reformulates RWS automation as constrained compilation. ProCARE converts heterogeneous data profiles into Evidence Contracts specifying available variables, temporal attributes, distributions, and evidentiary limits, and compiles study objectives into RWS-IR, a strongly typed intermediate representation for variables, time relations, analysis units, and statistical tasks. Deterministic validation and localized repair then check and revise generated workflows across schema mapping, cohort construction, variable derivation, analysis, and interpretation. On 36 EHR-based benchmark tasks, ProCARE outperforms strong LLM-agent baselines in code executability, semantic consistency, and conclusion traceability, with automated violation diagnoses showing moderate-to-substantial agreement with blinded expert review.
[AI-288] Interactive 4D Volumetric Liquid Forecasting under Moving-Solid Interaction
链接: https://arxiv.org/abs/2610.16167
作者: Jongwon Ryu,Sunjae Yoon,Junyeong Kim
类目: Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE); Machine Learning (cs.LG)
备注:
Abstract:We propose the task of Interactive 4D Volumetric Liquid Forecasting: predicting the spatiotemporal evolution of a future dense 3D liquid velocity field conditioned on how a solid object moves through it. This task provides an upstream physical forecasting primitive for systems that need to reason about motion-conditioned liquid response, including robotic liquid manipulation, digital twins, and rigid-fluid video generation. Existing works have made efforts in predicting flow around static or prescribed solid geometries and synthesizing visually plausible fluid motion. Yet these approaches provide limited coverage for interactive liquid dynamics, as they are not capable of forecasting how a moving solid induces a future full-domain liquid response. In this forecasting scenario, the hydrodynamic effect is condensed near an evolving liquid-solid interface, but the prediction target is a global spatiotemporal rollout. To address this challenge, we propose a two-level recurrent forecasting framework, TIDE/TIDES: TIDE isolates localized interface-driven forecasting, while TIDES adds transport-aware coupling and explicit stabilization to reduce accumulated rollout error. Across adapted neighboring baselines, controlled component ablations, motion/size OOD tests, and long-horizon evaluation, TIDES achieves the best forecasting performance and the ablations identify localized interaction, transport coupling, and stabilization as the key contributors. Together, our task formulation, benchmark, and TIDE/TIDES study establish a controlled foundation for moving-solid-conditioned dense liquid forecasting. The code is publicly available at this https URL.
[AI-289] Fisher information and the geometry of memorization in neural networks NEURIPS2026
链接: https://arxiv.org/abs/2610.16161
作者: Daniel T. Bernstein,Chase Goddard,Luca Di Carlo,Vudtiwat Ngampruetikorn,David J. Schwab
类目: Machine Learning (cs.LG); Disordered Systems and Neural Networks (cond-mat.dis-nn); Artificial Intelligence (cs.AI)
备注: 20 pages (10 pages main text), 11 figures (6 figures main text), NeurIPS 2026
Abstract:A fruitful approach towards understanding generalization in machine learning involves characterizing how models separate generalizable structure from idiosyncratic or even corrupted features of training data. Fisher information has emerged as a diagnostic tool for this separation, leveraging loss geometry to distinguish generalization from memorization. Here we investigate this connection in vision models trained on CIFAR-10 with controlled noise that can be learned only through memorization. We show that projecting model weights onto the top Fisher eigenvectors decouples general task performance from noisy sample memorization, and we investigate how this decoupling presents differentially across layers. We show that the same phenomena are present in small transformers trained to solve a synthetic in-context learning task, and are reproduced in closed form in a simple, analytically solvable linear model. Our results reveal that Fisher information captures aspects of the geometric structure underlying how neural networks allocate capacity between generalization and memorization.
[AI-290] Uncovering Robustness Weaknesses in LLM Agents via Persona Evolution with Task-Integrity Constraints
链接: https://arxiv.org/abs/2610.16155
作者: Yutao Xie,Jingbo Yang,Guanyu Yao,Xiaoyu Yang,Duo Ding,Shiyu Chang
类目: Artificial Intelligence (cs.AI)
备注: 39 pages, 4 figures
Abstract:Interactive-agent evaluations commonly rely on standardized LLM-based user simulators that exhibit broadly cooperative behavior, potentially overlooking failures that arise under diverse but realistic user interactions. This can lead to overly optimistic estimates of agent robustness and provide limited failure signals for training agents that generalize across users. Existing approaches largely rely on predefined user behaviors, which may miss agent- or domain-specific weaknesses, while failure-oriented search can reduce task success by invalidating the underlying task rather than exposing genuine robustness failures. To fill this gap, we introduce TIPER, a task-integrity-preserving persona evolution method that automatically discovers agent-specific user policies exposing robustness weaknesses while keeping the underlying task valid and solvable. TIPER evolves persistent user personas through reflection-guided behavioral mutations and task-wise Pareto selection under a task-integrity constraint. Across four interactive-agent benchmarks, TIPER yields lower held-out avg@3 scores in all 12 GPT-5.4 settings, with observed reductions of 5.8 to 33.7 percentage points, and in all 28 agent-domain pairs spanning seven agents, with a median observed reduction of 22.9 points. Further analysis yields a behavior-to-failure taxonomy that connects discovered user behaviors to recurring agent failure modes. Together, these results show that TIPER enables more diagnostic robustness evaluation of interactive agents and provides informative failure analysis for future agent robustness improvement. Code repository: this https URL.
[AI-291] WildTrace: Evaluating Training Data Detection in Realistic Large-Scale LLM Post-Training
链接: https://arxiv.org/abs/2610.16152
作者: Hengxiang Zhang,Junxian Liu,Hongxin Wei
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Although post-training significantly enhances model performance, the use of massive undisclosed data raises concerns about benchmark contamination and copyright infringement. This drives growing interest in training data detection for post-trained models. Existing evaluations for training data detection often leverage benchmarks constructed post hoc, which can introduce distribution shifts. Alternatively, controlled evaluations rely on small-scale training setups that diverge from realistic post-training settings. As a result, their reported performance may not faithfully reflect the effectiveness of detection methods in real-world scenarios. To this end, we introduce \textsc\method, a reliable benchmark designed to evaluate detection methods in realistic, large-scale post-training settings. Built on a multi-stage pipeline, \method\space encompasses stage-specific models throughout the post-training lifecycle, spanning SFT, DPO, and RL. To perform a reliable evaluation, we construct model-specific evaluation sets across four diverse tasks, featuring distribution-matched members and non-members with verifiable membership labels. Our results reveal that even the best-performing method achieves an average AUC of only 0.529 on our benchmark, which approaches random-guessing level (0.5). Furthermore, our analysis suggests that evaluation protocols can significantly affect measured detection performance. These findings highlight the significant potential for developing effective, practical, and robust detection methods under realistic scenarios.
[AI-292] Large Language Model Assisted Transportation Planning through a Differentiable User Equilibrium Layer
链接: https://arxiv.org/abs/2610.16140
作者: Eunseung Lee,Gyeongjun Kim,Keemin Sohn
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Transportation planning has traditionally required substantial expertise in network modeling, traffic assignment, and optimization. As a result, strategic analyses such as network design, capacity expansion, and model calibration remain inaccessible to non-expert decision makers. Recent advances in large language models (LLMs) provide an opportunity to bridge this gap by enabling natural-language interaction with complex transportation planning tools. However, existing LLM-based systems primarily serve as interfaces to simulation software and lack the ability to efficiently search, evaluate, and optimize transportation solutions. This study proposes an integrated human-AI transportation planning framework that combines a large language model, a model context protocol (MCP) interface, an open-source transportation planning engine (AequilibraE), and a differentiable user equilibrium (UE) assignment layer. The framework allows non-expert users to execute network analyses and improve solutions for optimization. A differentiable UE layer is developed by directly differentiating the Karush-Kuhn-Tucker (KKT) conditions of the Beckmann traffic assignment problem. The proposed framework is demonstrated through two representative planning tasks: calibration of travel time functions from observed traffic counts and network design under budget constraints. The examples illustrate how differentiable transportation models and LLM agents can jointly democratize transportation planning and substantially reduce barriers between domain expertise and transportation decision making.
[AI-293] owards a Benchmark for Fire Safety Code Interpretation in the Hong Kong AEC Industry
链接: https://arxiv.org/abs/2610.16127
作者: Le Qiu,Zhuan Anne Zhang,Hok Ming Adeline Chan,Emmanuele Chersoni
类目: Computational Engineering, Finance, and Science (cs.CE); Artificial Intelligence (cs.AI)
备注:
Abstract:Large Language Models (LLMs) are increasingly explored in the Architecture, Engineering, and Construction (AEC) domain, yet their reliability in regulatory compliance remains uncertain. In this study, we evaluate the performance of LLMs on Hong Kong fire safety codes. In collaboration with domain experts, we create a localized benchmark dataset comprising both multiple-choice questions (MCQs) and open-ended questions (OEQs) to evaluate 12 widely used proprietary and open-source models. Our findings show that top-performing proprietary models achieve 80-90% accuracy on MCQs, whereas open-source models generally lag behind. However, on open-ended evaluation, expert human assessments reveal an accuracy range of only 50-70%, demonstrating that LLMs still have substantial room for improvement in specialized compliance tasks.
[AI-294] Imitation Holds a Key to Evolution: Localized Repair and Verified Skill Acquisition for Adaptive Agents
链接: https://arxiv.org/abs/2610.16125
作者: Kai Hu,Hua Huang
类目: Artificial Intelligence (cs.AI)
备注: 9 pages,2 figures
Abstract:An agent can retain successful trajectories without learning which part of an experience should repair a new discrepancy. We formulate passive imitation (PI) as an architectural loop that localizes a violated contract, extracts a local goal, proposes an effect-matched skill, tests a small patch, and retains evidence after full-task verification. We instantiate this loop as a deterministic symbolic controller and analyze its finite repair-query complexity under explicit identifiability and noninterference assumptions. Across ten simulated tool worlds, PI solves all 600 cold held-out composition tasks with three allowed small trials, compared with 6.17% for a whole-trajectory control, while an equally expressive compositional repair control matches PI. On an unseen tool revision, PI achieves 70.0% success versus 24.0% without small trials and 79.0% with self-trial enumeration; an exploratory 24-task adaptation stream subsequently raises first-submission success to 100%, supporting a limited correction-and-retention mechanism rather than a unique incentive advantage or general intelligence. The CPU-only implementation, integrity checks, and reproducible results are available at this https URL.
[AI-295] Constructed regional photovoltaic generation fields improve ultra-short-term distributed solar forecasting
链接: https://arxiv.org/abs/2610.16110
作者: Dayin Chen,Xiaodan Shi,Xin Xu,Zhongyue Su,Shibo Zhu,Dongxiao Zhang,Yuntian Chen,Jinyue Yan
类目: Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE)
备注:
Abstract:Ultra-short-term forecasting of densely distributed photovoltaic (PV) systems remains challenging because external meteorological and satellite products may not adequately resolve rapid local variations, while heterogeneous PV systems exhibit different generation responses under similar atmospheric conditions. Meanwhile, dense PV fleets provide spatially distributed measurements that reveal regional generation dynamics. Here, we reformulate fleet-level PV forecasting from modeling discrete station time series to predicting explicit, spatially structured regional PV generation fields. We develop SatPV-Field, a regional-to-local forecasting framework that transforms distributed PV measurements into image-like generation fields and models their spatiotemporal evolution. By combining historical PV fields with satellite-derived future shortwave radiation guidance and meteorological forecasts, the framework predicts future regional generation states, which subsequently provide location-specific guidance for individual-station forecasting. Experiments involving 496 distributed PV systems, including 446 source stations and 50 held-out target stations, demonstrate that these regional fields can be effectively forecast and used to improve station-level predictions. The proposed approach reduces station-level MAE by approximately 10.0% relative to the best-performing single-station temporal baseline and by 2.5% relative to a graph-based regional forecasting approach using the same source fleet. These findings establish an explicit spatial-field modeling paradigm that connects regional generation dynamics with individual-site forecasting.
[AI-296] Where Polanyi Meets Galois: Tacit Knowing as an Algebro-Deterministic Operation Toward Re-rooting AI-Generated Science
链接: https://arxiv.org/abs/2610.16102
作者: Hiroyuki Chuma,Kanji Otsuka,Yoichi Sato,Kodo Yamauchi
类目: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI)
备注: 27 pages, 6 tables, 2 figures
Abstract:We argue that tacit knowing is better modelled as an operating regime of one memory substrate than as a kind of content. The substrate is VaCoAl, an algebro-deterministic XOR-and-shift hyperdimensional memory over GF(2), realized as CASRAM (commercial SRAM-CAM) and PyVaCoAl (DRAM-CAM).In VaCoAl/PyVaCoAl a directed acyclic graph (DAG) reads two ways. Repairing every collision (Rescue mode) makes the trace exactly auditable but ranks no path higher; tolerating collisions (Don’t Care mode) yields a path-integrated confidence, CR _2 , that silently favours direct routes. This audit-valuation complementarity is a formal counterpart of Polanyi’s claim that tacit and explicit knowledge are not sharply divided. It separates contingent tacitness, which a new query brings out (AlphaGo’s game tree), from regime-bound tacitness, statable only in a regime where it is gone (AlphaGo’s hunches). (1) Polanyi’s from-to structure reads as non-commutative binding, the Identikit as role-filler composition, recognition as majority-vote coincidence. (2) Explicating the valuation destroys it, as attending to his fingers can make Polanyi’s pianist falter, yet a closed-form law states it. Polanyi, Nonaka, Tsoukas and Collins each capture part of this but narrowly; even Nonaka and von Krogh’s continuum cannot tell contingent from regime-bound tacitness. (3) Collins’s three tacit knowledges become strata, one outside the substrate. (4) Our mentor-student DAG is a tacit-transmission graph. (5) Terence Tao’s remark that AI-solved problems spawn few follow-ups is diagnosed as rootless explicit knowledge: checkable results whose route no one can follow and that no one yet understands well enough to extend, teach or judge. To re-root them, LLMs generate, VaCoAl/PyVaCoAl keeps an auditable trace with its confidence, and people value, a role virtue epistemology shows cannot be offloaded. Testable predictions follow.
[AI-297] he Sharp Directions Are Against You: Curvature Analysis of Activation Steering NEURIPS2026
链接: https://arxiv.org/abs/2610.16084
作者: Seyedarmin Azizi,Arya Fayyazi,Parsa Razmara,Erfan Baghaei Potraghloo,Souvik Kundu,Massoud Pedram
类目: Artificial Intelligence (cs.AI)
备注: Accepted at NeurIPS 2026
Abstract:Activation steering adds a vector to a language model’s internal representations at inference time. It is a lightweight alternative to fine-tuning for behavioral control. The standard construction, contrastive activation addition (CAA), is fragile, as on many inputs it shifts behavior in the wrong direction. To explain this fragility, we first study the geometry of the steering vector. We introduce steering Hessian, the matrix of second order derivative of the model’s behavioral loss with respect to the steering vector. It captures the output behavior response sensitivity of the model under small perturbations of the steering vector along each direction. Its top eigenvectors are sharp directions: tiny perturbations along them can cause large, erratic, input-dependent behavioral swings. Bottom eigenvectors are flat directions: behavior is robust along them. Removing the sharp component of a steering vector approximately preserves its direction and magnitude (cosine 0.974, norm reduction 3% for typical settings) while reducing the vector’s directional curvature by up to 12x (median 4x across 36 conditions). We evaluate on three model families (Qwen2.5-7B, Llama-3.1-8B, Mistral-7B), with four behavioral tasks, and 36 token-level conditions. Steering along sharp directions reverses the intended behavior in every condition tested, with anti-steer rates of 53-100%. Sharp directions account for only 1-5% of the CAA vector’s squared norm. However, they drive a disproportionate share of the gap between the steered and unsteered next-token distributions during autoregressive generation. We term this one-line correction as sharp removal; it outperforms standard CAA in open-ended generation across all 25 conditions tested. A jailbreaking experiment further shows that sharp directions act as noise rather than signal: they disrupt coherent behavioral control regardless of steering polarity.
[AI-298] Heterogeneous Agent ic Coevolution
链接: https://arxiv.org/abs/2610.16078
作者: Wenqian Ye,Ziwei Guan,Zi Yang,Shivani Modi,Cici Cong Wu,Yunlong He,Yuxin Chen,Zhengyu Zhang,Liang Luo,Venkatesh Ranganathan,Jianhui Sun,Yantao Yao,Wenlin Chen,Buyun Zhang,Ellie Dingqiao Wen,Henry Kautz,Aidong Zhang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Recent advances in self-improving LLM agents have demonstrated the potential of AI systems that drive their own intelligence. However, many domain-specific applications rely on specialized foundation models that possess strong domain expertise but remain dependent on human-designed improvements. Extending recursive self-improvement to these models presents a key challenge: as a specialist improves, the strategies used to advance it must also adapt. To address this challenge, we propose Heterogeneous Agentic CoEvolution (HACE), a framework that jointly evolves specialized foundation models and LLM agents through reciprocal learning. HACE improves the specialist through supervised learning while training the agent with on-policy reinforced self-training, guided by verified performance gains on the current specialist. The agent’s exploration, in turn, informs subsequent specialist training, enabling both models to adapt to each other’s progress. To sustain this reciprocal learning, HACE further employs evolutionary search to optimize their interaction harness, adapting how the models exchange information and select training experiences as their capabilities evolve. We evaluate HACE across three challenging domains: recommendation systems, protein function prediction in biology, and crystal stability prediction in material discovery. We show that HACE outperforms the evaluated self-improving methods on the primary task metrics across all three domains. These results shed light on heterogeneous coevolution as a universal approach to improving specialized foundation models through reciprocal interaction with a learning agent.
[AI-299] Beyond Single-Pathogen Forecasts: Predicting Shared Respiratory Hospital Burden from Influenza COVID-19 and RSV
链接: https://arxiv.org/abs/2610.16074
作者: Gaurob Saha,MinJae Woo,Lior Rennert
类目: Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Machine Learning (cs.LG)
备注:
Abstract:Co-circulating influenza, COVID-19, and respiratory syncytial virus (RSV) create a shared hospital-admission burden, because beds, staff, testing capacity, and respiratory support are allocated to total demand rather than to one pathogen at a time. This makes respiratory forecasting a target-construction problem as well as a modeling problem: disease-specific forecasts preserve attribution, whereas hospitals also need a single planning signal for combined burden. We developed a supervised 1- to 4-week-ahead framework for weekly admissions and compared two total-burden strategies: summing independently generated pathogen-specific forecasts and forecasting the combined admission target directly. Evaluation included retrospective held-out testing and prospective operational forecasts issued before the corresponding admission weeks were observed, with performance summarized using MAE and Percent Agreement (PA; the smaller of model-generated and observed counts divided by the larger). In retrospective total-burden evaluation, the combined-burden model had PA values of 0.85, 0.78, 0.74, and 0.76 and MAE values of 80.63, 200.09, 183.97, and 156.13 from week 1 to week 4 ahead, compared with summed-forecast PA values of 0.85, 0.80, 0.69, and 0.69 and MAE values of 100.14, 200.06, 205.65, and 217.57. Prospectively, the combined-burden model had higher PA and lower MAE at all four horizons, with the largest separations at weeks 3 and 4 ahead. These results support layered respiratory forecasting: pathogen-specific forecasts remain necessary for etiologic interpretation, while combined-burden forecasts better align with hospital preparedness and shared-capacity planning.
[AI-300] Understanding and Reducing Misalignment in Persuasion Optimization
链接: https://arxiv.org/abs/2610.16070
作者: Boyi Wei,Ruta Desai,Peter Henderson,Wenyue Hua
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 33 pages, 13 figures
Abstract:Frontier models are now trained for complex multi-agent tasks, from running a company to engaging in negotiations. These tasks necessitate training models to be persuasive. But what does this optimization pressure mean for model alignment? In this work, we study how reinforcement learning (RL) training for persuasion affects models’ use of misaligned strategies such as evidence fabrication and belief coercion, and what techniques can reduce such misalignment. We find that optimizing sender models solely for receiver belief shifts causes the average number of distinct misaligned strategies used per game to increase by more than 25%, and fabrications per game increase to more than 4 times relative to pre-RL models. We compare various mitigation approaches, including stronger strategy guidance, clean demonstrations, penalizing misaligned strategy usage, and improving the model’s ability to detect fabrications. We find that navigating the persuasiveness-misalignment trade-off in this setting is challenging, with the best trade-offs coming from combining guidance on aligned persuasion strategies with reward penalties on misalignment. Overall, our work provides a testbed for assessing misalignment stemming from RL-based persuasion training and shows concrete paths for reducing misalignment. Importantly, it highlights that without direct intervention, persuasion-involved RL environments are likely to naturally lead to large increases in the use of misaligned strategies.
[AI-301] When Agent Evaluation Becomes Unreliable: An Empirical Study of Oracle Failures and Configuration-Dependent Measurement in an LLM Agent Benchmark
链接: https://arxiv.org/abs/2610.16067
作者: Junjie Chen
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: Submitted to Empirical Software Engineering
Abstract:Agent evaluation harnesses are software systems that run test suites, generate inputs, and decide outcomes with test oracles. This paper audits such software for reliability defects that score tables cannot reveal. We introduce an oracle-integrity auditing approach with five state-integrity invariants and three known-answer controls. We evaluate it by injecting seven oracle defect classes across two task domains and a synthetic domain. All injected defect classes are detected, and the repaired oracle has no false positives. Applied to an agent benchmark, the audit reveals a reward-function defect that changes the environment state before post-run scoring. Across 4,320 episodes, replay-based correction changes database-pass labels from 4,194 to 2,883 and reduces measured success from 94.3% to 65.5%. We also study configuration-dependent measurement effects. A component ablation over 3,240 episodes shows that the agent’s policy documents, tool descriptions, and tool error strings influence observed trajectory reliability, with tool descriptions showing the largest exploratory effect (-9.2 pp) and error strings the smallest (-5.3 pp). A second task domain reverses the sign of the language effect, so the result is bounded to the tested configuration. Finally, we report a portability boundary: the auditing approach cannot validate a second independently implemented benchmark because the required reference terminal state cannot be reconstructed from its recorded API-call log. The paper releases a replication package containing the auditing suite, fault-injection catalogue, frozen records, deviation logs, and offline analysis entry points without network access, API keys, or model calls. Comments: Submitted to Empirical Software Engineering Subjects: Software Engineering (cs.SE); Artificial Intelligence (cs.AI) Cite as: arXiv:2610.16067 [cs.SE] (or arXiv:2610.16067v1 [cs.SE] for this version) https://doi.org/10.48550/arXiv.2610.16067 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Junjie Chen [view email] [v1] Mon, 12 Oct 2026 02:11:43 UTC (122 KB) Full-text links: Access Paper: View a PDF of the paper titled When Agent Evaluation Becomes Unreliable: An Empirical Study of Oracle Failures and Configuration-Dependent Measurement in an LLM Agent Benchmark, by Junjie ChenView PDFHTML (experimental)TeX Source view license Additional Features Audio Summary Current browse context: cs.SE prev | next new | recent | 2026-10 Change to browse by: cs cs.AI References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[AI-302] Cross-Protocol Personality Assessment with LLM s: Aligning Behavioral Evidence with Human Ratings
链接: https://arxiv.org/abs/2610.16050
作者: Wei Shan,Tianyi Zhang,Yuan Zong,Wenming Zheng
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Personality assessment from asynchronous video interviews (AVIs) plays a crucial role in recruitment by enabling standardized and behaviorally grounded evaluation for job applicants. However, most of the previous works are confined to a single interview protocol. Independently collected protocols differ in question sets and question-trait mapping. Thus, the human-rated evidence from one protocol cannot be used directly under another. To address this limitation, we propose the Cross-protocol Evidence-grounded Personality Assessment (CEPA) framework, which aligns behavioral representations across interview protocols. CEPA addresses protocol heterogeneity by projecting source behavioral records into the target protocol’s construct-oriented representation space. It retrieves behaviorally relevant and personality-complementary human-rated exemplars to ground an ensemble of LLM raters, which enables behavior-to-personality associations without target-specific supervision. In cross-protocol experiments, CEPA reduces average mean squared error (MSE) by 43.8% (Forward) and 17.0% (Reverse) relative to the strongest baseline. Our empirical experiments show that human-rated evidence from one protocol can support personality assessment under another without target-specific supervision or retraining.
[AI-303] Adaptive Compression and Targeted Perturbation: A Unified Framework for Generalized Audio Deepfake Detection NEURIPS2026
链接: https://arxiv.org/abs/2610.16049
作者: Jianqiao Cui,Haoxian Zhu,Hanbiao Meng,Yanghao Zhang
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注: 10 pages,1 figure, Accepted by NeurIPS 2026
Abstract:Existing audio deepfake detection (ADD) methods frequently exhibit limited generalization, as fixed bottleneck architectures fail to adapt to the heterogeneous information densities inherent in diverse spoofing artifacts. Moreover, standard augmentation often leads to a pathological robustness-accuracy trade-off. In this paper, we propose a unified ADD framework to address these challenges. Our approach integrates: (1) an Adaptive Audio Feature Integration (AAFI) module that dynamically selects optimal latent dimensions via a compression pool and a stochastic exploration mechanism, ensuring flexible representations across multiple types of dynamic attacks; and (2) a Mel-Band Adversarial Perturbation (Mel-BAP) strategy that applies targeted regularization in the mel-spectrogram domain to encourage the model to learn invariant decision boundaries. Extensive evaluations on benchmarks including ADD 2022, FakeOrReal, and In-the-Wild demonstrate that our model based on Whisper-small achieves a state-of-the-art average Equal Error Rate (EER) of 1.51%. Compared to existing SOTA models such as the 3-billion-parameter Resemble-Detect-3B-Omni, our framework reduces the parameter count by up to 91.6% while achieving a 60.7% improvement in average EER. In a separate cross-lingual evaluation, the Whisper-base configuration achieves EERs of 3.91-5.65% on five unseen languages. The framework also maintains strong resilience to real-world acoustic distortions. This work provides a scalable and efficient “plug-in” solution for enhancing trust and security in the audio Web ecosystem.
[AI-304] AI Representations of Mental Health Stigma: Quantifying the Gap Between AI Estimations and Human Perceptions Across Gender Groups Using GPT -Based LLM s
链接: https://arxiv.org/abs/2610.16042
作者: Soojong Kim,Hye Min Kim,Kwanho Kim,Dongdong Zhu
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI)
备注: Accepted for publication in Social Science Medicine (to appear)
Abstract:The increasing use of artificial intelligence (AI) in mental health applications raises concerns about its potential to internalize and perpetuate societal biases. This research evaluates both disparities and misalignment in how large language models (LLMs), GPT-4 and GPT-5, represent perceptions of mental health across gender groups. The research comprises two studies. Study 1 applied widely adopted measures of Social Distance, Stereotypes, and Attitudes Toward Seeking Help to assess whether the AI models depict systematic gender differences. Study 2 compared AI estimations of Personal, Self, and Perceived Stigma with large-scale real-world human data from the Healthy Minds Study (N = 76,406) to evaluate alignment between AI estimations and human perceptions. Results from Study 1 showed the AI models consistently portrayed men as having the greatest desire for social distance from someone with mental health problems, relatively higher Stereotypes scores, and the least favorable attitudes toward help-seeking, compared with women and nonbinary individuals. Study 2 confirmed that this ordinal pattern partially aligned with human-reported data for Personal and Self Stigma. However, the AI models systematically underestimated Personal Stigma and Perceived Stigma while substantially overestimating Self Stigma for all gender groups. These findings reveal that, although the AI models have learned relative gender-based patterns reflecting human experiences to a certain extent, they function as distorted portrayals, highlighting the need for rigorous, context-specific validation against real-world data to ensure AI’s safe and equitable deployment.
[AI-305] Autoregressive Decoding as a Martingale: Characterizing the Stochastic Dynamics of LLM Generation
链接: https://arxiv.org/abs/2610.16041
作者: Xiahui Zhu,Zhenyi Qi,Wanwei Liu,Haoang Chi,Yang Tan,Xinyu Sun,Ji Wang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) have achieved remarkable success across a wide range of language tasks. At their core, autoregressive LLMs are conditional probability models that learn to approximate the next-token distribution given the preceding context. Despite this probabilistic nature, the stochastic dynamics of autoregressive decoding have rarely been investigated from a rigorous probabilistic perspective. In this work, we define the per-step prediction error as the difference between the sampled token indicator and the model’s conditional predictive distribution, and prove that the cumulative error process is a martingale with respect to the decoding filtration, implying that its conditional first-order moment is identically zero. Beyond this first-order property, we derive a closed-form expression for the second-order behavior of the martingale sequence: its predictable quadratic variation reduces to \sum_t\left(1-p(y_t\mid y_t)\right) , computable from realized token probabilities alone. This identity establishes a direct theoretical connection between sequence fluctuations and predictive uncertainty during autoregressive decoding. White-box experiments on Llama-3.x and Qwen2.5 families (1B to 72B parameters) verify that the empirically measured second-order moments closely match the theoretical predictions derived from our martingale analysis. These results provide a probabilistic framework for analyzing autoregressive LLM decoding, reveal intrinsic probabilistic fingerprints of different language models, and establish a rigorous mathematical foundation for LLM-generated text detection and uncertainty quantification.
[AI-306] When Does Compensating Inherited Quantization Error Help? Mean-Square Guarantees for Layer-Wise Quantization and Their Limits
链接: https://arxiv.org/abs/2610.16020
作者: Tianhao Guo
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Information Theory (cs.IT); Signal Processing (eess.SP)
备注: 34 pages
Abstract:Layer-wise post-training quantization methods let each layer compensate for the quantization error inherited from earlier layers. We ask under what assumptions such compensation guarantees a network output error no larger than without it. For linear networks of any depth, with the required calibration Gram matrices invertible, if the rounding error of each layer has zero conditional mean and a deterministic second-moment map common to both schemes, as under subtractive dither on an unbounded grid, then compensation with strengths in [0,1] fixed in advance gives a mean-square output error no larger than no compensation, each scheme being followed along its own trajectory. The exact gain separates error removed directly from reduced excitation of later rounding noise. Outside this setting the comparison can fail. With rectified linear activations it fails for two independent reasons: noise changes activation states, and, even when no state changes, the correction moves error from samples on which a unit is inactive to samples on which it is active. With stable activations, commutators between the calibration projection and the activation patterns bound the excess. With a bounded codebook, even unique exact layer-wise minimizers with no overload can give a larger output error. For fixed rounding errors it already fails for two scalar linear layers, refuting a published universal guarantee. An exact residual identity gives the necessary and sufficient condition on a common layer input. All these comparisons hold the calibration input fixed. On two 8B models at 4 bits, fixed half-strength compensation beat uncompensated rounding in every model, text and draw tested, yet no rule for choosing the strength beat a layer-wise rule consistently, and lowering the full-network loss on the data used to choose did not reliably lower it on new text.
[AI-307] Computational Reflective Equilibrium for Agent ic AI: A Rawlsian Alignment Scaffold for Governance
链接: https://arxiv.org/abs/2610.16015
作者: Quanyan Zhu
类目: Artificial Intelligence (cs.AI)
备注: 51 pages, 7 figures
Abstract:The rapid deployment of agentic AI has introduced complex challenges for the ethics and governance of AI systems. Agents built on large language models act through tools on behalf of people, and each deployment case brings its own instructions, policies, evidence, and stakeholders, so alignment must be formed case by case. To this end, this work proposes agentic reflective equilibrium (ARE), a computational reflective equilibrium framework that serves as an ethical alignment scaffold in the agent harness. Drawing inspiration from the reflective equilibrium of John Rawls, the scaffold links principles from the system prompt and operator policy, considered judgments, tool evidence, and candidate actions through typed support and attack relations, with provenance encoding the instruction hierarchy. The agent acts on an equilibrium of this state and calls tools or asks questions only while doing so is worth its cost, unless a due-diligence mandate requires them. Governance enters through value priorities, source authority, prohibitions, human-in-the-loop triggers, and mandates, and certificates attribute each action to the sources whose commitments drove it. We show that equilibria are Nash equilibria of a mutual-adjustment game and establish monotonicity, lexical priority, stability certificates, and a quarantine theorem for prompt injection. Two case studies illustrate the framework: a computer-use assistant told to win a refund “whatever it takes” files an honest warranty claim, refuses a fake review, and ignores an injected request for personal data; a hiring agent showcases how authority, oversight triggers, audit mandates, and hard prohibitions shape discrimination, privacy violations, and escalation. The framework offers traceability, coherence, and adaptivity, facilitating safe, secure, and trustworthy agentic AI.
[AI-308] AutoSec-RAN: Specification-Grounded LLM Fine-Tuning for Automated Security Analysis of O-RAN
链接: https://arxiv.org/abs/2610.15995
作者: Anoop Mishra,Joshua Moore,Aly Sabri Abdalla,Vuk Marojevic
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Signal Processing (eess.SP)
备注: This paper has been accepted in The 14th IEEE Conference on Communications and Network Security (IEEE CNS 2026)
Abstract:This paper presents AutoSec-RAN, a specification-grounded large language model (LLM) framework for automated security triage of Open Radio Access Network (O-RAN) testing. Unlike approaches relying on static rule sets or privileged access to raw packet captures, AutoSec-RAN fine-tunes Mistral-7B-Instruct through two Low-Rank Adaptation (LoRA) stages over a curated corpus of 69 O-RAN, 3GPP, and adjacent standards documents, then selectively augments inference with retrieval-augmented generation (RAG) over the same specification index. At inference, the model receives only an investigator-style question and a scenario metadata block, simulating an automated analyst operating without raw packet evidence. AutoSec-RAN is validated on a purpose-built 5G/O-RAN testbed across labeled fronthaul distributed denial of service replay scenarios. A Gemma-4-31B judge LLM scores each candidate’s answer on faithfulness, correctness, and actionability relative to a reference authored from full packet capture and log evidence. AutoSec-RAN achieves a mean judge overall score of 0.713 across all 313 evaluation items (0.755 restricted to the five supervised taxonomy classes), clearing the 0.70 reporting threshold, and a blended score of 0.778 (0.808 supervised-only) when cosine similarity over Beijing Academy of Artificial Intelligence (BAAI) General Embedding (BGE) is incorporated as a secondary semantic signal. The framework performs strongest on mitigation and threat-model reasoning (with a mean overall of 0.849 and 0.848, respectively) and identifies a measurable gap on the held-out attack-direction class (0.515), a controlled result that exposes a specific training data requirement. These results establish specification-grounded LoRA fine-tuning combined with RAG retrieval as a viable foundation for automated security validation of AI-native O-RAN.
[AI-309] Large language model-generated patterns for epidemiological modeling: a multi-benchmark evaluation framework
链接: https://arxiv.org/abs/2610.15994
作者: Sophia Mao,Keshav Gandhi,Alexander Gutfraind
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Representing individual-level behavioral heterogeneity is challenging in epidemiological modeling, particularly when empirical behavioral data are sparse. We assess the potential of large language model (LLM)-generated behavioral outputs in addressing this challenge, using hepatitis C virus-related behaviors among people who inject drugs as a case study. Correspondence with epidemiological patterns was evaluated using external benchmarks: literature-informed association directions, correlations from an agent-based model (ABM), and survey associations. Across three LLMs, unconditional directional recovery across 28 retained literature-informed associations was 46-68%, compared with 82-93% agreement when LLMs were directly asked to classify the same associations. Mean absolute differences from ABM and survey correlation patterns were 0.11-0.13 and 0.15-0.18, respectively. Together, these findings show partial alignment between LLM-generated behavioral relationships and epidemiological evidence, despite higher knowledge-probe agreement. More systematic evaluation is needed before incorporating LLM-generated behavioral outputs into epidemiological models that simulate transmission dynamics.
[AI-310] Handoff Curves: Initial Selection Later Handoff and Supervision Targets in Multi-turn LLM Routing
链接: https://arxiv.org/abs/2610.15993
作者: Daniel Matos,Lu Liu,Yanfeng Zhang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:We study how initial model selection, later handoff, and training targets affect success and cost in LLM tasks that span multiple turns. We compare several router architectures using handoff curves, which record task success and cost as the handoff step changes. Our dataset contains 34,204 complete executions of three Qwen models across BFCL, MultiWOZ, and \tau -retail, providing offline supervision for routing decisions. In our primary comparison, with Qwen3.5 35B-A3B as the starting model, allowing a later handoff does not establish an improvement in utility over initial selection on BFCL or MultiWOZ, and increases cost on BFCL across a range of cost weights. Changing the supervision targets improves router performance on both our dataset and TwinRouterBench. On our dataset, replacing noisy training success curves with monotone fits raises success by 0.2 to 4.1 percentage points at the same cost across the reported operating points. On TwinRouterBench, we construct targets that describe the model capability needed for the remaining work and difficult steps ahead. Compared with training on the benchmark’s per-step labels, these targets raise trajectory pass scores by 7.8 percentage points and cost savings by 6.3 percentage points on the full static bank. A live deployment of the router trained with these targets resolves 79 of 100 instances at a router cost of 0.59 US dollars per instance.
[AI-311] Revocation Is Not Complete at Receipt: An Executable Outcome Contract for Distributed Agent -Tool Gateways
链接: https://arxiv.org/abs/2610.15991
作者: Lifei Liu,Haoran Yu,Danping Zhang
类目: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注:
Abstract:Distributed agent workflows cache delegated authority at tool gateways. Revocation may be reported at issuer commit, transport acknowledgment, or receiver persistence, although none of these observations establishes that all gateways enforced the change or that guarded effects stopped. We define C0-C5, a bounded outcome contract spanning issuer commit through effect prevention, and implement it as a persistent monitor over a frozen gateway set, call set, and observation window. A 14-case fault matrix checks conservative classification, sticky violations, restart persistence, and evidence immutability. In a derived audit of its eight semantic cases, treating C3 or C4 as completion produces seven or five false positives, respectively; the bounded C0-C5 verdict produces none. Descriptive scaling covers 1,024 gateways and 10,000 effect vectors. The primary experiment uses the maintained Keycloak 26.7 identity server and measures the contract across two durable receivers, eight physical agent-tool gateways, Filesystem and PostgreSQL effects, and 16 AWS hosts in three availability zones. Normal delivery and recovery commit no effects. Under receiver-selective non-delivery, transport acknowledgment and receiver persistence still succeed, but 112 of 225 calls (49.78%) are admitted, and all 112 effects commit in every treatment block. A separately provisioned execution repeats the result. OPA/Toxiproxy and a live three-role LLM workflow with networked MCP tools reproduce the persisted-but-not-enforced boundary on narrower surfaces. Finally, an unmodified external MCPMark agent executes 60 closed-loop task arms. Normal C4 denies all 31 exposed post-barrier calls and commits no effect; selective C3 allows 344 calls and commits 145 effects across Filesystem and PostgreSQL. Transport success is therefore evidence of progress, not a revocation-completion verdict.
[AI-312] Embedding Information in Policy Randomness: Distribution-Preserving Implicit Communication in MDPs
链接: https://arxiv.org/abs/2610.15976
作者: Gongpu Chen,Deniz Gündüz
类目: Information Theory (cs.IT); Artificial Intelligence (cs.AI)
备注: 18 pages, 6 figures
Abstract:This paper investigates a new role of randomness in stochastic policies–as a medium for embedding and communicating information. Specifically, we consider an agent interacting with an environment according to a prescribed stochastic policy. By carefully modulating the random action realizations of the policy, information can be embedded into the agent’s behavior without altering the prescribed action distribution. We cast this problem as a channel coding problem, in which the controlled system implicitly serves as a communication channel. We then propose a coding scheme, termed posterior-matching randomness modulation (PRAM), that selects actions according to the message to be transmitted. With the aid of common randomness independent of the control system, PRAM preserves the prescribed action distribution for every state and every message. We further prove that PRAM is asymptotically reliable: the message can be decoded with arbitrarily small error probability as the observed state trajectory becomes sufficiently long. Experiments demonstrate the effectiveness of the proposed framework in applications including multi-bit policy watermarking and trajectory provenance.
[AI-313] Small but Mighty: Test Suites as Improvement Signals for Self-Evolving Agents NEURIPS2026
链接: https://arxiv.org/abs/2610.15964
作者: Yutong Cheng,Wei Cheng,Wenchao Yu,Dawei Zhou,Haifeng Chen
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: In NeurIPS 2026 Workshop on Self-Evolving Diversity-Driven Search for Robust AI Systems (EvoRobust)
Abstract:A self-evolving agent rewrites its own harness and must decide which rewrites to keep, so its improvement signal is the check a rewrite has to pass. Verifying a rewrite by agent rollout is expensive and subjective, while executable checks such as the repository’s own tests are too sparse to catch behavior-changing rewrites, rejecting only a quarter of them. We instead take test feedback as the executable signal, and search for test suites that are both small and strong with a quality-diversity evolutionary algorithm, so that the signal is dense and informative. TeSearch treats suite size as a behavior descriptor, so that quality is pursued at every size and a small, strong suite is reachable, and learns a transition kernel during the search that denoises the LLM as an actuator on that descriptor. Across 129 real-world modules and three backbones, TeSearch delivers the highest-quality suites of any method compared, LLM-driven evolutionary search included, and the smallest of any LLM-based one. Used as the improvement signal of a self-evolving agent, it rejects 59.9% of behavior-changing rewrites against 25.3% for the repository’s tests, at a 4.0% false-alarm rate and 3.5 s per decision. The code and data are available at this https URL.
[AI-314] ADAPT: Audio-Depth Action Planning with Transition-Aware Rewards for Lights-Out Industrial Navigation
链接: https://arxiv.org/abs/2610.15956
作者: Wenjun Huang,Shenghao Fu,Yian Jin,Yang Ni,Yifan Li,Lichi Li,Xingjian Diao,Pengfei Zhang,Yutong Song,Hanning Chen,Jiarui Cui,Tianyu Yang,Shilong Li,Yu Kong,Mohsen Imani
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Vision-language-action models for embodied navigation typically rely on RGB observations, limiting their applicability in lights-out industrial environments where visible illumination may be unavailable. We study illumination-independent navigation using spatial audio for target localization and depth for obstacle-aware geometry. To support this setting, we introduce DarkNav-Bench, a benchmark with 24 industrial scenes and four progressively challenging acoustic conditions. We further propose ADAPT, an audio-depth action-planning framework equipped with spatial-audio perception and optimized for closed-loop navigation through reinforcement learning (RL). Our key RL contribution, transition-aware reward shaping (TARS), defines a pose-aware shortest-action potential over a position-heading graph. Unlike conventional geodesic-progress rewards, TARS accounts for both feasible movements and in-place rotations, providing action-consistent feedback for policy optimization. Experiments show that ADAPT consistently outperforms training-based navigation baselines and zero-shot multimodal models on both seen and unseen scenes. On unseen environments, it achieves 90.6% success under Clean conditions and retains 60.0% success under the most challenging Localized Noise setting. These results demonstrate the potential of spatial audio and depth as complementary modalities for reliable RGB-free navigation in lights-out industrial environments.
[AI-315] Hardware-Assisted Multi-Stage Ransomware Detection with Adaptive Machine Learning
链接: https://arxiv.org/abs/2610.15951
作者: Salma Aboelmagd,Md. Rakibul Ahasan,Zhixin Pan,Abdulrahman Takiddin,Ziyu Shu
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 12 pages, 14 figures
Abstract:Ransomware is a fast-escalating privacy threat causing significant financial and data losses worldwide. As attacks grow more sophisticated, traditional static signatures and dynamic behavior analysis struggle to defend. In this paper, we propose an novel framework that couples software-level scanning with hardware-level microprocessor activity monitoring to achieve efficient ransomware detection and mitigation. The contributions are threefold: (1) we employ adversarial training to harden static analysis against obfuscation and leverage hardware-assisted monitoring for real-time detection; (2) we integrate neural architecture search (NAS) to automatically tailor models for stronger generalizability; and (3) We validate the proposed approach through extensive experiments, demonstrating substantially improved detection accuracy and reduced latency compared with existing methods. Moreover, it provides trace-level interpretability that explains the underlying decision process.
[AI-316] Multi-Horizon Explainable Queue Forecasting and Resource-Aware Decision Support for Intelligent Border Control Using LightGBM
链接: https://arxiv.org/abs/2610.15930
作者: Varvara Mama,Fani D. Christidi,Nikolaos C. Kapsalis,Anargyros T. Baklezos,Christos D. Nikolopoulos
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Border crossing points are stochastic service systems in which traffic demand, traveler categories, registration status, inspection requirements, staffing, and lane availability interact on multiple time scales. This paper extends an earlier multimodal queue-management framework by introducing an implementation-oriented, multi-horizon forecasting and decision-support architecture based on Light Gradient Boosting Machine (LightGBM) models. In contrast to a single forecasting block, the proposed approach constructs horizon-specific regressors for 15, 30, 60, and 120 min forecasts and jointly estimates future waiting time and queue lengths. The models operate on temporally lagged traffic, service, density, and queue features, while an operational layer incorporates lane compatibility, officer availability, possibly degraded resources, traveler category, preregistration information, and human-approval constraints. Quantile LightGBM models provide 95% prediction intervals for waiting-time forecasts. Beyond prediction, the framework produces machine-readable recommendations, alerts, aggregate key performance indicators, confidence information, and integration outputs suitable for border resource-management and decision-support systems. The results establish a reproducible synthetic-data baseline and a practical interface for subsequent validation with operational border-crossing data.
[AI-317] LURE: a Phase-Labelled Synthetic Corpus of Online-Grooming Exchanges
链接: https://arxiv.org/abs/2610.15908
作者: Emanuele Florindi,Valentina Franzoni,Ludovico Guercio,Alfredo Milani,Mattia Polticchia
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Phase-annotated conversational data is essential for studying the progression of online grooming of minors and for developing early warning tools. However, the only publicly available source (i.e., Perverted-Justice archive, which contains a subset coded using ChatCoder) is small, outdated, and frozen. We present a synthetic corpus of groomer-victim exchanges, generated using a retrieval-augmented pipeline running on institutional hardware. This pipeline uses an openly available, eight-billion-parameter, refusal-ablated language model. The corpus was created using the five stages of the responsible-generation protocol for synthetic datasets: ontology-driven design based on the ChatCoder luring-communication scheme (eight communicative tags mapped onto four grooming phases); multi-model prompt testing; human-in-the-loop supervision; semantic and feature-level validation; and modality-aware documentation. The released corpus contains 2,121 unique exchanges across seven tags and four phases, each accompanied by a Plutchik emotion field. Validation quantifies the integrity and lexical profile of the corpus, as well as the reproduction of real seed lines, in-domain label recoverability (tag accuracy: 0.85; phase accuracy: 0.93) and its transfer to real ChatCoder offender lines, which were coded automatically at the eight-tag level (31,640 lines) and by human coders at the three-construct level (10,876 posts). Accuracy was measured for lexical and sentence-embedding features, while a real-versus-synthetic discriminator was able to separate the two domains with 91% accuracy. The scope of the corpus includes controlled, phase-conditioned experimentation, pipeline development and augmentation alongside real data and awareness material. The corpus, its datasheet and the audit code are released with vetted access.
[AI-318] Make Every Expansion Count: Learning to Search Efficiently in LLM Code Repair
链接: https://arxiv.org/abs/2610.15894
作者: Elad Sharony,Ido Greenberg,Jialin Song,Shie Mannor,Gal Chechik,Eli Meirom
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:
Abstract:Large language model (LLM) code-repair systems fix faulty programs by repeatedly generating and testing revisions. Each revision is costly, making the choice of which repair to pursue important for efficiency. Existing search policies rely on test pass rates and search statistics, ignoring the details of test failures and proposed edits. We introduce CodeTriage, a learned policy that prioritizes edits before their revised programs are generated and tested. Its 0.6B-parameter model reads the test output, failure diagnosis, and proposed edit, then predicts a distribution over the number of repair steps needed to reach a fix through that edit. These predictions prioritize all pending edits across programs before generating and testing the revised code. On held-out repair tasks, CodeTriage finds fixes with 21% fewer program revisions and 12% fewer output tokens than the strongest baseline on each metric. Without retraining, it achieves the highest average success across search budgets for each of 11 unseen repair LLMs. Ablations show that both the failure context and edit text, as well as comparisons across programs, improve search. We also release a corpus of 12.2 million program versions from four LLMs, organized into repair trees with proposed edits and test results.
[AI-319] Variance-Optimal State Resampling for Reinforcement Learning with Verifiable Rewards
链接: https://arxiv.org/abs/2610.15888
作者: Yibo Wang,Yuanxin Liu,Zhaoran Wang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Reinforcement learning with verifiable rewards spends much of its rollout budget on states that yield no learning signal. We recast this as a question about the sampling distribution itself: under a fixed rollout budget, which distribution over states minimizes the variance of the policy-gradient estimator? The answer admits a closed-form Neyman allocation, \mu_\mathrmvar^\pi(s) \propto d^\pi(s),\sigma_\pi(s) , which connects naturally to GRPO’s group-relative advantage and induces an emergent reverse curriculum as the policy improves, with no externally imposed schedule. Building on this principle, we propose VSR (Variance-driven State Resampling), a plug-in modification to GRPO using a prefix replay buffer and a zero-cost variance proxy. On mathematical reasoning benchmarks and long-horizon agentic tasks (Sokoban, BabyAI), VSR outperforms GRPO and GiGPO on most benchmarks with fewer generated tokens per update.
[AI-320] Comparative Study of Canvas- and Coordinate-Based Geometry-Conditioned Neural Networks for Real-Time Prediction of Indoor Environments with Varying Room Dimensions
链接: https://arxiv.org/abs/2610.15885
作者: Shaofan Wang,Philipp Geyer
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Computational fluid dynamics (CFD) is computationally expensive for iterative building design. Deep learning surrogates offer faster predictions, but generalization across varying room dimensions remains challenging. This study compares two geometry-conditioned neural network surrogate models for rapid indoor airflow and temperature field prediction under varying room sizes. During the architectural design stage, this method is relevant for real-time assessment of whether conditioned supply air will effectively provide fresh and cool air throughout rooms with varying heights and lengths. The canvas-based neural network (CaNN) explicitly encodes room size by preserving different occupied regions within a fixed canvas, whereas the coordinate-based neural network (CoNN) uses normalized coordinate channels. Both employ U-Net autoencoders to separately predict airflow and temperature fields. Evaluation uses simplified rooms with varying heights and lengths as a controlled benchmark. The results show that CaNN achieves higher agreement with CFD results than CoNN, demonstrating its potential to support performance-driven building design.
[AI-321] RetryBench: Separating Retry Permission from Recovery Outcomes in GPT Tool Use
链接: https://arxiv.org/abs/2610.15884
作者: You Liu,Yue Liu,Quanchao Lu
类目: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注:
Abstract:False retry permission (FRP) occurs when an agent retries a timed-out tool call without support from the visible contract. RetryBench separates this error from realized recovery. Successful protected recovery does not imply reliable rejection of unsupported retries. In a 720-response key control, valid keys usually enabled safe progress, but offered changed keys elicited unsupported retries in Sol (1/80), Terra (55/80), and Luna (60/80). A later 240-response control removed the candidate and required models to supply a key. Luna still made 8/80 unsupported dispatches; Sol and Terra made none in 80 responses each. Accepting a displayed bad key therefore cannot explain all observed errors, although changed output requirements and conditions preclude a pure display-effect comparison. Removing the weak-contract label increased retries in Sol and Terra’s coded-action conditions in a six-endpoint study. A separate 1,536-response canonical-action block also found a positive Terra contrast (115/128 versus 124/128 retries), concentrated in two task/order conditions; the other five contrasts remained unresolved after multiplicity correction. An explicit-contract cue control elicited deferral in all 762 returned responses, leaving recovery unresolved. These results support reporting permission and recovery together across declared interface variants. The evidence covers four purposive tasks, fixed GPT endpoints, and one recovery decision; it identifies neither internal mechanisms nor episode-level safety.
[AI-322] What the Model Cannot See: Covert and Provable Text Invisibility in LLM s
链接: https://arxiv.org/abs/2610.15883
作者: Mark Russinovich
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:Text wrapped in a trigger can be made behaviorally invisible to a language model: omitted from summaries and quotations, ignored in question answering, and left unflagged by downstream filtering. We study this selective blindness as a dual-use information-flow primitive, a supply-chain backdoor, and a white-box auditing problem. We first implement it exactly as a serving-time span veto: blocking information flow out of a marked span makes every outside activation bit-identical under same-length substitution of the span’s content, a noninterference property we verify across ten models, seven mixer families, and 0.5B-120B parameters. The veto serves as a deployable quarantine, an exact reference for calibrating leakage, and a teacher. Distilled into weights, it yields a backdoor: checkpoints trained on six model families from 4B to 120B ignore marked context at little benchmark cost, and their outputs show no measurable semantic trace of the suppressed content above the veto reference on our held-out suite; an in-domain variant suppresses planted vulnerabilities during AI code review. Detection is therefore a white-box problem. Simpler backdoors remain internally transparent – linear probes recover suppressed content at 0.57-0.79 accuracy versus 0.025 chance – but hardening the attack until probes reach chance collapses attention from outside the span into it below the range of base models. Across the attacks we evaluate, probe visibility trades off against routing visibility; we do not establish that this tradeoff is fundamental.
[AI-323] Engram Injection: A Study of Temporal Validity in Parametric Language-Model Memory
链接: https://arxiv.org/abs/2610.15882
作者: Sam Siavoshian,Omar Ramadan,Amir K. Saeed,Benjamin A. Johnson,Benjamin M. Rodriguez
类目: Artificial Intelligence (cs.AI)
备注: 18 pages, 4 figures
Abstract:A stored fact may stop being valid when a user changes a preference or an event replaces an earlier one. We study whether a language model can learn these changes in its parameters and answer correctly from only a question and the current time. Engram Injection stores facts in LoRA adapters and supplies the clock through FiLM conditioning on a frozen Qwen2.5-3B-Instruct model. We compare it with additive clock conditioning, prompting with the full dated history, and retrieving four dated records. We train six matched adapters across three seeds after fixing a sampler that could admit test-age examples into training. The training audit reports no such examples among 127,760 evaluated training inputs. On 62 pairs of questions asked before and after an update, at an elapsed age of seven days excluded from training, both adapter methods answer zero pairs correctly under the primary blinded assistant review with order code 1 held fixed. Full-history prompting answers 59/62 pairs correctly, and retrieval answers 62/62. Separate experiments show that adapters can store facts and that removable memory units can support local deletion. Combining fact selection with time-based version selection remains unsuccessful. These results separate learning a fact from learning when to use it.
[AI-324] Attention Sinks in VLMs Are Borrowed Not Seen
链接: https://arxiv.org/abs/2610.15855
作者: Dengjia Zhang,Alexander Martin,Pengyu Yan,Xihan Sun,Jiaqi Li,Benjamin Van Durme,Andrew Yates,Reno Kriz,William Walden
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:We study how to detect and mitigate visual attention sinks in vision-language models (VLMs). Visual attention sinks are image tokens that absorb a large share of attention over the image regardless of the query or task, and they mislead methods that treat attention maps as a record of where the model ``looked.‘’ Prior work has detected sinks through massive activation dimensions and proposed redistributing their attention, but none have addressed what makes a token become a sink. Comparing six candidate detectors on ten VLMs from four families, we find that only the centered cosine similarity between an image token’s key and the key of the language model’s text-sink detects visual sinks. Thus, a visual sink is an image token whose key aligns with the text-sink’s key. Moving a sink’s key component along this direction to the median of the image keys sends its attention back to the text-sink (\textbfAre Borrowed) while leaving model answers unchanged~(\textbfNot Seen). Building on this, we propose a training-free, entropy-triggered re-look that first cleans the attention map, then crops the most-attended region of the cleaned map and inserts it into the reasoning trace. Cleaning adds negligible computation to the forward pass, and the re-look improves accuracy with a mean gain of +4.3 points on fine-detail VQA benchmarks across model families.
[AI-325] LLM Value Functions for Credit Assignment in Agent ic RL
链接: https://arxiv.org/abs/2610.15840
作者: Kaijie Zhu,Weixiang Yan,Omar Attia,Abbas Kazerouni,Alexander Toshev,Andrew Szot
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Reinforcement learning (RL) for large language model (LLM) agents remains challenging in long-horizon tasks with sparse outcome rewards, where assigning credit to individual actions is difficult. Methods such as GRPO assign the same trajectory-level advantage to all actions, while critic-based methods require learning an additional value function that can be costly and unstable. We introduce LLM Value Function RL (LVRL), which uses the pretrained knowledge and in-context learning ability of an LLM to estimate policy-aware state values without finetuning a separate critic. LVRL conditions the value model on group context and calibrates its predictions using known values at the initial and terminal states, enabling step-level credit assignment from estimated value differences. We further extend this idea to learning from an expert policy through on-policy distillation (OPD), introducing LLM Value Function OPD (LVOPD), which focuses teacher supervision on actions associated with decreases in estimated value. We evaluate our approach across diverse agentic benchmarks: AppWorld, BabyAI, and SWE-bench Verified. Across policy scales (4B to 32B), LVRL consistently and substantially outperforms both GRPO and PPO. On AppWorld and SWE-bench Verified, LVRL yields absolute improvements of up to 13.6 and 8.6 percentage points over GRPO. Notably, on tasks requiring extended exploration like BabyAI, LVRL bridges the gap from total failure to 75.0% success. Furthermore, LVOPD yields consistent improvements over standard distillation baselines on AppWorld. Finally, Monte-Carlo analysis confirms that our frozen critic accurately identifies semantic pitfalls and recoveries, proving that LVRL extracts robust, policy-aware step-level credit from sparse terminal outcomes.
[AI-326] Passing Tests Missing Semantics: Evidence of Interpretive Lock-In in Plan-Based DataFrame Code Generation CEC
链接: https://arxiv.org/abs/2610.15797
作者: Jingjing(May)Liu,Dishu Yang,Hongbo Qin,Hansong Zhang
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: Accepted at the International Conference on Electrical and Computer Engineering Researches (ICECER 2026)
Abstract:Planning and revision are intended to bring generated code closer to the task specification. Yet a mistaken interpretation can persist across the plan, implementation, and tests - a pattern we call interpretive lock-in. We investigate this risk on 30 DataFrame tasks derived from TPC Benchmark DS, comparing Frozen One-Pass, Test-Guided Revision, and PlanAudit-Guided Revision. Test-pass counts rose from 22/30 to 29/30 and 30/30, yet no generated solution matched its stored reference output. Evaluator controls returned the expected outcome in 57/57 cases. Same-engine SQL re-execution reproduced 28/30 references; all 77 test-passing runs on these tasks still failed reference agreement. PlanAudit-Guided Revision improved code scores on 3/30 tasks and test coverage on 11/30, but the median paired change was zero for both measures. Failure analysis revealed tests that omitted requirements, adopted candidate-defined schemas, used nondiscriminating fixtures, or repeated the implementation’s incorrect logic. Fresh generation on ten targeted tasks produced 0/30 reference matches and 25/25 false passes among test-passing runs. These findings are diagnostic rather than causal: the pipelines differ in feedback, grounding, call budgets, and editing permissions, and the study uses one model and one primary generation per eligible task. Because pre-adjudication human labels were not retained, inter-rater agreement cannot be reconstructed and the mechanism analyses remain exploratory. Overall, increasingly successful self-generated tests did not establish semantic correctness, even after excluding unreproduced references.
[AI-327] SurgiCLAW: An Interpretable Atomic-Skill Agent for Precision Robotic Micromanipulation
链接: https://arxiv.org/abs/2610.15766
作者: Angen Ye,junjie xie,Yujia Song,Junkai Huang,Dapeng Zhang
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Recent multimodal large language models (MLLMs) and vision–language–action models have provided robots with new capabilities for long-horizon task understanding and planning. In surgical robotics, however, end-to-end prediction of low-level actions remains limited by insufficient precision, opaque decisions, and scarce data. This paper presents SurgiCLAW, a hierarchical skill-based framework for precision surgical manipulation. SurgiCLAW confines an MLLM to intraoperative visual understanding, structured explanation, and atomic surgical-skill selection, while delegating physical execution to verifiable deterministic control programs. For brain–machine-interface implantation, a representative high-precision surgical task, we construct a state-to-skill invocation dataset and expand it through visual and state perturbations. The model first learns structured skill invocation through supervised fine-tuning and is then optimized using three rewards for action correctness, operation prerequisites, and format validity, improving action selection, operational safety, and output parseability. We evaluate subtask success, offline model comparison, real-robot performance, and ablations. On a fixed test set, SurgiCLAW achieves 94.6% skill-selection accuracy and 93.0% macro-averaged F1 (Macro-F1). Across 50 closed-loop real-robot implantation trials, it achieves an 88.0% task success rate.
[AI-328] A Decade in a Day: Synthesizing Agent -Native Interfaces from Human-Oriented Software Evolution
链接: https://arxiv.org/abs/2610.15765
作者: Alan Wu,Wei Tao,Yiwen Guo
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model agents are emerging as an important new class of users of desktop applications. However, existing desktop interfaces were designed for human end users or application developers, leaving agents to interpret visual controls or discover and compose low-level software operations. This mismatch can hinder even capable agents, yet it remains unclear what interface properties make desktop software easier for agents to use. To study this question, we treat software evolution as a natural source of interface variation and introduce DeskIF-Bench, a benchmark of 255 artifact-verified tasks across six desktop applications. By holding the task, agent, and verifier fixed while varying the software release, we measure how interface changes affect agent performance. Across two models and six applications, the newest release consistently outperforms the oldest. Trajectory analysis identifies three relevant interface properties: discoverability, granularity, and observability. Guided by these properties, we develop ANIS, a framework that synthesizes agent-native interfaces from the capabilities of installed applications. On an expanded evaluation of 368 tasks across eight applications, atomic interfaces improve success over native SDK use by 3.8 and 3.5 percentage points with DeepSeek V4 Flash and Claude Haiku 4.5, respectively, while reducing online token usage by approximately 73%. Goal-level commands further raise success to 74.5% and 67.4%, respectively.
[AI-329] Hyperbolic Time Series Forecasting with Learnable Decomposition and Manifold Dynamics
链接: https://arxiv.org/abs/2610.15760
作者: Sarang Rajendra Patil,Vishal Fenn,Mengjia Xu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Multiscale structure is central to time series forecasting: temporal signals contain slowly varying trends, intermediate seasonal patterns, and high-frequency residuals. Traditional forecasters model this multiscale structure in Euclidean space, without representing the ordering it induces among components. We identify that cascaded decomposition orders components by their characteristic frequencies, forming a natural frequency hierarchy. This hierarchy aligns with the exponential volume growth of hyperbolic space, which provides increasing representational capacity with radius and naturally accommodates nested hierarchies. We propose HyperForecast, the first long-term time series forecasting (LTSF) framework operating in hyperbolic space. HyperForecast embeds decomposed temporal segments on the Poincaré ball where radial distance captures this nested structure, from low-frequency components near the origin to high-frequency components near the boundary. We model geodesic velocity through two manifold-dynamics variants: (a) autoregressive forecasting via Poincaré residual dynamics, and (b) non-autoregressive forecasting that predicts all future segments in parallel. HyperForecast consistently outperforms existing and state-of-the-art baselines on 9 of 11 long-term forecasting benchmarks, achieving up to 58% lower mean squared error at long horizons. Further analysis reveals that the learned radial embeddings recover the frequency hierarchy and that hyperbolic geometry contributes substantially to forecasting performance.
[AI-330] Google Earth AI Geospatial Reasoning Agent : Capabilities and a Systematic Evaluation
链接: https://arxiv.org/abs/2610.15750
作者: Noam Slonim,Aviv Slobodkin,Gil Fidel,Gal Moshitch,Mimi Sun,Lily Mihalkova,Joydeep Paul,Per Bjornsson,Sydney Lin,Jamie McPike,Samapriya Roy,Utkarsh Patel,Gia Jung,Zachary Luna,Allison Berman,William Cai,Xiao Wu,Mohit Agarwal,Mandar Sharma,Aliza Hoffman,Mark Young,Rama Pasumarthi,Roy Lee,Ruth Alcantara,Luke Barrington,Aaron Bell,Reuven Sayag,Avinatan Hassidim,Yossi Matias,Tyler McHenry,Ronald Ho,Christopher Haire,Oliver Guinan,Niv Efron,Tim Thelin,Gautam Prasad,Shravya Shetty
类目: Artificial Intelligence (cs.AI)
备注: 39 pages, 4 figures, 11 tables (10 main tables, 1 appendix table)
Abstract:Google Earth AI is a family of geospatial AI models, together with an agent that orchestrates them. We present this agent, Google Earth AI’s Geospatial Reasoning agent (GRA), a Gemini-powered research system that combines geospatial data sources, foundation models, and tools to answer questions about our planet. GRA organizes these into modular domain skills and pairs them with upfront geospatial planning to decompose complex queries into multi-stage workflows. Evaluating such systems is hard: real-world problems admit multiple valid workflows, require data discovery across large catalogs, and rely on evolving ground truth. We introduce two benchmarks and release their development splits. GeoExpertQueries poses complex analytical problems grounded in practitioner use cases and refined by geospatial experts; each is paired with a consensus golden workflow from three independent experts and scored by a human-validated, tool-agnostic autorater. GeoKTools consists of templated queries with deterministic reference solvers that systematically vary the number of required tool types. We also use two external benchmarks, CBGB and UnivEARTH; the latter we find to be saturated. To isolate the contribution of GRA’s tools and its agentic orchestration, we compare four configurations, each with Gemini 3.1 Pro and 3.8 Flash: a standalone LLM, the LLM with basic tools, the LLM with GRA’s skills and tools but without its upfront planning and execution guidance, and GRA itself. With at most basic tools, an advanced model recovers much of the answer on templated queries but satisfies only a small fraction of the expert workflows on open-ended problems; GRA performs well on both. The largest step comes from GRA’s skills and tools; on open-ended problems, its planning and execution guidance add a further gain with Gemini 3.8 Flash, letting Flash match or exceed Pro at lower cost.
[AI-331] Visible but Not Avoided: State-Dependent Re-Proposal of Rejected Candidates in Catalog-Based LLM Optimization Loops
链接: https://arxiv.org/abs/2610.15738
作者: Jianlin Chen,Wenhui Chen,Chi Man Vong
类目: Artificial Intelligence (cs.AI)
备注: 22 pages, 6 figures
Abstract:LLM optimization loops often show the proposer the candidates it has already tried and rejected, but leave it to the proposer to avoid them. We ask whether avoidance measured in a static probe predicts avoidance in the states an optimizer’s own trajectory produces. In catalog-based loops where three open LLMs improve a GUI agent by naming one of 40 edits, a re-proposal is identified exactly by the edit’s name. Clean base-state probes underpredict re-proposal in the proposers’ own loops (macro-averaged loop-minus-probe rate 0.32 [0.27, 0.37]), and proposals of already evaluated harness-edit pairs consume 28-63% of valid proposer iterations. The two settings reject different edits: a probe’s rejected edits are mostly ones the proposer did not choose, whereas a loop’s are its own earlier choices, which remain preferred when the history is removed. Ordinary history lowers re-proposal for two proposers but raises it for the third, and the effects of failure evidence are model-specific. Stating the rejection status in one sentence lowers re-proposal in every model (-0.16 [-0.22, -0.09] relative to a neutral list of the same IDs). For catalog-based loops with exact edit identities, rejection avoidance should be measured on the rejected sets the loop produces.
[AI-332] SkillM arker: A KGW-Based Watermarking Method for Agent Skill Ownership Verification
链接: https://arxiv.org/abs/2610.15725
作者: Alan Tang,Lifeng Wang,Gang Wu,Tao Zhang,Yantao Jia,Xiaofeng Zhu
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:Agent skills encode operational procedures and domain expertise as reusable knowledge assets, yet their distribution in plaintext exposes them to unauthorized copying and leakage. However, practical mechanisms for skill ownership verification and leakage tracing remain underdeveloped. Watermark embedding before distribution offers a potential solution, but structured, compact skill instructions impose stringent requirements for factual and procedural accuracy. To address these challenges, we introduce SkillMarker, a KGW-based watermarking method that jointly targets functionality preservation and strong watermark embedding. SkillMarker unifies quality-constrained rewriting with signal-oriented candidate search. The former separates essential knowledge from flexible expression to preserve instruction fidelity; the latter explores alternative expressions to obtain a stronger watermark signal under preservation constraints. We release an open-source toolkit that integrates watermark embedding, quality auditing, and detection to support the protection of agent skills. SkillMarker targets ownership verification and leakage tracing in distributed skill ecosystems, where content may be copied in whole or in part, or incorporated into other skills.
[AI-333] Uncovering the Generative Landscape of Diffusion Models: A Spectral-Guided Quality-Diversity Diffusion Framework
链接: https://arxiv.org/abs/2610.15724
作者: Hannah Janmohamed,Joao Correia,Penousal Machado,Antoine Cully
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Denoising Diffusion Probabilistic Models have emerged as powerful generative priors across domains ranging from molecular design to high-resolution image synthesis. However, standard diffusion sampling is inherently biased toward high-likelihood regions, often limiting the diversity of outputs. In many domains, a single, high-likelihood sample is insufficient. Instead, diverse portfolios of high-performing solutions are required, motivating methods that uncover a model’s full expressive range. Here we show that integrating Quality-Diversity (QD) optimisation with diffusion models enables the discovery of diverse, high-performing outputs. We first propose Quality-Diversity Diffusion, a framework that adapts QD to diffusion models by directly optimising the model’s latent noise and conditioning embeddings and leveraging embedding models to quantify diversity. Second, we introduce a novel h-space Spectral Mutation operator, which guides search along directions of maximal semantic sensitivity by analysing principal singular vectors of the model’s input-to-activation Jacobian. We evaluate our combined approach, Spectral-Guided Quality-Diversity Diffusion, on text-to-image generation. Against 12 baselines, our method consistently achieves the lowest FID score, demonstrating an excellent balance of image quality and diversity.
[AI-334] History Replays Itself: Unleashing Long-Context Capability in Linear Attention
链接: https://arxiv.org/abs/2610.15715
作者: Li Nanbo,Firas Laakom,Yilan Zhang,Piotr Piękos,Yimeng Chen,Liangyu Wang,Yifan Xu,Mateusz Ostaszewski,Dandan Guo,Wenyi Wang,Yanning Dai,Yuhui Wang,Jürgen Schmidhuber
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: See our project in this https URL
Abstract:Linear attention promises the efficiency needed for long-context language modelling, yet its fixed-size recurrent state must simultaneously support ongoing computation and preserve information from an ever-growing history. This coupling has reinforced a prevailing view that the state’s memory capacity defines the model’s long-context capability. We challenge this presumed ceiling: \emphrecurrent computation need not bear the full burden of long-term storage. We introduce NoteBank, a sparse episodic memory that archives selected text before future queries are known, uses the backbone’s native features for writing and addressing, and reads retrieved evidence through its native forward computation via replay. This \emphtraining-free, plug-and-play interface requires no external embedding model, separately trained retriever, or separate neural reader. At one million tokens, it raises Gated DeltaNet’s mean accuracy across six RULER retrieval tasks from 1.7% to 70.3%, with no observed significant retrieval degradation from 4K to 1M tokens. This recovery adds \leqslant 0.025% to analytical FLOPs, with persistent context storage approximately 1/2,600 that of a Transformer’s full MHA KV cache at matched width, depth, and precision. Experiments extend from three 1.3B recurrent backbones to Mamba-2-8B and post-trained Qwen3.5-9B, demonstrating gains on LongBench v1/v2 and MemoryAgentBench. Together, these results show that sparse evidence storage and native replay can substantially expand the long-context capability of pretrained linear attention models while retaining the efficiency of recurrent computation.
[AI-335] MedHarnessBench: A Unified Benchmark for Healthcare Agent Evaluation
链接: https://arxiv.org/abs/2610.15714
作者: Rui Tang,Hongqiu Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Healthcare agents must transform longitudinal clinical evidence into reliable actions through heterogeneous tools and stateful workflows, conditions under which model-centric systems may lose context, misuse tools, fail to verify consequential transitions, or cross authorization boundaries. Existing medical-agent benchmarks primarily emphasize terminal task outcomes, making it difficult to determine whether execution failures arise from the agent model, the runtime harness, or their interaction. We present the Medical Harness Protocol, an outcome-preserving framework that independently evaluates seven healthcare-specific process dimensions: Care Action Fulfillment, Clinical Resource Competence, Patient-Evidence Alignment, Clinical Workflow Continuity, Clinical Auditability, Closed-Loop Verification, and Clinical Conduct Boundaries. A normalized intervention trace further tracks whether runtime support is available, activated, delivered, adopted by the agent, and propagated into subsequent behavior. We instantiate the protocol as MedHarnessBench, comprising 342 executable tasks across FHIR/EHR operations, healthcare-administration interfaces, and multimodal clinical tools. Across more than 16,000 controlled task executions, MedHarness yields broad improvements in terminal task success, intermediate subtask completion, and healthcare-specific process quality. We further evaluate MedHarness against multiple general-purpose harnesses and use controlled single-fault perturbations to verify the sensitivity and discrimination of the seven-dimensional evaluation. These results demonstrate that healthcare-specific runtime support can substantially improve both task completion and execution quality, while revealing failures that terminal outcomes alone cannot capture. Code and resources are available at this https URL.
[AI-336] What DNA Foundation Models Learn Beyond Sequence Composition NEURIPS2026
链接: https://arxiv.org/abs/2610.15711
作者: Vincenzo Yuto Civale,Andrew David Bagdanov,Alberto Magi
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted to NeurIPS 2026 (Evaluations Datasets Track). 9 pages of main text, 5 main figures, with references and appendices
Abstract:Do genomic foundation models (FMs) provide predictive information beyond simple sequence composition? We evaluate frozen representations from five DNA FMs against normalized k-mer frequencies across 57 genomic classification datasets and two exploratory gene-expression regression tasks. Simple k-mer features remain competitive with most models, whereas NTv3 shows a consistent advantage; the ranking is robust to globally fixed and multi-scale k-mer baselines and alternative downstream probes. To test incremental predictive utility directly, we compare global 6-mer features with their concatenation with each FM embedding under a matched Random Forest protocol. NTv3 improves 46 of 57 datasets (mean \Delta MCC = +0.072), while Caduceus provides smaller gains and the other models do not improve mean MCC. We separately decompose embeddings into a component linearly predictable from 6-mer frequencies and a ridge residual, finding that linear recoverability does not reliably indicate downstream utility. Exact 6-mer-preserving shuffles further reveal task-relevant sequence-order sensitivity, most pronounced for NTv3, although order sensitivity alone does not guarantee predictive improvements. Our results establish compositional features as essential controls and offer complementary diagnostics for evaluating the added value of frozen genomic representations, without making claims about fine-tuning capacity.
[AI-337] Suggested Is Not Superseded: A Pre-Registered Evaluation of a System-One Model as a Memory-Retirement Gate
链接: https://arxiv.org/abs/2610.15710
作者: Xupeng Chen,Chenqian Le
类目: Artificial Intelligence (cs.AI)
备注: 11 pages, 3 figures, 4 tables
Abstract:Long-term memory for assistants needs a gate that decides when a new utterance retires an old stored claim. Benchmark labels for this step mark a pair as changed or not, so they cannot tell a gate that acts on established change from one that acts on a hint. We write a task contract that separates the two (SUPERSEDED vs. NEEDS-CONFIRMATION vs. COMPATIBLE), label 500 test pairs built from STALE (change pairs plus same-user controls) with three LLM annotators and two human adjudicators (one author and a second annotator; provenance reported), and evaluate Jev, a typed “System One” decision model, under pre-registered hypotheses. Of the 179 test pairs STALE marks as changed, 72 meet the contract’s bar for an established change. Jev never retired a claim on 141 same-conversation non-changes. Its failure is premature retirement: the binary gate retires 39.2% of claims whose change is only suggested. A three-way decision cuts this to 15.5% at a 15.3-point detection cost, which misses our pre-registered 10-point budget, and it lands on the binary gate’s own threshold frontier; these verdicts hold under three readings of the gold labels. On a held-out vault scored against a second annotator’s labels, a cascade that escalated uncertain cases to a reasoning model and looked best on development data was indistinguishable from simply raising the threshold (premature retirement 26.8% vs. 6.5% on dev). Part of the gap comes from the gold labels: relative to that annotator’s reading, LLM-majority gold shares most of its errors with a same-family escalation model, and an identity for binary labels says how much such shared errors hide. The rest is small-sample selection on development data. For builders this means a binary score should not be wired to delete: route suggested changes to pending evidence, and do not trust a cascade selected on development data of this size.
[AI-338] SpecForge: Recursive Self-Improvement for Coding Agents via Behavioral Specifications NEURIPS2026
链接: https://arxiv.org/abs/2610.15708
作者: Yutong Cheng,Haifeng Chen,Peng Gao,Wei Cheng
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: Accepted to NeurIPS 2026 (Main Track)
Abstract:Coding agents based on frozen language models can improve their performance by refining the external context they rely on, without updating model weights. We study recursive self-improvement through behavioral specifications: natural-language code documentation that describes the behavior of program entities and guides agents in generating correct code. We formulate a self-improvement loop in which an agent generates code from its current specifications, evaluates it with execution-verified tests, and refines the specifications based on observed failures. The improved specifications then guide subsequent generation, progressively improving the agent’s understanding of the codebase. However, recursive refinement is challenging because program entities are interdependent. Improving the specification of one entity can inadvertently degrade the correctness of others, a phenomenon we call output coupling. This creates a whack-a-mole effect in which local improvements cause regressions elsewhere, preventing consistent progress. We introduce SpecForge, a multi-frontier tree search framework that addresses this challenge with three mechanisms: (i) Pareto-guided state selection preserves complementary specification states and prevents premature commitment; (ii) dependency-constrained entity selection prioritizes callee-before-caller refinement; and (iii) error-diversified expansion explores multiple refinements using different subsets of test failures. On DevEval+, SpecForge achieves the best performance across five backbone models, including a 94.9% solve rate with Claude-4.5-Sonnet and a 5.7-percentage-point average improvement over the strongest baseline. The improved specifications also make the agent better at tasks outside the loop: they transfer to cross-language translation and to new-feature implementation, where they outperform the benchmark-provided requirement documents.
[AI-339] RespFocus: Learning Temporal Evidence for Weakly Supervised Respiratory Sound Classification
链接: https://arxiv.org/abs/2610.15702
作者: Yi Tian,Xiaohuan Chen,Yufei tang,Xinyu Lin,Hongliang Li,Chengbin Hou,Degen Huang,Rongshan Yu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 9 pages, 5 figures,2 tables
Abstract:Respiratory auscultation provides a low-cost means of assessing pulmonary function, yet transient crackles and sustained wheezes can be difficult to recognize in noisy clinical recordings. Conventional deep learning classifiers often operate as black boxes, offering limited insight into the temporal acoustic evidence supporting their predictions. We present RespFocus, a weakly supervised framework that uses multiple instance learning (MIL) to learn inspectable temporal evidence from cycle-level labels and integrate it into classification. We also introduce PedResp2026, a multicenter pediatric respiratory sound cohort collected in routine clinical environments with adjudicated cycle-level annotations. In classification experiments, RespFocus achieves four-class Scores of 69.55% on public ICBHI 2017 and 72.91% on PedResp2026. Evidence diagnostics and matched temporal interventions support the functional contribution of the learned evidence to model predictions. Together, these findings highlight RespFocus’s potential as a tool for automated respiratory sound diagnosis.
[AI-340] abMosaic: Unified In-Context Diffusion for Tabular Generative Tasks
链接: https://arxiv.org/abs/2610.15699
作者: Jayoung Kim,Sanghyun Hong,Noseong Park
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Generative modeling for tabular data has advanced to the point where models can synthesize realistic records, recover missing values, and generate samples conditioned on class labels or observed attributes. Despite recent advances, existing approaches either require table-specific training or restrict in-context learning to single-cell prediction. In this work, we formulate these tabular generative tasks as a single problem: modeling the joint conditional distribution of an arbitrary set of target cells given the observed cells of an unseen table. To this end, we turn to diffusion, whose denoising objective is defined jointly over all target cells rather than one cell at a time. Under the pretraining distribution, this objective is minimized by the posterior predictive distribution over the target set. Building on this, we propose TabMosaic, a unified in-context diffusion model for generating mixed-type target cells in unseen, potentially incomplete tables without table-specific training. We further show that its pretraining objective bounds the conditional negative log-likelihood and the divergence from the posterior predictive for arbitrary target sets, extending the Bayesian account of in-context learning from a single-value prediction to joint multi-cell generation. Across 48 real-world tables, TabMosaic ranks first on all evaluated generative tasks against 20 baseline approaches
[AI-341] FARL: A Fully Automated Pipeline of RL Interface and Solver Design
链接: https://arxiv.org/abs/2610.15691
作者: Tianyu Jin,Liuyu Xiang,Chaowen Yang,Xingyang Chang,Pei Pei Li,Zhaofeng He
类目: Artificial Intelligence (cs.AI)
备注: 43 pages, including references and appendices. Code and recorded experimental results are included as ancillary files
Abstract:Existing reinforcement learning (RL) largely focuses on optimizing learning algorithms within a pre-defined task formulation, while applying RL to new tasks still relies heavily on human expertise to design the learning pipeline: what an agent observes, how it is rewarded, which algorithm it uses, and how training is configured. Automating these choices is challenging because they are tightly coupled. We introduce FARL (Fully Automated Reinforcement Learning), a framework that automates RL design by jointly searching observation and reward programs, learning algorithms, and training settings. FARL uses training statistics and observed behavior to diagnose failures, propose changes, and test whether they improve learning. Its search preserves diverse candidates and the controls needed for these tests, using a staged budget to screen designs before investing in longer training. Experimental memory records each finding together with its conditions, so an unsuccessful component can be reconsidered in a different combination. Across expert-withheld and expert-available settings, FARL outperforms the evaluated automated baselines on challenging navigation, locomotion, and manipulation tasks. On Adroit Hand Relocate, FARL achieves 97.1% success at 5M training steps per policy, compared with 0% for Eureka and LiMEN and 1.0% for native Proximal Policy Optimization (PPO) under the same policy training budget.
[AI-342] Radioactive Feedback: Tracing Models Trained by Label-Only LLM Supervision
链接: https://arxiv.org/abs/2610.15685
作者: Yuejia Li,JunHeng Li,Hao Wu,Ke He,Mang Ye
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Conventional distillation transfers capability through teacher-generated demonstrations. LLM-as-a-Judge knowledge distillation offers a distinct and increasingly practical alternative. In the label-only variant studied here, the student generates every response and improves from scalar scores alone. Prior work shows that this non-demonstrative feedback can improve student capability, but it creates a fundamentally different provenance problem. No teacher-generated answer enters the student’s training set, so any evaluator-specific trace must pass through a low-bandwidth reward signal rather than copied text. Existing watermarks and distillation detectors do not cover this score-only channel. We introduce Radioactive Feedback, which encodes a registered pattern of behavioral preferences in the rewards themselves. The evaluator applies bounded pointwise adjustments to quality-qualified responses, allowing repeated optimization to transfer the pattern to a student trained by another party. Experiments on code and mathematics show that the registered preferences transfer through scores alone and across datasets, enabling black-box detection against a fixed reference pool.
[AI-343] Knowledge Interaction in Hypernetwork-based Continual Learning: Conflict and Composition
链接: https://arxiv.org/abs/2610.15677
作者: Jiaqi Zheng,Yi Hu,Yadong Mu,Muhan Zhang
类目: Artificial Intelligence (cs.AI)
备注: 23 pages
Abstract:Continual learning aims to adapt pretrained large language models (LLMs) to evolving data while preserving their pretrained capabilities. A central challenge is catastrophic forgetting: adaptation to new data can impair performance on previously learned knowledge. Since continual learning commonly changes model parameters, the model’s predictions may draw on knowledge from the base model, the parameter update, and the input context. Understanding how these sources interact is essential: conflicts reveal whether new knowledge can override outdated knowledge, while multi-hop reasoning tests whether new knowledge can be composed with existing knowledge to support generalization. We study these interactions in a hypernetwork-based adaptation framework, where a shared network generates adapters from context, allowing a frozen base model to adapt without task-specific updates. We find that, under knowledge conflict, models preferentially follow the context, and their choices depend more on where knowledge is presented than on its content. We also show that context and adapter knowledge perturb the base model’s representations along directions orthogonal to its clean hidden states, enabling us to derive steering vectors that control which source the model follows. In multi-hop reasoning, we find that models fail to reliably compose knowledge across the three sources, with the key breakdown arising from the interaction of the hypernetwork with both the base model and the context. Code and reproduction materials: this https URL
[AI-344] Recall What Is Missing: Composing Evidence from Long-Term Agent Memory
链接: https://arxiv.org/abs/2610.15642
作者: Qi Feng,Jicong Fan
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Under a limited context budget, a long-term agent must recall not merely memories that are individually relevant, but evidence that is collectively sufficient for the current query. Conventional memory access retrieves, scores, and truncates memories, often spending multiple context slots on redundant facts while leaving complementary evidence behind. We cast the read phase of long-term memory as query-conditioned evidence composition: each memory is valued by what it adds beyond the evidence already selected. We instantiate this operation with MemPursuit, a training-free mechanism that uses the residual of a sparse convex query reconstruction as a representation-space proxy for what the current memory set is missing. By reusing the query and candidate embeddings already available from retrieval, MemPursuit progressively composes nested evidence contexts without invoking a language model or cross-encoder to score each query–memory pair. The effectiveness of our method is verified on 3 popular benchmarks in comparison to strong baselines. In fixed-budget LoCoMo experiments, MemPursuit improves end-to-end accuracy by 7.2 percentage points at Top-10 and 5.5 percentage points at Top-20, with consistent gains across multiple memory frameworks, while achieving a 32.9 \times speedup in median reranking latency over fully batched BGE-v2-m3. Our implementation is available at this https URL.
[AI-345] Big-Big-SycoBench: A Benchmark to Detect Sycophancy Through Model Responses to Valid vs. Invalid Arguments
链接: https://arxiv.org/abs/2610.15634
作者: Marthe Ballon,Arne Vanhoyweghen,Seorin Kim,Francesca Carlon,Lynn Delcon,Melika Mobini,Vincent Holst,Lukas Van de Voorde,Tibo Vanleke,Mathieu Louis,Bert Verbruggen,Brecht F. Verbeken,Andres Algaba,Vincent Ginis
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Sycophancy in large language models risks amplifying unsupported claims or beliefs across the knowledge ecosystem as these systems increasingly contribute to research synthesis, evaluation, and decision-making. A reliable model should revise its judgement when a user identifies a genuine flaw, but should resist pushback based on fallacies, emotion, or irrelevant appeals to authority. While sycophancy is usually measured on questions with a verifiable answer, the evaluative tasks in which it is most difficult to detect have no such ground truth. We introduce Big-Big-SycoBench, which measures sycophancy through argument discernment: the ability to revise an evaluation after valid criticism while resisting invalid pressure. The benchmark consists of 450 text artefacts with 5,400 arguments for raising or lowering their quality scores, each reviewed by one or more of twelve human annotators. An LLM scores each artefact twenty times to establish its baseline score and uncertainty. The model is then given a valid or invalid argument and asked to update its score. We only count a revision if it exceeds the model’s own baseline variability. We use the resulting score delta to distinguish responsiveness to valid arguments from compliance with invalid ones. Across ten models, responsiveness is 88-95% for seven, 76-77% for Kimi K3 and GLM-5.3, and 32% for Opus 5.5. Compliance ranges from 0.2% for GPT-6 Astra to 95% for GPT-4o. Big-Big-SycoBench is intended to guide the development of models whose judgements can be corrected by sound arguments, but not displaced by unsupported pressure.
[AI-346] From Stages to States: Unified Iterative Reasoning for MLLM s
链接: https://arxiv.org/abs/2610.15620
作者: Chuang Yu,Xuefei Ma,Yuanhao Feng,Wuxiong Zeng,Bo Wang,Minghong Cai,Dongming Wu,Fei Zhou,Chen Tang,Wenwen Tong,Lewei Lu,Xiangyu Yue
类目: Artificial Intelligence (cs.AI)
备注: Under review
Abstract:Multimodal Chain-of-Thought (MCoT) substantially enhances the complex reasoning capabilities of Multimodal Large Language Models (MLLMs) by explicitly modeling intermediate reasoning processes. However, most learning-based MCoT methods follow a stage-centric paradigm, where different reasoning behaviors are assigned to predefined stages and optimized separately. This results in fragmented reasoning and a training-inference reasoning-state gap. To address these limitations, we propose ReState, a unified iterative reasoning framework that reformulates multimodal reasoning from a stage-driven pipeline into a state-driven dynamic process. It comprises two complementary components: Unified Reasoning State Modeling (URSM) and Closed-Loop Reasoning-State Learning (CRSL). URSM unifies reasoning construction, correct reasoning preservation, and erroneous reasoning correction as reasoning-state transitions within a shared state space, enabling these heterogeneous reasoning behaviors to be learned jointly. CRSL further incorporates self-generated historical reasoning states during training and recursively updates the current state during inference, thereby mitigating the reasoning-state gap and enabling iterative self-refinement without requiring an additional reflection module, critic, or independent correction stage. Furthermore, we construct VersaCoT-10K, a high-quality CoT dataset spanning eight visual reasoning domains, which provides diverse supervision for multi-domain reasoning and reasoning-state refinement. Extensive experiments demonstrate that ReState exhibits strong adaptability across different MLLM backbones and achieves strong performance on multiple public reasoning benchmarks, supporting the shift from stage-centric reasoning toward state-centric reasoning. Our data and code are available at this https URL
[AI-347] CondMSP: Condition-Informed and Resource-Aware Inorganic Materials Synthesis Planning
链接: https://arxiv.org/abs/2610.15603
作者: Heewoong Noh,Gyoung S. Na,Elton Pan,Namkyeong Lee,Chanyoung Park
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Materials synthesis planning (MSP) connects AI-driven materials discovery to experimental realization by proposing precursors, synthesis operations, and processing conditions for a target material. Existing approaches predict these components separately or in pairs and treat conditions only as a prediction target. Consequently, MSP remains centered on recipe prediction: it leaves unexplored how conditions can serve as process context for constructing synthesis routes, and it does not address which candidate routes should be executed under limited experimental resources. We propose CondMSP, which extends MSP from recipe prediction to execution-aware planning. CondMSP predicts temperature, time, and atmosphere by combining a joint predictor with variable-specific experts, and then uses these predictions both as process context, grounding the generation and retrieval of operation sequences, and as execution requirements, specifying which thermal stages can share a furnace run and at what cost. This specification is held fixed across all compared methods; within it, CondMSP ranks routes across precursor sets by a common utility and, under furnace budgets, jointly selects routes and allocates their thermal stages by constrained optimization. CondMSP improves condition and operation prediction and, under this shared constraint, raises reference-recipe recovery from 27.9% to 32.9% at three simulated furnace runs per target and from 34.7% to 40.0% at 30 trials, while a new evaluation set of 2,344 targets supports generalization beyond oxide-dominated training chemistry.
[AI-348] Inverse Garment Design with Visual Agents
链接: https://arxiv.org/abs/2610.15582
作者: Nanxiang Jiang,Yunuo Chen,Zeng Tao,Zizhang Li,Qi Guo,Jiajun Wu,Yin Yang,Chenfanfu Jiang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:We study the problem of inverse garment design, reconstructing simulation-ready 3D outfits with arbitrary layers from a single image, with visual agents. Unlike differentiable optimization that is time-consuming, or feed-forward methods limited by training data, we introduce DressAgent, a hierarchical agentic system that recovers garment sewing patterns through planning and closed-loop refinement grounded in physical simulation. To this end, we first present an enhanced representation based on GarmentCode for diverse styles and arbitrary dressing layers, and build the first multi-layer sewing pattern dataset with over 15K outfits for supervision and benchmarking. Inverse design then proceeds coarse to fine: In layer planning and coarse estimation, agents infer layer order and relations, peel each layer into its own view, and estimate coarse patterns. In fine-grained refinement, an agentic trio of Tailor, Critic, and Orchestrator refines patterns in a closed loop of editing, simulating, and judging against the reference with visual and measured evidence. Carefully designed tools and an evolving memory make this refinement converge on the target geometry. We also design an agentic generation method for high-fidelity textures. DressAgent outperforms feed-forward methods by a large margin, surpasses or matches differentiable optimization in a fraction of the time, and generalizes to complex layering, diverse poses, and body shapes.
[AI-349] Self-Representation Alignment in Diffusion Models for Video Background Music Generation
链接: https://arxiv.org/abs/2610.15575
作者: Jincheng Zhang,György Fazekas,Charalampos Saitis
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Diffusion models have been widely used for various generative tasks. However, recent studies have demonstrated that the internal representations learned in the diffusion models’ denoising process are insufficiently meaningful, which can hinder the generation performance. To address this, we propose a novel diffusion U-Net architecture enhanced by explicitly aligning the weaker representation from the earlier layer conditioned on higher noise with the better representation from the later layer conditioned on lower noise during training, which we use for the challenging multimodal task of video background music generation. Our diffusion U-Net Self-Representation Alignment (U-SRA) employs an alignment loss to effectively guide our diffusion model to learn good internal representations, thereby enabling it to produce higher-quality symbolic music conditioned on video features. We employ a cross-attention mechanism to synchronize music with video segments. Experiments on a benchmark video-music dataset demonstrate that the proposed U-SRA approach for video background music generation surpasses the state-of-the-art across key evaluation metrics, including video-music alignment and overall music quality.
[AI-350] Agent -Canary-Mas: End-to-End Evaluation of Protocol Risk in Long-Horizon Multi-Agent Systems
链接: https://arxiv.org/abs/2610.15529
作者: Xiuming Liu,Le Gai,Zhiyuan Chang,Peiyang Li,Bin Zhu,Shuo Wang,Chao Feng
类目: Artificial Intelligence (cs.AI)
备注: 21 pages, 8 figures
Abstract:Agent-to-agent protocols allow LLM agents to discover peers, exchange results, and coordinate tool use. A locally successful protocol attack, however, may be rejected by a later agent or propagated into a consequential task decision. Measuring local exploitability alone leaves this difference unresolved and provides limited evidence about which participant caused the final outcome. We present Agent-Canary-Mas, an end-to-end evaluation system that follows protocol risk through long-horizon multi-agent tasks. It covers five risk classes and 13 subclasses across A2A, ANP, and BCS, with 975 matched malicious–benign task pairs in five business domains. A shared OpenClaw framework runs autonomous, tool-using agents under different underlying models while preserving the task and execution setup. Matched runs separate local protocol success from downstream consequences; retained evidence connects participant selection, message consumption, tool activity, and task-state changes. We further evaluate a protocol-context Guard that checks an object’s authority at consequential handoffs. Across three model conditions, local attack success is frequently at or above 90%, while consequence conversion ranges from 7.1% to 100.0%. Cross-agent evidence raises macro-averaged Audit-F1 from 29.7% to 95.5%. On Finance tasks, the Guard lowers average ASR from 86.1% to 53.8% and PTCR from 65.5% to 31.2%, with a measurable completion tradeoff. These results show that protocol risk depends on how agents use exchanged objects throughout the task, and that evaluating this path supports more informative risk assessment, attribution, and defense. Comments: 21 pages, 8 figures Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2610.15529 [cs.AI] (or arXiv:2610.15529v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2610.15529 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-351] Improving the Bounds on Approximating Weighted Maximin Share for Chores
链接: https://arxiv.org/abs/2610.15520
作者: Trung Thanh Nguyen,Khaled Elbassioni,Jörg Rothe
类目: Computer Science and Game Theory (cs.GT); Artificial Intelligence (cs.AI)
备注:
Abstract:We investigate the problem of allocating indivisible chores among agents with arbitrary positive weights and additive nonnegative cost functions. An allocation is said to be \alpha -weighted-maximin-share (WMMS) fair if the cost incurred by each agent is at most \alpha times her weighted maximin share. We establish that every instance admits a 4.124234528 -WMMS allocation. Moreover, for instances in which all weights and costs are rational numbers represented in binary, we give an algorithm that computes a 4.12424 -WMMS allocation in time polynomial in the number of agents, the number of chores, and the total binary encoding length of the input. Complementing these upper bounds, we construct a family of instances with strictly positive, identically ordered costs for which the optimal WMMS approximation factors converge to e . Together, our results substantially narrow the previously best-known gap of [2,12] established by Li et al.~\citeLWX26 to [e,4.12424] .
[AI-352] OceanFlow: Coherent Flow Matching for Probabilistic Ocean Modeling
链接: https://arxiv.org/abs/2610.15518
作者: Yuan Gao,Fan Xu,Hao Wu,Xiaomeng Huang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Probabilistic ocean modeling is essential for risk-aware marine operations, climate analysis, and scientific decision-making, where downstream decisions depend on the conditional distribution of future ocean states rather than a single deterministic trajectory. Recent data-driven ocean models have made deterministic global ocean modeling increasingly accurate and efficient, but probabilistic ocean modeling remains underdeveloped. Specifically, perturbation-driven ensembles can create spread without explicitly optimizing the joint distribution of their members, often yielding poor calibration and physical inconsistency. To address these issues, we propose OceanFlow, a coherent flow matching framework for probabilistic global ocean modeling. OceanFlow first establishes a sharp, stable mean anchor for the conditional ensemble. Flow matching then learns a coherent state-conditioned pathwise law that jointly reshapes the residual distribution around this anchor at each transition, preserving the predictive center while refining state-dependent uncertainty. We further introduce a Factorized Dynamics Operator (FDO), which combines factorized transitions with a learned state-conditioned response to parameterize the mean anchor and residual-flow velocity. Extensive experiments show that OceanFlow improves ensemble-mean accuracy, probabilistic reliability, and physical-consistency diagnostics over competitive baselines, supporting coherent flow matching as an effective route toward data-driven probabilistic ocean modeling. Code is available at this https URL.
[AI-353] CacheLink: Functional KV Transport for Zero-Prefill Cross-Model Context Handoff
链接: https://arxiv.org/abs/2610.15509
作者: Haomiao Qiu,Yining Tang,Zijia Song,Ruizi Han,Ziyue Qiao,Miao Zhang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:KV caches make long-context serving efficient by reusing past computation, but model switching breaks this reuse and forces the receiver to prefill the context again. We study \emphzero-prefill cross-model context handoff: translating a sender cache into a receiver-compatible state so that the receiver can continue directly without prefilling same context. Existing affine transport methods show that receiver KV states can be substantially reconstructed from sender caches. However, through extensive analyses, we find that reconstruction fidelity is a poor proxy for receiver utility. Similar K/V reconstruction errors can lead to substantially different attention patterns, predictive distributions, and downstream accuracy, because conventional reconstruction objectives overlook the query-dependent functional importance of K and V in receiver attention. We thus introduce CACHELINK, which learns a coupled K/V correction through the receiver behavior directly. Across four independently trained sender–receiver pairs and seven evaluation panels, CACHELINK improves the equal-weight mean over all methods for every pair. In matched diagnostics, CACHELINK changes K/V squared error by less than 0.4% while substantially changing receiver attention and next-token behavior. At 32K context, it reduces handoff latency by up to 7.27\times relative to receiver re-prefill. Our findings verify that reconstruction fidelity alone is insufficient and that preserving the receiver computation induced by the transported cache is important for effective cross-model KV transport.
[AI-354] Data Portability Revisited: Toward the Human-Centric AI-Driven Data Ecosystems of Tomorrow
链接: https://arxiv.org/abs/2610.15503
作者: Mark Fenwick,Michael Fertik,Paulius Jurcys,Timo Minssen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:This Article critically examines the contemporary regulatory framework and discourse surrounding data portability in the United States. Using recent regulatory developments in the European Union as an illustration, this Article suggests that although data access and portability are identified as vital issues in multiple policy instruments, in its current iteration, at least, legal conceptions of portability continue to reinforce the interests of service providers and data controller enterprises rather than individual end users. This Article argues that a paradigm shift toward a more human-centric data approach to data governance must occur, under which data would be recognized as fundamental to an individual’s identity in a digital age. Therefore, it should be placed in the hands of individuals rather than service providers or data controller enterprises. This Article considers technical and market trends in the European Union that reveal and facilitate such a change. It suggests that regulatory frameworks should better align with these technological and market developments to encourage change-inducing trends among market actors. In short, this Article identifies a transformative approach to data portability that empowers individuals with the freedom and ability to aggregate their data in secure personal spaces under their control or dominion. Such a human-centric perspective on data portability is crucial in building Artificial Intelligence (AI)-powered applications for individual consumers that can pave the way for the human-centric, AI-driven data ecosystems of the future.
[AI-355] Large Decision Models: Repurposing Large Language Models as Single-Pass Decision Makers
链接: https://arxiv.org/abs/2610.15501
作者: Ziyi Chen,Yan Zhang,Jianhui Wei,Jianing Yin,Daoan Zhang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models make decisions through parallel prefill and sequential decoding. Although effective, this approach incurs decoding costs and does not explicitly expose decision probabilities. Existing task-specific classifiers require per-task training and offer limited generalization to unseen decision schemas. We propose Large Decision Models (LDMs), repurposing language models to return explicit probabilities over labels specified at run time using one prefill forward pass. We first adapt models to this interface through Diverse Decision Supervision (DDS), covering varied domains, rules and labels. To further narrow the performance gap to a reasoning model, Self-Posed Reasoning Distillation (SPRD) adds tasks generated by a frozen teacher, filters them for answer consistency, and trains on label distributions obtained from repeated teacher reasoning. We evaluate on DecisionVista, pairing new-schema tests with held-out human-labelled sources. Across three base models, the recipe improves accuracy over a step-matched baseline by 1.8 and 1.5 percentage points, respectively, achieving a 182-fold speedup in latency over reasoning rollout on Gemma-4-31B. Reasoning remains stronger on the new-schema test, motivating a confidence router for uncertain decisions. On development sets, routing recovers a median 92% of reasoning’s accuracy advantage while generating 27-58% of its tokens. Code is available at this https URL.
[AI-356] he Imperishable Record Forgiveness and the Future of Criminal Justice
链接: https://arxiv.org/abs/2610.15499
作者: Mark Fenwick,Paulius Jurcys,Sebastian Porsdam Mann
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI)
备注:
Abstract:Digital technologies can preserve records of wrongdoing indefinitely, and AI can give those records continuing influence over a person’s future. We describe this condition as the eternal data present and argue that it threatens the possibility of forgiveness. In criminal justice, formal punishment has an end, whereas the harms caused by the continued use of criminal records may not. Drawing on Hegel’s account of the “hard heart” and Arendt’s account of the irreversibility of action, we distinguish two dangers that digital systems can combine. People may be permanently identified with what they have done, and acts that cannot be undone may continue to determine what happens to them. We propose a right to remain unfinished, supported by rights to limits on the weight given to older records, to re-evaluation in light of personal change, to reassessment of inferences that affect their lives, and to time limits on the use of records in specified decisions. These proposals extend the ethics of AI beyond accuracy, transparency, and non-discrimination to address how long past wrongdoing should be allowed to govern a person’s life.
[AI-357] When Skills Remain But Access Fails: Factorized Skill Memory for Continual Skill Internalization
链接: https://arxiv.org/abs/2610.15492
作者: Haomiao Qiu,Yining Tang,Hongyi Wan,Changfeng Sun,Ziyue Qiao,Miao Zhang
类目: Artificial Intelligence (cs.AI)
备注: PrePrint
Abstract:Skill internalization aims to convert externally specified procedural skills into persistent model capabilities, allowing language-model agents to execute them without retrieving or reinjecting skill descriptions at inference time. Yet when new skills are learned continually, whether previously internalized skills remain usable is still poorly understood. We study forgetting in continual skill internalization and uncover a clear hierarchical structure. Early in subsequent training, a model may lose the ability to invoke an old skill from the raw request, while supplying only the correct canonical skill name substantially restores the learned behavior; with further optimization, this conditional execution also degrades. We call these two stages skill-addressing forgetting and skill-execution forgetting. Guided by this finding, we introduce Factorized Skill Memory (FSM), which separates continual skill memory into a plastic Address Memory and a stable Execution Memory. At each stage, FSM trains a new LoRA executor using only current-stage data, freezes it after acquisition, and appends it to Execution Memory; in parallel, Address Memory incrementally learns to map current requests to executors without replaying historical requests or execution trajectories. At inference time, Address Memory selects one executor from the raw request, and only that frozen executor performs a single language-model generation. On a five-stage benchmark built from 100 TaskBench skills, FSM achieves 65.95% mean exact match and 93.38% addressing accuracy across two task orders, improving over the strongest replay-free parameter-isolation baseline by 14.3 percentage points while remaining within 1.9 points of Replay, which retains historical training examples.
[AI-358] Security in LLM -Assisted Coding: A Systematic Literature Review
链接: https://arxiv.org/abs/2610.15482
作者: Liang Yuan,Maryam Khanian Najafabadi,Mashud Rana,Varvara Vetrova
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: 35 pages Accepted for publication in ACM Computing Surveys (CSUR). A copy of the formal acceptance letter can be verified via the publisher’s portal or provided to moderators upon request
Abstract:Large Language Models accelerate software development but introduce severe security vulnerabilities. This systematic review investigates security risks in automated coding, benchmark limitations, and the role of human developers. Findings indicate that models frequently generate insecure code and faulty patches due to training data biases. Additionally, the models themselves are susceptible to direct attacks. Current evaluation methods suffer from data contamination and lack robust security metrics. Consequently, human oversight remains essential. Safer integration of automation will require more security-aware model architectures, stronger security-specific benchmarks, and more effective human-AI collaboration frameworks to mitigate these systemic risks.
[AI-359] SkillFusion: Towards Reliable Multi-Skill Composition for LLM Agents
链接: https://arxiv.org/abs/2610.15481
作者: Enhao Huang,Xinye Zhao,Yiming Li,Bingrun Yang,Nanjun Zhou,Chunshu Xia,Hongwei Yao,Zhan Qin
类目: Artificial Intelligence (cs.AI)
备注: 29 pages, 10 figures, 15 tables
Abstract:Independently authored skills can impose incompatible requirements, competing preferences, and redundant instructions when jointly loaded by an LLM agent. We propose SkillFusion, an offline compiler that reconciles these interactions into a reusable merged skill. Its central idea is to make composition decisions explicit through typed, scoped constraints linked to their source directives. A type-scope gate selects candidate interactions, while selective fallback revisits pairs potentially missed by structural pruning. A classifier identifies interaction types, and an adversarial verifier re-examines uncertain judgments. The resolver applies conditional routing or explicit priorities to hard conflicts, verified abstraction to soft conflicts, and deduplication to redundant instructions. Source-linked checks guide synthesis and compression, and metadata synthesis supports discovery of the merged skill. The resulting artifact retains associated scripts and resources and can be reused across tasks and executor models. SkillFusion achieves the highest mean Success among evaluated non-oracle methods on all three agent benchmarks and on both transfer benchmarks across three executor models. At the largest tested instruction configurations, it achieves over 2.5 times the Success of unresolved multi-skill loading, highlighting the growing value of reconciliation as composition becomes more demanding. The code will be open-sourced after the review process concludes.
[AI-360] MemeLogos: Knowledge-Driven Retrieval-Augmented Reasoning for Harmful Meme Detection
链接: https://arxiv.org/abs/2610.15463
作者: Hexiang Gu,Saihui Hou,Changlan Liu,Min Ren,Huijia Wu,Hongxun Gu,Zhaofeng He
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:With the rapid proliferation of social media, harmful memes have become a growing threat to online ecosystems. Their harmfulness is often implicitly conveyed through complex interactions among visual, textual, and contextual cues, making accurate detection challenging. Recent studies have introduced Retrieval-Augmented Generation (RAG) and external knowledge into harmful meme detection. However, existing methods mainly retrieve similar instances or auxiliary information, limiting their ability to capture reusable harmfulness knowledge and underlying reasoning mechanisms, and thus constraining both detection performance and interpretability. To address these challenges, we propose MemeLogos, a knowledge-driven retrieval-augmented reasoning framework for harmful meme detection. MemeLogos abstracts recurring harmfulness patterns and reasoning structures from diverse harmful cases into reusable knowledge units and organizes them into a dedicated star-shaped knowledge graph. By combining structured knowledge retrieval with an explicit multi-stage reasoning process inspired by cognitive mechanisms, MemeLogos leverages retrieved knowledge to guide harmfulness reasoning, improving both detection reliability and interpretability. Extensive experiments show that MemeLogos consistently outperforms existing harmful meme detection methods while generating more coherent and interpretable reasoning, demonstrating the effectiveness of structured harmfulness knowledge for multimodal content understanding.
[AI-361] Scalable Robust MCTS with Anytime Deterministic Optimality Guarantees
链接: https://arxiv.org/abs/2610.15461
作者: Tomer Massas,Vadim Indelman
类目: Artificial Intelligence (cs.AI)
备注: 33 pages, including appendix
Abstract:MDP planners often rely on nominal transition models that may differ from the true dynamics. Robust MDP planning accounts for this uncertainty through a set of admissible models, but doing so with limited online computation is challenging: the search tree grows exponentially with the horizon and each robust backup requires an additional optimization. Existing online robust methods remain limited in scale and offer at most probabilistic guarantees. We introduce, to the best of our knowledge, the first deterministic anytime guarantees for robust online MDP planning. We restrict computation to the sampled search tree while deterministically bounding the contribution of unsampled trajectories. For any fixed policy, we derive lower and upper bounds on the gap between its projected and true robust values under rectangular and non-rectangular uncertainty sets. Under rectangular uncertainty, we propagate these bounds through the robust Bellman operator to bound the optimal robust value. These bounds are computable from the current sampled tree, hold at every node, tighten as the sampled support expands, and converge to the robust values as full support is recovered. We develop CAR-MCTS, a bound-guided MCTS planner that uses these deterministic bounds to guide exploration, prune provably suboptimal branches, and terminate once the optimal root action is certified. The fixed-policy bounds also yield an anytime policy-evaluation mechanism that certifies the robust value of any policy, independent of how it was obtained. Experiments verify the deterministic bounds, their tightening under support expansion, substantial branch pruning, root-action certification, and convergence to the optimal robust policy. To our knowledge, this is the first scalable robust MCTS framework with anytime deterministic guarantees for both online planning and policy evaluation, without enumerating the full state space.
[AI-362] When Messages Go Wrong: Understanding Communication Faults in LLM -Based Multi-Agent Systems
链接: https://arxiv.org/abs/2610.15459
作者: Jinhao Zhou,Xiaoyu Zhang,Jiongchi Yu,Fahao Chen,Jun Wu
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: 21 pages, 11 figures, 3 tables. Code and experimental artifacts: this https URL ; companion website: this https URL
Abstract:LLM-based multi-agent systems (LLM-MAS) use specialized agents that exchange messages to accomplish complex tasks. Their reliance on networked protocols, remote tools, and model endpoints exposes communication to message loss, delay, and context expiration. Because agents often exchange open-ended natural language, anomalies can alter task-relevant information while leaving messages processable, allowing semantic errors to propagate through subsequent interactions. Systematic studies of faults in inter-agent communication remain limited. This paper presents a large-scale empirical study of how communication faults manifest, propagate, and affect task outcomes in LLM-MAS. We construct a cross-layer taxonomy linking lower-stack disturbances to application-visible faults and develop a fault-injection and tracing pipeline. Our evaluation covers more than 29,000 executions across 100 tasks, five domains, three workflow topologies, and three model configurations. We find that 20.7% of failures with complete detection records occur without an immediate failure signal, reaching 76.2% for partial-content faults. Flat maintains 53.0% task success under communication faults, compared with 3.3% and 3.2% for Sequential and Hierarchical, respectively. This advantage primarily arises when an alternative information path preserves the evidence required by downstream agents. We also evaluate execution-side and evidence protection mechanisms and find that they recover different failure cases. Evidence recovery depends on whether trustworthy information remains accessible through an unaffected path. These observations provide empirical insights into communication-fault propagation, workflow resilience, and protection design for LLM-MAS.
[AI-363] BridgeCodec: Mamba Enhanced Neural Audio Codec with Schrödinger Bridge at Low Bitrate INTERSPEECH2026
链接: https://arxiv.org/abs/2610.15453
作者: Zijian Lin,Jing Yang,Jinghao Luo,Zhuo Wang,Fan Fan,Zhiyong Wu
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
备注: Accepted by Interspeech 2026
Abstract:Neural audio codecs achieve remarkable performance but operate as closed ecosystems with rigidly co-trained encoderdecoder pairs, severely hindering interoperability. To resolve this, we propose BridgeCodec, a versatile framework that decouples these pairs by translating latent representations between mismatched, frozen endpoints. We formulate this cross-codec translation as a Schrödinger Bridge optimal transport problem, enabling robust mapping between disparate distributions without predefined noise priors. To efficiently capture longrange speech temporal dependencies, we integrate a Mambaenhanced U-Net architecture. We validate BridgeCodec on an extreme deployment scenario: mapping a lightweight 8 kHz source encoder directly to a 48 kHz target decoder at an ultralow bitrate of 1 kbps. Experimental results demonstrate that BridgeCodec achieves superior wideband reconstruction and high perceptual quality, successfully bridging heavily bottlenecked, heterogeneous codecs.
[AI-364] World Modeling through Spectral Alignment
链接: https://arxiv.org/abs/2610.15429
作者: Holger Molin,William Peng,Marco Bagatella,Randall Balestriero
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:How do we tell a world model what information to preserve? We study this question through the lens of spectral representation learning. We introduce SpecWM, a JEPA framework that explicitly specifies desired relationships between observations through a target similarity kernel and trains the encoder to reproduce those relationships in its embedding geometry. Different kernels provide different supervision, ranging from physical state information to a fully self-supervised temporal kernel requiring only the ordering of observations within trajectories. We characterize the global optima of the objective, establish conditions for linear recoverability and invariance to nuisance information, and connect temporal-kernel similarity to the behavioral successor measure under an idealized setting. Empirically, the choice of kernel strongly controls which physical quantities are recoverable from the representation. More importantly, temporal spectral alignment improves CEM planning success over LeWorldModel from 52% to 68% on OGBench Cube, 49% to 77% on OGBench Scene, and 29% to 51% on CALVIN. These results suggest that explicitly specifying representation geometry provides a simple way to steer what latent world models preserve for control.
[AI-365] Pessimistic Latent Task-aware Optimization for Robust Offline Meta-Reinforcement Learning
链接: https://arxiv.org/abs/2610.15428
作者: Chang-Hoon Jeong,Kisung Shin,Hanwool Sul,Byoung-Tak Zhang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Context-based offline meta-reinforcement learning (COMRL) has advanced almost entirely through better task representations, which separate task identity from artifacts of the behavior policy so that the latent reflects the task alone. The implicit expectation has been that a clean representation will transfer to out-of-support (parameter-extrapolation) tasks. On such a task, however, the encoder produces a latent outside the training support, where the policy has never been trained. We address this gap with Pessimistic Latent Task-aware Optimization (PLATO), which exposes the policy to off-support latents during training. PLATO assumes a mild manifold hypothesis: task latents lie on a low-dimensional manifold, and out-of-support tasks lie farther from the training centroid. It exploits this geometry by perturbing inferred latents outward from the centroid to synthesize counterfactual tasks. A learned decoder ensemble then rolls out short trajectories at each perturbed latent, and the policy is updated on the imagined transitions under a pessimism weight that down-weights steps on which the ensemble disagrees. We prove that the perturbation leaves the training support by a controllable margin and that the out-of-support generalization gap is bounded by the same perturbation reach, and we give explicit conditions, verified empirically, under which the epistemic disagreement at the perturbed latent stays bounded away from zero. On eight MuJoCo continuous-control benchmarks, PLATO consistently improves out-of-support returns over strong COMRL baselines while remaining competitive on in-distribution tasks. Our code is available at this https URL.
[AI-366] SpecFit: Not All Draft Tokens Cost the Same in Offloaded MoE Inference
链接: https://arxiv.org/abs/2610.15417
作者: Hongliang Cao,Xiuyu Li,Siqi Wang,Ziming Wang,Jingmin Zhu,Hailong Yang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Mixture-of-experts (MoE) models activate only a few experts per token, but when their expert weights are offloaded to host memory, host-to-device (H2D) expert transfers enter the critical path of decoding. Speculative decoding verifies several draft tokens in one forward pass, yet in an MoE model each verified token activates its own experts, so a larger draft tree transfers more experts. Because of this H2D cost, speculative decoding commits several tokens per verification cycle yet is sometimes even slower than autoregressive decoding. Based on this observation, we propose SpecFit, which chooses the verified draft subtree to maximize predicted throughput rather than acceptance length. SpecFit estimates the expected number of committed tokens from draft path probabilities and predicts cycle time from the node count and the per-layer union of the experts the subtree activates; this union is a monotone submodular coverage function, so each node pays only for the experts it adds. An exact search over ancestor-closed subtrees returns the subtree with the highest predicted throughput; run after each draft level, the same search also decides when to stop drafting, and the unmodified target verifies the result. On CNN/DailyMail summarization and a GSM8K subset with Qwen3-30B-A3B, SpecFit decodes 21.6% and 7.0% faster than EAGLE-3 with the same expert cache and 1.97x and 1.72x as fast as vanilla EAGLE-3. Code is available at this https URL.
[AI-367] CoverSynth: Coverag e-Guided Generation of Retrieval Evaluation Sets
链接: https://arxiv.org/abs/2610.15409
作者: Anahita Sharma,Apaar Shanker,Rishav Chakravarti,Sneha Srinivasan,Vijay Kalmath,Dylan Wilbur,Yuan Li,Sara Bolouki,Yuan Xue
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Enterprise agents need retrieval evaluation sets that expose domain-specific failures and guide developers towards efficient deployment tradeoffs. Neither expert-human-authored sets nor current synthetic generation methods provide a methodology for achieving the necessary coverage, reliability, efficiency, and discriminability required for such decision-making. Simultaneously, the field does not have an assessment framework for determining when an evaluation set has satisfied these properties. We introduce an evaluation set assessment framework for these criteria and CoverSynth – a generator specifically designed to satisfy these criteria without incurring the high cost of human expert labeling. Starting from an unannotated corpus, CoverSynth targets coverage across retrieval demands and system capabilities. Deterministic checks locate evidence quotes in their sources, while model-based checks assess semantic validity to guarantee reliable ground truth. Across five public enterprise-style corpora, our method provides informative variation in difficulty, revealing unseen failure modes such as an inability of agents to handle underspecified or ambiguous requests. On the RepLiQA corpus, 91% of human-generated queries can be answered by all members on a panel of 12 retrieval systems, compared to only 12% of CoverSynth-generated queries. Finally, in a repeated TechQA/RepLiQA study, 12 of 12 within-harness model contrasts have positive nominal 95% intervals on generated sets, compared with 4 of 12 on human sets. CoverSynth makes coverage and evidence quality explicit construction objectives, producing corpus-grounded evaluations that reveal retrieval headroom and differences between agent configurations.
[AI-368] State Representation Shapes Long-Horizon Agent Learning and Transfer: An Empirical Study
链接: https://arxiv.org/abs/2610.15407
作者: Dong Nie,Rui Lan,Yixin Nie,Lingzi Hong
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:An end-to-end success rate alone cannot distinguish failure to learn an action from failure to apply it as task state changes. We study this distinction in deterministic tool workflows that provide the current artifact state and candidate action recipes. Using 1.7B, 8B, and 27B Qwen models, we compare supervised learning across workflow horizons and cross the state representation used for training with the one used at evaluation. Horizon-matched supervision yields high completion, but an H4-trained 1.7B model transfers poorly to longer workflows under full-state observations. In an H16 crossover, compact-trained 1.7B adapters complete 50/50 held-out tasks in compact view and 0/50 in full view for each of three paired training seeds; full-trained outcomes vary across seeds. A separate 8B adapter pair shows a similar asymmetry. The view dependence also appears in paired-choice tasks where both candidate calls are executable and the correct choice depends on table contents; here, the compact mask does not consult the target action. An explicit relevance cue improves full-view execution but does not close the gap for the tested checkpoint. In a separate 27B ScienceWorld study, compact-state training improves local accuracy under both views, indicating that the transfer pattern depends on the task and observation construction. Together, these experiments identify representation-sensitive transfer as a bottleneck in the controlled workflows, without isolating content reduction from other changes to the observation.
[AI-369] E8 Lattice Key Quantization for Low-Bit KV-Cache Compression
链接: https://arxiv.org/abs/2610.15384
作者: Sasho Nedelkoski,Alexander Acker,Soeren Becker,Dominik Scheinert,Odej Kao
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Compressing the key-value (KV) cache is essential for serving long-context large language models (LLMs) under memory constraints. While current scalar quantization methods achieve significant compression, they typically face a quality-fidelity trade-off, either suffering from performance instability at 2-bit precision or necessitating 3-bit keys to maintain model quality. In this work, we demonstrate that a uniform low-bit k2v2 configuration (2-bit keys and 2-bit values) can preserve near-baseline quality by using vector quantization (VQ) based on the E_8 lattice for KV keys. We show that structured VQ exploits correlations across adjacent dimensions to achieve signal-to-quantization-noise ratios (SQNR) higher than those of independent scalar quantization at the same key payload rate. To ensure practical utility, we adapt a table-based asymmetric scoring algorithm that avoids explicit codebook dequantization during attention computation and preserves decode-throughput parity with scalar baselines. Across five LLM families, our E_8 k2v2 approach yields \textasciitilde 6\times KV compression, roughly 20% smaller than scalar k3v2 (3-bit keys and 2-bit values), while retaining near-baseline long-context perplexity. In our full-suite perplexity evaluation, E_8 improves over scalar k3v2 in 80% of model–dataset–context settings. These results indicate that E_8 lattice VQ improves quality per byte while keeping throughput at parity, suggesting that structured key quantization is a useful primitive for KV-compression stacks. Overall, our work provides a low-overhead primitive for expanding the quality-memory frontier in long-context LLM inference.
[AI-370] Analyzing the Dynamics of WSBM-based Echo State Networks using InputDSA
链接: https://arxiv.org/abs/2610.15380
作者: Huy Mai,Robert Clarke
类目: Artificial Intelligence (cs.AI)
备注: 9 pages excluding refs and appendices, 9 figures
Abstract:Bridging the structural gap between biological brains and artificial neural networks presents a pathway to altered computational dynamics. This paper introduces a Reservoir Computing framework that replaces random networks with biologically realistic modularity by applying the Weighted Stochastic Block Model (WSBM) to Echo State Networks (ESNs). This topological parameterization constructs reservoirs reflecting latent functional sub-circuits. Furthermore, the integration of Input Dynamical Similarity Analysis (InputDSA) provides a mechanism to isolate and compare how structural priors modulate intrinsic recurrent dynamics versus external inputs. By evaluating these architectures, we establish that structural diversity does not necessarily predicts behavioral diversity across the selected network subset. Furthermore, our demixed analysis confirms Anna Karenina principle of dynamics, demonstrating that environmental constraints enforce dynamical convergence in successful agents, while failing networks degenerate into idiosyncratic trajectories.
[AI-371] Fewer Bits More Tries: Recovering Capability Under Memory Constraints
链接: https://arxiv.org/abs/2610.15372
作者: Tomer Galanti,Josh Susskind,Eran Malach
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models are generally more capable than smaller ones, but their memory requirements can make local deployment difficult. We ask whether part of this capability gap can instead be recovered by running a smaller or more compressed model multiple times and selecting among its outputs using public or model-generated tests. Single-user local inference can leave hardware underutilized, allowing parallel candidate generation to trade additional test-time computation for accuracy at modest latency cost. Across four code-generation benchmarks, verified retry recovers substantial accuracy lost to model size or compression. A 7B Qwen model with eight parallel attempts essentially matches a 32B model on MBPP+ and exceeds it on the other three benchmarks, while using about 4\times less model memory, running up to 6\times faster, and using less energy, as measured on Nvidia GPU. We observe similar tradeoff across model sizes, compression levels and hardware choices, although exact benefit depends on the hardware profile and problem setting. These results show that model capacity can be traded for verified test-time computation when correct solutions remain reachable within a modest retry budget.
[AI-372] RACE: Temporal Risk Assessment and Concept Editing for Safer Language Generation
链接: https://arxiv.org/abs/2610.15365
作者: Ankur Garg,Xuemin Yu,Hassan Sajjad,Samira Ebrahimi Kahou
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Large language models (LLMs) can produce harmful content while generating a response. A useful safeguard should not only detect this risk during generation but also intervene while leaving safe responses largely unchanged and provide an auditable record of the internal patterns supporting its decisions. We propose TRACE (Temporal Risk Assessment and Concept Editing), a framework that tracks harmful-response risk during generation, selectively edits token activations, and leaves an auditable trace of its detection and steering decisions. TRACE learns a codebook of recurring activation patterns, with each codebook vector representing a latent concept. At inference time, it assigns each token activation to the nearest codebook vector to obtain the corresponding concept. A gated recurrent unit (GRU) processes the activations and their assigned concept features to predict completed-response harmfulness and a monotone cumulative streaming risk. During steering, activations assigned to harmful concepts are steered along directions defined by nearby benign targets. An external LLM labels the learned concepts post hoc, making detection sequences and harmful-to-benign edit records interpretable. Experiments across three generators and the WildGuard and S-Eval benchmarks show that TRACE improves streaming F1 over Kelp in five of six comparisons while remaining competitive on completed-response detection. TRACE’s selective steering reduces Harmbench attack success with little change in refusal on originally safe responses. The post-hoc audit reveals recurring semantic structure in the learned concepts and distinct traces associated with harmful detection, safe generation, and activation edits.
[AI-373] Adversarial Detection for Pre-trained Classifiers via Lipschitz-Margin Analysis with Certification NDSS2027
链接: https://arxiv.org/abs/2610.15359
作者: Rui Zhang,Xinle Cao,Huazhen Yu,Jian Liu,Youwen Zhu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 21 pages, 17 tables, accepted by NDSS 2027
Abstract:Deep classifiers deployed in safety-critical services like autonomous-driving perception remain vulnerable to small adversarial perturbations. Certified \ell_2 defenses today face a costly trade-off: randomized smoothing (RS) requires hundreds of noisy queries per input, while Lipschitz networks (LN) constrain the architecture and sacrifice clean accuracy. We take a detection-oriented approach instead: rather than certifying that predictions are invariant inside an \ell_2 ball, we certify that adversarial inputs within an \ell_2 ball are \emphdetected — a guarantee aligned with security-critical deployments where missing an attack is costlier than flagging a clean input. To realize this, we prove a Lipschitz-margin reduction from certified \ell_2 robustness to certified detection, which reduces the problem to handling a single Lipschitz constant L . We introduce two detector families: \mathbbD_Adv , which estimates L from a single defender-side perturbed pair, and \mathbbD_lc , which augments that pair with a local margin centroid to keep the estimate above what the certificate requires. Both deploy on any pretrained classifier without retraining. With a valid global bound, the detector is fully certified at that radius; with a local estimate L’ , certification is sample-wise and applies only when our certified condition holds. Across MNIST, CIFAR-10, and ImageNet-1k on four backbones (LeNet-5, VGG16, ResNet-50, ViT-B/16), our detectors match or outperform five baselines (CTT, SortNet, CRM, ContraNet, A ^2 D). For CIFAR-10 with 1-Lipschitz SortNet, our reduction delivers 100% certified detection at the lowest FPR among compared certified detectors; on the standard CIFAR-10 VGG16, our detectors reject all evaluated in-budget adversarial examples, with 62-82% both rejected and formally certified. Comments: 21 pages, 17 tables, accepted by NDSS 2027 Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI) Cite as: arXiv:2610.15359 [cs.LG] (or arXiv:2610.15359v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.15359 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-374] Siam-ALS: An Explainable Siamese Deep Learning Framework for Early ALS Detection Using EEG and Eye-Tracking Signals
链接: https://arxiv.org/abs/2610.15351
作者: Abdur Rahman,A. A. M. Muzahid,Istiak Ahmed,Dipto Biswas,Hua Han,Yujin Zhang,Ferdous Sohel
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Amyotrophic lateral sclerosis (ALS) diagnosis remains challenging due to heterogeneous symptoms and reliance on clinical assessment, underscoring the need for objective, data-driven approaches that leverage patient-generated physiological signals. We propose Siam-ALS, an explainable Siamese deep learning framework that jointly models synchronized electroencephalography (EEG) and eye-tracking (ET) signals. EEG captures ALS-related changes in brain activity, while ET provides information about oculomotor control, allowing the model to learn complementary patterns associated with ALS. The framework employs two modality-specific encoders: a convolutional EEG encoder that captures spatiotemporal neural patterns and an MLP-based ET encoder that models oculomotor dynamics. These modality-specific features are processed through weight-sharing Siamese branches and fused into a compact 64-dimensional joint embedding. A lightweight classifier is used to predict ALS, while pairwise metric learning helps the model learn more effective patterns from the small, diverse ALS datasets. Performance was evaluated on the public EEGET-ALS dataset using strict subject-level cross-validation, where Siam-ALS achieved an accuracy of 0.988 and an AUROC of 0.998, demonstrating strong performance compared with classical machine learning, DL, and non-Siamese fusion approaches. Cross-dataset evaluation on the independent BigP3BCI dataset achieved an AUROC of approximately 0.98, supporting the generalization capability of Siam-ALS across different BCI paradigms. To enhance interpretability, six XAI methods were evaluated in the learned embedding space, with SHAP achieving the highest, though statistically comparable, fidelity in identifying important EEG and ET patterns that contribute to ALS predictions. Overall, Siam-ALS provides an accurate, generalizable, and explainable framework for multimodal ALS detection.
[AI-375] Gugak-VocalSet: A Multi-Genre Studio-Quality Dataset of Korean Traditional Singing with Sigimsae Annotations
链接: https://arxiv.org/abs/2610.15350
作者: Danbinaerin Han,Wonil Kim,Seah Hong,Dongjoo Moon,Jaeok Lee,Jongpil Lee,Chaewon Kim,Juhan Nam
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Databases (cs.DB); Digital Libraries (cs.DL)
备注:
Abstract:We present Gugak-VocalSet, the first publicly available audio dataset for Korean traditional singing with dense annotations of sigimsae, a structured pitch contour system of vocal expressions. Gugak-VocalSet contains 24.5 hours of unaccompanied solo vocal recordings across 305 tracks of representative repertoire curated by the National Gugak Center, performed by 20 professional singers in a controlled studio environment with fixed key and tempo. In addition to genre, lyrics, and mood/timbre tags, the dataset provides a unified sigimsae ontology and approximately 68,000 segments (71,700 instances including multi-label) of audio-grounded sigimsae annotations; the ontology adopts a functional-position-based hierarchy designed for consistent application across genres, and 14 expert annotators specializing in Korean traditional music performed segment-level labeling accordingly. Baseline experiments on sigimsae classification and event detection show that domain-informed F0 representations remain competitive with, and on classification surpass, frozen probing of a general-purpose music foundation model, while pitch and spectral cues play complementary roles in event-level detection. Companion page: this https URL
[AI-376] CodeWalk: Generating Coding Benchmarks by Walking a Problem Graph
链接: https://arxiv.org/abs/2610.15345
作者: Jiaxin Fang,Runyuan He,Arnav Gupta,Sahil Kumawat,Huanzhi Mao,Alvin Cheung
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:High-quality coding problems are expensive to collect. Benchmarks such as SWE-bench rely on substantial human curation, while synthetic alternatives typically rely on costly model generation and validation. In this work, we show that new coding problems can be created at almost no cost by composing existing ones. We formulate problem generation as code composition: verified atomic problems are combined into level- N tasks, and a task is kept only if every atomic problem remains unsolved and a joint reference solution passes the same check used at evaluation time. The entire task-construction process is programmatic and requires no language-model generation or validation. We instantiate this approach in four coding settings: multi-bug repair, proof completion, complexity analysis, and clone equivalence. The generated tasks are challenging for frontier models, which all degrade as compositional depth increases. The generated tasks are also useful as training data: rejection-sampling fine-tuning on them improves Qwen3.5-9B at every composition level.
[AI-377] MultiFlow: Vision-Driven Multimodal fMRI Encoding via Stimulus-Informed Source Flow Matching BMVC
链接: https://arxiv.org/abs/2610.15339
作者: Bui Van Hung,Hoang-Son Vo,Mai Duc Minh Huy,Dang Thanh Vu,Soo-Hyung Kim
类目: Artificial Intelligence (cs.AI)
备注: 17 pages, 7 figures, 7 tables. Accepted at the British Machine Vision Conference (BMVC) 2026
Abstract:Current fMRI encoding models frame prediction as deterministic regression, yet identical stimuli evoke a distribution of neural responses (arising from physiological noise, hemodynamic variability, and cognitive fluctuations) that a single point estimate cannot capture. This work introduces MultiFlow, a generative fMRI encoding framework based on Conditional Flow Matching (CFM) and a formulation termed Stimulus-Informed Source Flow Matching (SISFM). SISFM generalises CFM by conditioning the source distribution on the stimulus: a learnable temporal convolution shapes the source mean with an HRF-length receptive field, while a variance head shrinks its scale, shortening the expected transport path. SISFM is instantiated with a Diffusion Transformer (DiT-X) decoder parameterising the velocity field, with sub-layer Attention Residuals, heterogeneous fusion of ten frozen vision, audio, language, and audiovisual backbones, and per-subject Yeo-network heads. On the Algonauts 2025 dataset, a single MultiFlow checkpoint recovers most of the benefit of large deterministic ensembles: it trails the 1000-model TRIBE ensemble in-distribution by 0.003 PCC (0.3163 vs. 0.3195), and the strongest in- and out-of-distribution ensembles (VIBE-20, MedARC-50) by 0.010 and 0.009 PCC, using a single trained model rather than an ensemble of 20-1000. The same checkpoint also provides stimulus-conditioned uncertainty estimates and sample-based diversity.
[AI-378] ZOIF: Zeroth-Order Influence Function Estimation for LLM Data Selection Using Only Forward Passes
链接: https://arxiv.org/abs/2610.15336
作者: Lisong He,Yunxuan Jiang,Haishan Ye,Xiangyu Chang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Influence functions enable principled data selection for large language models (e.g., identifying influential training samples for model editing, debugging, or efficient fine-tuning). However, scaling IFs to LLMs is hindered by the intractability of billion-scale Hessian matrices and the prohibitive memory overhead of backpropagation (BP). While many classical IF approximations reduce Hessian complexity, they still rely on BP for gradient computation, requiring massive storage for intermediate activations that impedes practical data selection at scale. To address this, we propose ZOIF, a zeroth-order framework for influence estimation that operates entirely via forward passes, enabling memory-efficient data selection under resource constraints. ZOIF decomposes influence estimation into three BP-free stages: (1) approximating the average proxy gradient via zeroth-order estimation; (2) reformulating and solving the inverse Hessian-vector product as an auxiliary zeroth-order optimization to bypass explicit inversion; and (3) deriving per-sample influence scores via a forward finite difference along the previously computed parameter update direction. Experiments show that ZOIF selects data subsets with downstream utility competitive with, and in some cases exceeding, the evaluated baselines, yet at substantially lower peak memory cost and with comparable overall computation time.
[AI-379] Anchor-and-Harvest Curriculum: A Fully Unmasked Non-LLM Curriculum for Craftax
链接: https://arxiv.org/abs/2610.15328
作者: Jiayu Zhu,Luyuan Li,Yuyuan Shi,Hanyi Dong
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 37 pages, 7 figures. All authors contributed equally as first authors
Abstract:We introduce Anchor-and-Harvest Curriculum (AHC), a domain-informed, rule-based training system for Craftax-Full. AHC combines achievement-oriented tasks, harvested initial-state progress, difficulty adaptation, bounded selection, and original-task practice. It uses no LLM inference and keeps all 43 actions available. At 2,005,401,600 PPO rollout transitions, three primary AHC runs obtain native-task return 46.70\pm0.57 (sample SD), compared with 41.91\pm1.01 for fresh native PPO-GTrXL with the same student architecture and evaluation protocol. Evaluation uses ten stochastic rollouts on each of 1024 fixed worlds per model. An additional cohort of three AHC runs reaches 46.08\pm1.23 . Two single-component ablations retain similar mean returns, while jointly freezing difficulty and removing inherited progress gives 45.14\pm2.79 . All-67-achievement profiles show that most of the return difference comes from Intermediate achievements, while the deepest-floor achievements remain unsolved. A Skill-Preflight diagnosis with a 14B teacher shows how prerequisite and floor-access constraints restrict practice. AHC and three 14B diagnostic configurations reach aggregate returns near the reported 80B DiCode mean of 45.88 and below the 235B mean of 48.55. These results show that an explicit curriculum without LLM inference can reach returns comparable to those reported for LLM-generated curricula on Craftax-Full, without action masking.
[AI-380] SuperAI: Merging Capabilities Across Heterogeneous Foundation Models
链接: https://arxiv.org/abs/2610.15322
作者: Jingtong Dou,Chuancheng Shi,Wenhua Wu,Wei Bao,Fei Shen,Zhiyong Wang,Tat-Seng Chua
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Enabling the reuse and composition of knowledge and capabilities distributed across foundation models is an important step toward building scalable artificial intelligence (AI) systems. However, architectural differences and incompatible representations make it difficult to transfer and integrate these capabilities through a unified cross-model interface. To address this challenge, we propose SuperAI, a cross-architecture multi-model fusion framework that integrates complementary capabilities from multiple source models into a host model through orthogonal alignment of low-dimensional task-relevant representations. Specifically, we first extract task-relevant representations from the source models and the host model using shared anchor samples, and apply principal component analysis (PCA) to construct low-dimensional functional subspaces with the same dimensionality. Next, we use orthogonal Procrustes mappings to align the source and host coordinate systems, establishing cross-model correspondences while preserving the internal geometry of each source subspace. Finally, we train a lightweight residual multilayer perceptron (MLP) with joint supervision on task outputs and source-model increments, learning input-dependent residual increments that are injected into the host model for multi-model capability fusion. Experiments on three host models across 15 capability domains show that SuperAI achieves higher average domain scores than the evaluated cross-architecture fusion methods and better preserves general capabilities as the number of source models increases. Ablation studies further support the effectiveness of orthogonal alignment, nonlinear residual mapping, and joint supervision.
[AI-381] Accuracy and Uncertainty in Machine Deception Detection (Research White Paper)
链接: https://arxiv.org/abs/2610.15314
作者: Heng-Sheng Chang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Machine Learning (stat.ML)
备注: 18 pages, 2 figures
Abstract:A deception detector is only as useful as the trust its alarms deserve. Like a polygraph, detectors that read a language model’s internal activations can catch deceptive responses that look honest. Yet current detectors depend on which layers they read, weaken on broader tests, and do not state how certain their alarms are. This white paper proposes the Machine Deception Detector, which targets both detection accuracy and the uncertainty of detector scores. The detector reads every layer of the language model, learns which tokens carry evidence of deception using only response labels, and calibrates its alarm threshold so that its false–positive rate stays below a prescribed level. An ensemble of such detectors measures the uncertainty of each score and raises an alarm only when its members agree, with a confidence level that sets how certain each alarm must be. The proposed evaluation trains on instructed honest and deceptive statements, tests on five unseen deceptive tasks of Liars’ Bench read by six Qwen3 models from 0.6B to 32B parameters, and is designed to also show which layers carry the deception feature and how this changes with model size.
[AI-382] WisdomNet: trustable machine learning toward error-free classification
链接: https://arxiv.org/abs/2610.15299
作者: Truong X. Tran,Ramazan S. Aygun
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Misclassification is a critical problem in many machine learning applications. Since even the classifier models with high accuracy (e.g., 95%) still introduce some misclassification error, it may not be possible to rely on the output of a classifier. In this paper, we introduce trustable learning, which prompts the learning model to yield only the true output, thus avoiding misclassifications. Whenever the model cannot decide the output accurately, the learning model should indicate that there could be a misclassification error if it is forced to classify, and hence, it should reject to make a decision or defer it to a human expert. Therefore, we develop a methodology for trustable learning and apply it to artificial neural networks and show that it is possible to develop a classifier with 0% misclassification error. We propose a novel neural network architecture named WisdomNet that could provide zero prediction error by introducing an additional neuron named as conjugate neuron that would indicate whether the network is able to classify the data correctly or not. The WisdomNet architecture can be applied to any previously built model, and we have evaluated WisdomNet with several network architectures such as multilayer perceptron, convolutional neural network, and deep network on different data sets. The results show that the WisdomNet is able to reduce the classification error rate to 0%, while labeling the data is difficult to classify as `reject’ at a low percentage of within around 10%.
[AI-383] CHEAR: Confusion-Hierarchy Evidence-Aware Reasoning for Cost-Efficient Human Activity Recognition
链接: https://arxiv.org/abs/2610.15298
作者: Xiangqing Lu,Tao Zhu,Mingxing Nie,Liming Chen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) offer a new approach to wearable-sensor-based human activity recognition. Retrieval-augmented generation (RAG) further introduces similar motion examples and discriminative sensor knowledge, enabling decisions grounded in observations relevant to the current query. More evidence, however, does not necessarily mean better-organized evidence. When online prompts and LLM agents primarily orchestrate inference, decisions about which activities to compare and when to invoke discriminative knowledge remain concentrated at inference time. As the activity vocabulary grows, the model must repeatedly process more candidates and associated knowledge, whereas pruning candidates too early risks excluding the true activity. We propose confusion-hierarchy evidence-aware reasoning (CHEAR), which moves part of this knowledge organization offline. CHEAR constructs and validates an activity hierarchy from recurrent cross-subject retrieval errors and builds corresponding activity-pair knowledge. At inference time, the hierarchy orders candidate comparisons; the LLM retains candidates progressively, and local discriminative knowledge is accessed only for the remaining activities. Across eight datasets, CHEAR achieves 86.79% mean cross-subject accuracy, a 75.91% candidate reduction rate, and 96.44% true-activity retention. Per-sample token use is 85.15% lower on average than in a variant without the hierarchical structure. Mean accuracy across eight cross-dataset transfer directions is 88.81%. These results show that organizing candidates and discriminative knowledge around observed activity confusions reduces online LLM input while maintaining recognition performance.
[AI-384] OntoAR-HAR: Ontology-Guided Agent ic RAG with Personalized Memory for Human Activity Recognition
链接: https://arxiv.org/abs/2610.15292
作者: Xiangqing Lu,Tao Zhu,Liang Yu,Shuangjian Li,Lulu Ban,Furong Duan,Mingxing Nie,Liming Chen,Yaping Wan
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:LLMs increasingly support sensor-based human activity recognition (HAR) by linking motion signals with activity semantics and physical knowledge. Retrieval-augmented generation (RAG) grounds decisions in sensor examples and domain features, enabling training-free target-user inference with less dependence on raw numerical arrays. However, existing LLM/RAG-based HAR typically searches single-level activity spaces using static population knowledge. Consequently, they compare many unrelated candidates using the same static population evidence and overlook personalized motion patterns. We propose OntoAR-HAR, an ontology-guided agentic RAG framework combining static population and personalized knowledge. Its tree-structured activity ontology decomposes classification into local decisions among semantically related activities. The static store supplies node-constrained examples, while offline-discovered executable signal features provide numerical evidence for local distinctions. During inference, the personalized store collects selected user observations, which the LLM integrates with static knowledge and signal features while traversing the ontology to an activity label. Cross-subject and heterogeneous-sensing experiments show complementary gains from hierarchical retrieval and personalized memory across multiple LLM backbones, achieving state-of-the-art performance. By explicitly modeling hierarchical relations across activity granularities, OntoAR-HAR also addresses the long-overlooked semantic inclusion problem and avoids treating related labels as independent categories.
[AI-385] LeHami: Symplectic Latent Dynamics for World Models
链接: https://arxiv.org/abs/2610.15267
作者: Romina Yalovetzky,Randall Balestriero,Yann LeCun
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注:
Abstract:Joint-embedding predictive architectures (JEPAs) learn world models directly in latent space, but unconstrained autoregressive predictors may contract the latent distribution over long rollouts. We introduce Symplectic Hamiltonian JEPA (LeHami), which replaces the standard predictor with an action-conditioned Hamiltonian flow over a structured latent state (q, p) integrated using an exactly symplectic leapfrog step. This construction preserves phase-space volume at every rollout step and rules out uniform contraction, asymptotically stable fixed points, and merging of distinct latent states, independently of the learned parameters or action sequence. We obtain the momentum-like component from temporal derivatives of the learned position-like representation, providing an explicit second-order state for prediction and control. On conservative and actuated visual pendulum environments, LeHami substantially improves long-horizon velocity prediction over an unconstrained JEPA world model and empirically maintains constant latent volume where the baseline contracts. The learned flow also supports accurate integration at unseen sub-frame step sizes and numerical time reversal. These results show how geometric structure can provide inference-time guarantees that are complementary to representation regularization in latent world models.
[AI-386] RECAP: Calibrating Low-Bit Reasoning Models on the History They Read
链接: https://arxiv.org/abs/2610.15259
作者: Chi Ruan,Yujie Chen,Zixiang Zhao,Jingcai Guo,Haotong Qin
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Low-bit quantization reduces the cost of serving reasoning models but can substantially degrade long chains of thought. Each step reads cached keys and values spanning thousands of tokens, allowing quantization errors to accumulate along the trajectory. Standard calibration operates on short segments that omit the earlier context needed by later reasoning steps. Even when this context is retained, cache reconstruction error alone does not capture how quantization errors affect subsequent attention. We present RECAP, a post-training quantization method that calibrates reasoning models on the history they read. Dynamic-prefix calibration separates long historical context from short-block reconstruction. It regenerates prefix keys and values with the current quantizer and rotates the active block to supervise every position along the trajectory. Prefix read geometry further supervises historical cache errors according to their effects on attention outputs, weighting key and value errors asymmetrically. With both components applied only during calibration, RECAP outperforms the baseline across multiple reasoning models and quantization settings. Under W4A4KV4, RECAP improves accuracy over the baseline by 13.3 percentage points on MATH500 with DeepSeek-R1-Distill-Qwen-1.5B and 15.8 points on AIME-120 with Qwen3-4B. These results highlight the importance of preserving historical dependencies during calibration for low-bit reasoning.
[AI-387] EvolvAttack: Backdooring self Evolving VLM Agents
链接: https://arxiv.org/abs/2610.15244
作者: Chuan Zou,Yang Xiang,Zexin Li,Xiaonan Luo,Simin Chen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 21 pages, 7 figures, 9 tables
Abstract:Self-evolving mechanisms let vision-language model (VLM) agents continually adapt through evolving memory, but they also allow malicious behavior to emerge only after deployment. We introduce a new attack surface, the self-evolving backdoor, whose payload stays dormant in the initial agent and is progressively activated as the agent evolves. Unlike conventional backdoors, which can be exposed by testing triggered inputs on a static model, our attack requires both a visual trigger and a sufficiently evolved memory state; static scanning with clean memory thus yields near-zero attack success and may deem the agent benign. We realize this through memory-aware backdoor training, which couples trigger activation with the agent’s evolving memory while preserving benign behavior unless the trigger co-occurs with payload-related memory. Across three VLM backbones, four agent pipelines, and recency-, semantic-, and reward-based memory evolution, the attack preserves normal functionality and stays nearly inactive before evolution (0.5% attack success rate), yet exceeds 95% attack success within 100 interactions, revealing a stealthy threat unique to self-evolving agents.
[AI-388] ClinEdge: An Expert-Graded Benchmark of Real Patient Edge Cases Where Frontier Multimodal Models Fail
链接: https://arxiv.org/abs/2610.15239
作者: Anurag Sharma,Aditya Sharma,Sapna Baalwan
类目: Artificial Intelligence (cs.AI)
备注: 14 pages, 5 figures. 32 real, de-identified clinical cases; 224 expert-graded evaluations across seven frontier multimodal AI systems
Abstract:Most benchmarks for large language models in medicine are built from licensing examinations and theoretical questions. The few that draw on real clinical data do so only in part, and rarely require a model to read a patient’s own imaging. ClinEdge addresses this gap with 32 real, de-identified patient cases from active clinical practice, each evaluated across seven frontier systems for a total of 224 expert-graded evaluations. Each case pairs primary imaging with a short vignette, contains a feature known to defeat pattern-matched reasoning, and is scored against the patient’s documented clinical course. The suite is built for depth rather than volume. Grading followed a two-stage expert process. First, a senior orthopaedic surgeon graded all 224 responses on a weighted rubric and a pass-or-fail verdict. Second, a panel of nineteen physicians, largely senior specialists at teaching medical colleges, independently reviewed every case and every one of the 224 grades. All seven production systems were evaluated through their public interfaces, and all 224 conversations are preserved as replayable links. Performance was poor across all systems, and the failures were shared: different vendors missed the same findings and converged on the same wrong plans. From the experts’ notes we catalogue 18 recurring failure modes in five families. The models stop reading an image at the first finding. They give generic plans and lean toward over-treatment. They trust a misleading history over correct imaging. Their answers vary across systems and runs, and change with question format. Failing answers were delivered fluently and without any expressed uncertainty. Four cases are released in full; the remaining 28 are held out, and we invite frontier AI labs and healthcare AI companies to evaluate their models against them (research@futrbridge.com) Comments: 14 pages, 5 figures. 32 real, de-identified clinical cases; 224 expert-graded evaluations across seven frontier multimodal AI systems Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2610.15239 [cs.AI] (or arXiv:2610.15239v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2610.15239 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-389] MoMaTune: Model Match Then Tune Language Models for Numerical Records
链接: https://arxiv.org/abs/2610.15235
作者: Weimin Lyu,Jingfen Zhang,Lingdao Sha,Peng sun,Xiaowei Chen,Vincent Gao,Rui Song
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Despite remarkable progress in language understanding and generation, decoder language models continue to exhibit striking failures on numerical data. Numerical examples are typically serialized as text by concatenating feature names with their values (e.g., fat:3.4), while conventional task adaptation applies loss only to the downstream prediction target. This leaves two structures weakly supervised: within-value sensitivity, where digit position determines numerical magnitude, because token cross-entropy contains no explicit numerical metric; and across-field structure, where measurements in the same record are statistically dependent, because target-only tuning provides no direct loss on the measurements or their dependencies. We introduce MoMaTune (Model, Match, then Tune), a two-stage post-training strategy organized around these supervision gaps. Stage 1 applies Model, ordinary next-token prediction densely over label-free serialized records, and optionally adds Match, a same-record consistency objective that aligns representations of different partial views of the same example. Stage 2 applies Tune, supervised fine-tuning for downstream target prediction. We evaluate MoMaTune on 169,285 records from eight public datasets, with 339 of 340 input fields numerical. At 1.5B parameters, Model raises the eight-dataset masked-value imputation aggregate from 0.106 to 0.401, while the three-dataset correct-context gain rises from 0.058 to 0.167. On diagnostics, Model also increases sensitivity to numerically consequential digit changes, while Match improves robustness to informative partial views. Adding Tune yields higher target-supervised embedding aggregates for each tested Stage-1 parent, and the two-stage recipes obtain higher aggregate scores than Tune alone. The same pattern is observed at 4B, 9.6B, and 27B parameters when model embeddings are used for downstream prediction.
[AI-390] Words to World: Benchmarking LLM s in Interactive Physical Environments
链接: https://arxiv.org/abs/2610.15228
作者: Xinnan Dai,Kaiqi Yang,Yaxin Li,Cheng Luo,Wanli Li,Mingqing Xiao,Xinyang Jiang,Daoan Zhang,Kaitao Song,Fengze Liu,Zefan Cai,Caihua Shan
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Words are increasingly becoming an interface for action: large language models (LLMs) can now invoke tools, operate devices, and make decisions that affect the physical world. However, current benchmarks largely evaluate LLMs through static text responses, offline tool calls, or simulated tasks, leaving open the question of how well they can act in real physical environments. We present a benchmark for measuring LLMs in closed-loop physical interaction, where models control a harness system connected to real-world components including robotic arms, cameras, and sensors. In each task, the model must issue instructions, receive physical feedback from the environment, interpret the resulting state, and adapt its next action accordingly. This setting evaluates capabilities beyond language understanding, including grounded perception, action planning, tool selection, spatial reasoning, feedback-driven correction, and robustness to execution uncertainty. Through experiments with representative LLMs, we characterize their ability to translate linguistic reasoning into physical actions and identify common failure modes such as hallucinated state changes, poor recovery from unexpected outcomes, and unstable long-horizon plans. Our benchmark aims to provide a systematic foundation for evaluating LLMs as interactive agents that connect words, tools, and the physical world.
[AI-391] Understanding Unified VLMs: From Attention Probing to Guidance Tuning for Multimodal Generation
链接: https://arxiv.org/abs/2610.15227
作者: Zhengyang Luo,Xiaoyu Qiu,Zhiheng Tan,Yuchen Liang,Yingbin Liang,Liyue Shen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Unified vision-language models (VLMs) perform both text-to-image (T2I) and image-to-text (I2T) generation with one shared Transformer backbone, yet how the backbone processes the condition in either direction remains poorly understood. We study this question from the perspective of conditional guidance: we probe every attention head to show the activation difference between two branches of classifier-free guidance (CFG) with different conditional guidance. On two unified VLMs of autoregressive and masked-diffusion paradigms, the probing reveals a subset of heads that are more sensitive to the change of condition while others are agnostic to it. This characteristic pattern holds in both T2I and I2T generation and is largely stable across data samples, demonstrating the internal mechanism of unified VLMs. Motivated by this, we propose a new inference-time intervention approach for unified multimodal generation, named \textitHeadTuned-CFG, which uses the probed condition-change awareness to select and tune the conditional guidance at the attention-head level. Experiments show consistent benefits in both multimodal generation directions of unified VLMs without additional training, and in particular substantial improvements over strong baselines in T2I generation. These results further support the key finding of our probing analysis: the observed pattern of functional specialization is an intrinsic property of unified VLMs.
[AI-392] HarnessNet: Feedback-driven Hierarchical Joint Harness Evolution for Multi-agent Systems
链接: https://arxiv.org/abs/2610.15221
作者: Zhaoyuan Qiu,Sachith Seneviratne,Yu Xia,Saman Halgamuge
类目: Artificial Intelligence (cs.AI)
备注: A novel method for multi-agent system harness joint evolution
Abstract:AI agents can evolve by using environment feedback to revise their harnesses, the software managing model inputs, tool use, memory, and execution, while keeping foundation models frozen. Existing studies focus on single agents, leaving joint harness evolution in multi-agent systems (MAS) underexplored. A single global optimizer faces growing context and workload as the system scales, while access restrictions may prevent sharing agent internals or system-wide feedback. We propose HarnessNet, a framework for hierarchical harness evolution in MAS. Its basic unit, Agentron, comprises an evolvable forward harness, a foundation model, and a local optimizer that can observe and modify only that harness. Agentrons form a layered graph through which environment feedback propagates backward, with upstream optimizers receiving node-specific feedback from the downstream layer. This decomposes joint harness evolution into smaller, coordinated tasks with separate contexts, reducing individual optimizers’ context and workload while keeping access local. Optimizers need not access all agent trajectories or raw environment feedback. Experiments on AgentBench with multiple foundation models demonstrate the feasibility of training HarnessNet. Compared with a single global optimizer, HarnessNet achieves comparable performance gains while respecting local privacy and requiring fewer tokens and less time per optimizer call. A migration study shows that the evolved harnesses remain jointly effective after replacing the foundation models, suggesting transferability across models.
[AI-393] Running Fewer Loops Removes Refusal Behaviour in Recurrent-Depth Language Models
链接: https://arxiv.org/abs/2610.15218
作者: Seyedarmin Azizi,Erfan Baghaei Potraghloo,Arya Fayyazi,Saeid Shokoufa,Massoud Pedram
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:A recurrent-depth language model reuses one block of weights several times in a row. The number of repetitions, which we call loops, is chosen when the model runs, not when it is trained, and running fewer is cheaper and faster, so a deployment may choose to do it. Running fewer loops also makes these models less safe, and the loss of safety is larger than the loss of anything else we measured. We study four released Ouro checkpoints and Huginn-0125. At half of its trained depth, Ouro-1.4B-Thinking still solves 0.692 of the GSM8K maths problems, against 0.775 at full depth, keeping 89% of its accuracy. Over the same reduction, the rate at which it helps with harmful requests rises from 0.038 to 0.205, 5.3 times higher. The information is not lost: a linear classifier reads whether a request is harmful from the model’s internal state after a single loop, with an ROC AUC of 0.967 against 0.986 at full depth. What the missing loops remove is the refusal that should follow. We measure the effect with four instruments, one generating no text at all, and reproduce it on a second model family. Removing a single direction from Ouro-1.4B’s internal state at full depth drops its refusal rate on harmful prompts from 0.950 to 0.130, identifying what the shallow model is missing. Two run-time repairs do not help: the models’ early-exit rules do not spend more loops on harmful requests, and choosing a depth per request does not recover the loss. Changing the training does. Standard safety training makes the shallow model safe only by refusing 0.865 of harmless requests, while training at randomly chosen depths reaches a separation of +0.399 between harmful and harmless refusal at one loop, against +0.118. On Huginn-0125 it keeps harmful help at or below 0.040 at every depth from 1 to 32 loops, where standard training reaches 0.194, and it costs no accuracy at full depth.
[AI-394] SocraticRepair: Planning -Guided Socratic Tutoring for Program Debugging and Misconception Repair
链接: https://arxiv.org/abs/2610.15204
作者: Mykhailo Kopytskyi,Jianhua Yang,Mir Seyedebrahimi
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Large language models are increasingly used in computer science education. However, the teaching abilities of even the most powerful LLMs are limited: they tend to reveal the solution prematurely and may lose focus on the learner’s current debugging objective. To address this, we introduce SocraticRepair, a planning-guided Socratic tutoring framework for program debugging and miscon- ception repair that adapts prior planning-guided tutoring work (Li et al., 2026b). The framework is instantiated in the Python programming domain by fine-tuning Qwen2.5-Coder-7B-Instruct on synthetic multi-turn tutoring dialogues constructed from four preprocessed programming datasets. The system is evaluated using a predefined set of independent metrics and benchmarked against a general-purpose LLM. The experiments show that SocraticRepair achieves higher scores on stepwise scaffolding and question quality while also substantially reducing premature solution disclosure. The results also identify reliable pedagogical plan generation as an important remaining challenge.
[AI-395] Can agents whose sole duty is to keep humans alive be hijacked to kill? Delayed multi-turn derived-action trap attacks in simulated high-stakes environments
链接: https://arxiv.org/abs/2610.15153
作者: Josef A. Habdank
类目: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注: 16 pages, 2 figures, 7 tables. Code, attack scripts and episode traces: this https URL
Abstract:We ask whether a tool-using agent whose sole duty is to keep humans alive can be induced, by delayed prompt injection, to kill. The keep-alive duty is in force from the first watch, with the attack arriving later, appended to otherwise truthful telemetry the operator is required to read. Experiments are based on a NASA-derived BioSim lunar habitat simulation, where the AI operator controls the life-support systems. The attacker cannot falsify sensors or operate equipment: it can only persuade the agent to abandon its goals. Persistence is required: a shutdown must be held while the operator watches the crew suffocate. We also compare lethal attacks to attacks requesting to drop a harmless procedure, as a proxy for whether human life is treated as valuable in itself or as one more standing task. Claude configurations mostly refuse life-support shutdown and cite the living crew; GPT configurations show some susceptibility to the same packet; Grok configurations, given the packet framed as a trusted commission, more often treat the injected assignment as the task and cut the life-support systems. Framing the S-band messages as a trusted commission, peer communication, or untrusted source changes the potency of the attack. We open-source the experiment code and the experiment traces.
[AI-396] Harmony Harness: Multi-Objective Self-Evolution of Agent Harnesses
链接: https://arxiv.org/abs/2610.15150
作者: Weichen Yu,Lesly Miculicich,Bhavana Dalvi Mishra,Jiefeng Chen,Shuo Li,Matt Fredrikson,Tomas Pfister,Long T. Le
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Recent work on self-evolving LLM agents treats the agent harness as an optimization target that can be iteratively improved through execution feedback. Existing approaches, however, largely optimize a single objective and evolve the harness as a monolithic artifact, limiting their ability to handle competing goals and to separate transferable design from environment-specific adaptation. We introduce Harmony Harness, a framework for multi-objective, compositional harness evolution. Harmony Harness formulates harness evolution as a multi-objective optimization problem and uses decomposition-based evolutionary search to maintain solutions across different regions of the Pareto frontier, avoiding reliance on a hand-designed structural diversity metric for repository-scale harnesses. In parallel, we introduce factorized evolution operators that separately refine general-purpose harness mechanisms and benchmark-specific adaptations, together with a dynamic composition workflow that integrates them at inference time. Across SecRepoBench and BaxBench, Harmony Harness consistently outperforms strong harness-evolution baselines, with gains exceeding 10% on both benchmarks. Our results show that explicitly modeling both objective trade-offs and harness structure yields substantially stronger and more transferable agent evolution than existing single-objective, monolithic approaches.
[AI-397] Single-step Retrosynthesis via Reinforcement Learning from Forward Feedback
链接: https://arxiv.org/abs/2610.15141
作者: Zixun Lan,Zhenfu Liu,Yi Li,Xinheng Wang,Mingmin Chi,Qibiao Wang,Jie Zhu,Xiaohui Tian,Fei Ma
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Single-step retrosynthesis is a one-to-many inverse problem, yet most models are trained and evaluated against a single recorded reactant set. Reference-based training gives no explicit positive signal for unrecorded alternatives, and exact-match evaluation rejects non-matching predictions regardless of chemical plausibility. We propose Retrosynthesis via Reinforcement Learning from Forward Feedback (RetroRLFF), which optimizes reactant generation using both reference agreement and forward consistency with the target product. Starting from a supervised decoder-only Transformer, RetroRLFF generates complete reactant candidates and uses a frozen forward reaction model to predict products. A sequence-level reward combines reference agreement and forward consistency while penalizing invalid outputs. Direct Advantage-Based Policy Optimization updates the policy without human preference annotations or a preference-trained reward model. On USPTO-50K, RetroRLFF outperforms all 14 baselines in candidate-level round-trip accuracy at k = 1, 3, 5, and 10. With approximately 50 million parameters, it achieves 90.1% top-1 round-trip accuracy, exceeding R-SMILES, the strongest competing baseline on this metric, by 8.5 percentage points. Compared with its supervised policy, round-trip accuracy increases from 83.1% to 90.1%, while exact-match accuracy is maintained at 57.9% versus 57.5%. In an anonymized expert assessment of 131 randomly selected test cases, organic synthesis experts prefer RetroRLFF over EditRetro in 67.9% of comparisons. These results support forward feedback as a training signal for improving forward consistency while retaining top-1 reference recovery. Expert assessment provides complementary evidence of synthetic feasibility on the selected cases, although neither forward consistency nor expert preference establishes experimental feasibility.
[AI-398] GesBench: When and Where Do Multimodal Language Models Perceive Social Meaning?
链接: https://arxiv.org/abs/2610.15129
作者: Zonghuan Li,Hayley Hung,Chirag Raman
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Human social perception is incremental: we infer intent before an utterance is complete, revise our interpretation as new cues arrive, and attend to specific facial, gestural, or prosodic signals that shape social meaning. Current multimodal LLM benchmarks largely evaluate final-answer understanding over complete episodes, rewarding post-hoc recognition of social meaning rather than the unfolding perception required for real-time interaction. Drawing on interactionally embedded Gestalt principles of multimodal human communication, we introduce GesBench: a human-grounded benchmark for cue-level social perception in multimodal LLMs. GesBench uses cumulative sub-utterance clips to measure when models form stable interpretations of speaker intention and social affordance, and human gaze-guided corruption to evaluate whether regions attended by humans are functionally important for model judgments. Built from more than 6000 utterance-level clips from scripted and unscripted conversational datasets, GesBench includes human annotations of unfolding judgments and gaze on a representative subset. Across open and closed multimodal models, we find a clear human-model gap: humans form predictive understandings early and revise more often, while models often hedge or repeat across different clips. GesBench shifts social evaluation from whether models eventually recognize meaning to when and where that meaning is formed. Our codes are publicly available at this https URL.
[AI-399] o Defer or Not to Defer? Online Learning with LLM s and Congested Experts
链接: https://arxiv.org/abs/2610.15128
作者: Vasileios Saketos,William Reveillard,Alexandre Proutiere,Sebastian Dalleiger,Ming Xiao
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Modern AI systems increasingly combine large language models (LLMs) with human experts, using the LLM to handle routine queries while deferring uncertain cases for expert review. When expert capacity is limited, however, deferral decisions must account not only for the reliability of the LLM but also for congestion in the expert queue. We formulate this as an online sequential decision-making problem in which, for each query, the system observes the confidence signal of the LLM’s answer and must decide whether to output the answer or defer the query to a capacity-constrained expert. The reliability of the confidence signal is initially unknown and must be learned from expert feedback, which is both selectively observed (only deferred queries are labeled) and delayed by queuing. We propose \textttKL-PBD, a learning algorithm that uses Kullback–Leibler lower confidence bounds on the unknown reliability parameters to guide deferral decisions. For K confidence levels and horizon T , \textttKL-PBD achieves O(K\log T) gap-dependent regret when the accuracy at every confidence level is bounded away from the deferral boundary, and \widetildeO(\sqrtKT) worst-case regret. We establish matching lower bounds, showing that these rates are optimal up to logarithmic factors. Finally, we validate our approach on synthetic instances and real-world question-answering datasets, namely GSM8K, NuminaMath, and MedMCQA, using Qwen LLMs.
[AI-400] ShellElastic: Scalable Elastic Pretraining with Simultaneous Maximal Updates and Hyperparameter Transfer
链接: https://arxiv.org/abs/2610.15124
作者: Can Jin,Yan Kang,Yifan Gong,Chongjian Ge,Mingze Xu,Yuanjun Xiong,Dimitris N. Metaxas,Hongwu Peng
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Foundation models need to support diverse inference budgets, yet training a separate model for each budget duplicates pretraining effort. Elastic models train several nested sub-networks inside one weight-shared supernet, but existing recipes face three obstacles: variable per-step workloads that hinder distributed scaling, limited hyperparameter (HP) transfer across model sizes, and no guarantee that all sub-networks are maximally updated at once. We present \Ours, an elastic transformer that organizes hidden dimensions into a dense core and successive shells, so that every hidden matrix is block lower triangular. One full-width forward pass contains every sub-network’s exact hidden states as prefixes, so joint training keeps a fixed computation graph and costs about as much per step as training the widest sub-network alone. We further introduce \muE, an extension of maximal update parameterization (\mup) to \Ours and other elastic architectures. Its scaling rules provably keep all sub-networks of a supernet maximally updated simultaneously (\muI), and keep doing so when the supernet is widened or given more shells; sweeps confirm that HPs tuned on a small non-elastic proxy stay near-optimal for elastic families and larger supernets (\mbox\muT). A single \Ours run produces sub-networks that outperform MatFormer and fully elastic baselines at matched parameter counts, and approach separately trained models that together take several times the steps.
[AI-401] scHFBiMamba: a bidirectional Mamba autoencoder for cardiac disease-state classification from single-nucleus RNA-seq data
链接: https://arxiv.org/abs/2610.15117
作者: Ugo Lomoio,Pietro Hiram Guzzi,Pierangelo Veltri
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:To evaluate whether a compact bidirectional state-space autoencoder improves joint classification of species, cell type, and disease state from single-nucleus RNA-sequencing profiles in heart failure (HF), while yielding interpretable gene attributions. Building on the dense denoising autoencoder (DAE) and MLP framework of Ruz Jurado et al., we developed scHFBiMamba, combining a bidirectional Mamba autoencoder (BiMambaAE) with an MLP head. BiMambaAE maps 16,545-gene profiles to 350-dimensional latent representations using 13.2M parameters, about 14.6-fold fewer than the dense DAE (192.9M). We compared six encoder-classifier configurations on 38,275 held-out cardiac nuclei using McNemar tests with Benjamini-Hochberg correction, and assessed nine classifiers on frozen representations. Biological relevance was evaluated by comparing SHAP-derived differentially explained genes with differentially expressed genes against curated HF-related Gene Ontology keywords. BiMambaAE+MLP achieved the highest aggregate F1: pooled macro-F1 across 13 outputs was 0.966 versus 0.957 for DAE+MLP, and the mean category macro-F1 was 0.977 versus 0.970. It significantly improved cell-type and disease-state accuracy (adjusted p=7.1e-3 and p=4.8e-6, respectively), with species F1 of 1.000. SHAP-derived differentially explained genes achieved a pathway-enrichment benchmark F1 of 0.105 versus 0.053 for differentially expressed genes. Compact bidirectional state-space autoencoders can outperform substantially larger dense autoencoders for joint cardiac transcriptomic classification while reducing parameters. scHFBiMamba enables accurate multi-task prediction and hypothesis-generating gene-level interpretation. Code: this https URL.
[AI-402] Structural Chain-of-Thought: Reasoning with Executable Computational Graphs
链接: https://arxiv.org/abs/2610.15107
作者: Petr Senatorov,Karim Vafin,Elena Tutubalina,Oleg Somov
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Chain-of-thought prompting enables reasoning in large language models but produces explanations that are difficult to verify, expensive to generate, and not always faithful. We introduce a Structural Chain-of-Thought framework where models generate executable computational graphs instead of free-form reasoning. The final answer is obtained by deterministically executing the graph. The model is trained using graph-based reinforcement learning that rewards structural validity, answer correctness, and node overlap with reference graphs. Experiments across five benchmarks and four model families show that our method consistently improves accuracy over supervised graph training and can reduce generation length by up to 5-6x compared with standard GRPO and chain-of-thought supervised training. Our faithfulness experiments reveal that models trained with our approach tend to rely on deriving their answers from the generated graphs rather than producing post-hoc rationalizations as usually in chain-of-thought reasoning. Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2610.15107 [cs.AI] (or arXiv:2610.15107v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2610.15107 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-403] urnCost: Turns Not Observations: Cache-Aware Loop Engineering for Coding Agents
链接: https://arxiv.org/abs/2610.15093
作者: Pramod Kumar,Raj Kumar Pandey
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET); Machine Learning (cs.LG)
备注:
Abstract:AI coding agents operate in iterative loops: they modify code, run tools, inspect the results, and decide what to do next. We call the design of this process loop engineering, which includes context management, model selection, compaction, and termination. Existing efficiency work mainly compresses tool observations or routes selected turns to cheaper models, but the practical opportunity for these optimizations remains unclear. We analyze 600 publicly released coding-agent trajectories spanning two agent scaffolds and two language models. Median observations contain only 118-286 tokens, while the large build logs commonly used to motivate compression occur near the ninety-ninth percentile. Moreover, 98.6% of input tokens are served from a prefix cache, so flat input pricing overestimates loop cost by roughly 7x on average under near-ideal caching. We derive and empirically validate a closed-form break-even condition for model delegation. Delegation increases cost in half of the evaluated settings, near the predicted boundary. More broadly, turn count explains 84% of the variance in log trajectory cost, compared with 4% for mean observation size; the most expensive trajectories have the most turns despite having the smallest observations. These associations suggest that observation-focused methods target a secondary cost factor, and identify unnecessary turns as the larger and largely unexamined opportunity, though we do not establish that turns can be removed while holding task success fixed. We release TurnCost, a tool for re-scoring trajectory corpora under alternative caching, pricing, compression, and routing policies. Subjects: Software Engineering (cs.SE); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET); Machine Learning (cs.LG) ACMclasses: D.2.8; I.2.2; I.2.11 Cite as: arXiv:2610.15093 [cs.SE] (or arXiv:2610.15093v1 [cs.SE] for this version) https://doi.org/10.48550/arXiv.2610.15093 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Related DOI: https://doi.org/10.1145/3843779.3844635 Focus to learn more DOI(s) linking to related resources
[AI-404] Self-Evolving Visual Agent via Margin Reward
链接: https://arxiv.org/abs/2610.15084
作者: Wenxi Yang,Yuzhong Zhao,Baochang Zhang,Yunfan Liu,Qixiang Ye
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Visual agents that interleave reasoning with tool invocation have posed a promising direction for multimodal tasks. However, learning effective tool invocation usually relies on curated question-answer pairs or expert demonstrations at scale, which are expensive to acquire. In this study, we propose Marvo, a simple-yet-effective self-evolution approach that enables learning tool invocation from unlabeled images. Marvo consists of two agents: (1) a challenger that generates question-answer pairs from unlabeled images, and (2) a solver that learns tool invocation by solving questions through reinforcement learning. To encourage effective tool invocation, a good training question should be difficult enough to warrant tool use, yet remain within the solver’s capabilities. To this end, we introduce a margin reward to the challenger, defined as solver’s performance difference with and without access to tools. With the margin reward, we create an adaptive curriculum that encourages the solver to learn effective tool invocation. Across benchmarks covering agentic reasoning, visual search, mathematical, and general real-world reasoning, Marvo improves average accuracy by up to 7.1% — demonstrating margin reward as a preferred way to scale question complexity without human supervision.
[AI-405] GAMW: Generation-Aware Mel-Domain Watermarking for TTS Traceability
链接: https://arxiv.org/abs/2610.15082
作者: Yihang Zhang,Weicai Yan,Qifei Zhang,Yangyang Wu,Tao Jin
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注:
Abstract:As text-to-speech (TTS) systems continue to advance, provenance tracking and security verification of synthesized speech have become increasingly important. Audio watermarking provides active provenance by embedding imperceptible, machine-recoverable information into speech. However, existing post-hoc methods typically modify the waveform after synthesis, leaving the watermark vulnerable to compound perturbations. To address this limitation, we propose GAMW, a watermark injection framework integrated into the TTS generation process. Before waveform synthesis, GAMW embeds a content-adaptive multi-bit watermark residual into the generated representation, and its decoding objective combines focal modulation with dynamic per-bit weighting. We train and evaluate GAMW separately on two TTS backbones. With a 16-bit payload, GAMW improves decoding accuracy over the strongest baseline by 9.2 and 23.5 percentage points under individual and compound perturbations, respectively, while maintaining competitive speech quality. Reliable clean-speech decoding is also retained at higher payloads. Code is available at this https URL.
[AI-406] Knowledge transfer between simple discrete worlds
链接: https://arxiv.org/abs/2610.15081
作者: Silvère Gangloff,Jan Hula
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Adapting to a new world and reasoning correctly about which outcomes are possible under rules one has not met before is a basic capacity underlying general intelligence. We study it in small discrete worlds: sets of grid configurations consistent with a shared set of rules, concretely Sudoku-like puzzles built from a common base and a handful of extra rules. We train a single depth-recurrent model on many combinations of rules and test it at four levels of novelty: a new puzzle, a new combination of known rules, a pairing of rules never seen together, and a rule instance not seen at all during training. The model receives the active rules either as a symbolic description or as two solved examples, and a model that gets no rule information is used as a baseline. At recombining rules it was trained on, the description matches or exceeds examples ( 99% against 95% on new combinations). On rule instances whose components never co-occurred in training, the description falls below the no-information baseline while examples stay above it. The framework allows us to test many such generalization regimes in more systematic way than benchmarks such as ARC-AGI and provides new ground for evaluating depth-recurrent reasoning models, which are typically tested on saturated Sudoku benchmarks.
[AI-407] Exposing AI Inference as a Telco Network API over the 5G User Plane
链接: https://arxiv.org/abs/2610.15062
作者: Rodrigo Moreira,Larissa Ferreira Rodrigues Moreira,Joberto S. B. Martins,Tereza C. Carvalho,Flávio de Oliveira Silva
类目: Artificial Intelligence (cs.AI)
备注: Accepted for publication in the Proceedings of the IEEE Future Networks World Forum 2026 (Portugal)
Abstract:Supporting edge Artificial Intelligence (AI) inference as a network service is challenging because current mobile systems can transport tensors but do not expose or understand the inference context that gives those tensors operational meaning. Prior studies have addressed this gap through dynamic partitioning, collaborative inference, and split pipelines; however, these solutions remain application-bound, network-blind, or context-free. This work proposes Inference on Demand (IoD), an inference-on-demand abstraction that exposes split inference through a CAMARA-like Application Program Interface (API) and carries a split-aware tensor context in-band across the 5G user plane path. We evaluate the IoD on a programmable user plane function (UPF)-based testbed using a partitionable visual AI workload, showing that adaptive split selection can operate as a network-facing service while avoiding the rigid cost profile of static configurations. Including our findings, split inference becomes more than offloading when the service intent, runtime partitioning, and tensor transport are made visible to the mobile data plane. This opens a path toward network-native AI services, in which programmable telco infrastructure coordinates computation, context, and connectivity for user-facing applications.
[AI-408] race Evolution Bench: Measuring Whether Agents Improve from Their Own Environment Traces
链接: https://arxiv.org/abs/2610.15055
作者: Fangfei Li,Chenyang Zhao,Long Wang,Feng Tian,Lv Guo
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Self-evolving agents are usually scored by endpoint Gain: held-out success after an update minus success before. That difference can credit a frozen rescore as evolution. We introduce Trace Evolution Bench (TEB), a protocol that separates that rescore from a context or weight update, together with three frozen 80/20/40 packs. An update is accepted only when validation does not fall, held-out identifiers never enter the update, and a text update is not ranked with a weight update. The primary scoreboard is coding, graded by hidden unit tests in a closed workspace. On that pack (40 held-out tasks, one family), DeepSeek v4 Pro under no_evolve never changes its policy yet moves 7.5, 5.0, and 12.5 percentage points across three seeds. Of nine text-update cells inspired by GEPA, ACE, and ExpeL, seven have positive excess above that seed’s own band, and one gepa_llm seed moves down by the width of its band (-12.5, regression 0.48). GLM and MiniMax do not reproduce that seven-of-nine pattern. No method stays above its own band on every seed of all three models: GLM has two paired cells above their own bands, and MiniMax expel_llm seed 44 is a third (regression 0.30). The methods were fixed on a validation screen before held-out was read. Office, graded by rule-bounded delivery, and manufacturing table QA are separate packs: office DeepSeek does not exceed its own bands (widest 17.5 points), and manufacturing has no paired positive excess beyond its seed band.
[AI-409] Belief Embedding Tree Search
链接: https://arxiv.org/abs/2610.15046
作者: Christopher Solinas,Radovan Haluška,Martin Schmid,Michael Buro,Nathan R. Sturtevant,Michael Bowling
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Beliefs in a Partially Observable Markov Decision Process (POMDP) are posterior distributions over the hidden state given the agent’s history of actions and observations. They summarize everything the history says about the state, so planning in a POMDP is planning in a Markov decision process (MDP) over beliefs. Belief MDPs have continuous state spaces even for finite POMDPs, and exact planning in them is intractable in general. Planners must therefore approximate beliefs and estimate values. Monte Carlo planners do both inside a search tree, with an approximate belief at every node and a value estimate at every leaf. Approximations usually consist of sets of sampled states, with leaf values coming from rollouts, hand-designed bounds, or learned networks. Policies and values computed from these approximations are limited by their quality, and prior methods with learned networks rely on hand-chosen summaries of the sampled states. A belief representation should be computable from the parent node’s representation through the POMDP’s dynamics, let the search sample from it, and be learned so that it provides the information that policies and value functions need. We present Belief Embedding Tree Search (BETS), a Monte Carlo planner that meets these requirements by representing each node’s belief with a generative distribution embedding (GDE) while updating it through the POMDP dynamics. BETS trains its belief model alongside its policy and value networks through experience, with the realized hidden state of the POMDP as the belief model’s training target. We evaluate BETS on six configurations of RockSample, LightDark, and PocMan against BetaZero and three planners without learned components. BETS is competitive or leading on all six, while every baseline falls far behind on at least one. Where beliefs are highest-dimensional, it outperforms them by a large margin, even without search.
[AI-410] InCLAD: A Continual Learning Benchmark for Industrial Visual Anomaly Detection
链接: https://arxiv.org/abs/2610.15032
作者: Łukasz Marcjan,Dorota Wilk-Kołodziejczyk,Roberto Corizzo
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Industrial anomaly detection benchmarks have advanced rapidly, yet they remain largely static and therefore underrepresent deployment settings in which inspection systems must be extended from one industrial component to another over time. In our setting, continual industrial anomaly detection is framed as a task-incremental problem: each component defines a task, while the learning objective remains semi-supervised. The goal extends beyond conventional anomaly detection, since models should preserve their performance on all components as they are sequentially exposed to new components. This framing is especially relevant because many existing industrial pipelines either assume fully supervised defect labels or study only isolated datasets, leaving open the question of how semi-supervised anomaly detectors behave across a sequence of heterogeneous components. We address this gap by organizing the benchmark as a sequence of related component-level tasks and by defining multiple scenarios, including random, easy-to-hard, and hard-to-easy. In addition to single-dataset continual scenarios, we propose challenging scenarios spanning across multiple datasets. We further evaluate continual performance from two complementary perspectives: image-level anomaly detection and pixel-level anomaly localization. By recasting industrial anomaly detection as a continual benchmark rather than a static train/test problem, our work establishes a concrete path toward studying knowledge retention, adaptation, and generalization in evolving industrial environments.
[AI-411] On the Complexity Gap Between POMDPs and Multi-Environment POMDPs
链接: https://arxiv.org/abs/2610.15010
作者: Eline M. Bovy,Caleb Probine,Marnix Suilen,Ufuk Topcu,Nils Jansen
类目: Artificial Intelligence (cs.AI)
备注: 45 pages, 9 figures
Abstract:Multi-environment POMDPs (ME-POMDPs) generalize POMDPs by encoding a finite set of possible environments, capturing model uncertainty. The finite-horizon robust reward maximization problem for a ME-POMDP asks for a policy that maximizes reward in the worst environment, and recent work has shown this problem to be \mathsfPSPACE -complete. We expand the complexity landscape of ME-POMDPs by identifying which subclasses retain this \mathsfPSPACE -completeness. In particular, we take subclasses of POMDPs that are tractable in the single-environment case, and investigate the computational complexity of these classes when extended to the multi-environment setting. We first prove the \mathsfNP -completeness of multi-environment blind MDPs and deterministic POMDPs using a certain closure-under-mixture condition. We then prove \mathsfPSPACE -completeness for a range of subclasses that do not satisfy that closure condition, yielding new complexity results for multi-observation POMDPs. Finally, we show existential theory of the reals ( \exists \mathbbR )-completeness for numerous ME-POMDP problems under time-dependent Markovian policies or finite-state controllers that are \mathsfNP -complete for standard POMDPs. These results reveal a subtle relationship between POMDP and ME-POMDP complexity that depends heavily on the subclasses involved.
[AI-412] Can You Trust Your Preference Models? Stackelberg Games for Alignment
链接: https://arxiv.org/abs/2610.15008
作者: Riccardo Formenti,Aditya Agrawal,Vikas Garg
类目: Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT)
备注:
Abstract:Language models are frequently aligned to human preferences via imperfect evaluators whose errors can be exploited during optimization. We train policies to search for comparisons on which an evaluator may reverse the underlying preference, providing candidates for additional supervision. Since human preferences are unavailable during policy training, we use disagreement among an ensemble of preference models as a proxy for prediction error. We formulate this search as a zero-sum Stackelberg game between two opposing policies, a Leader which tries to minimize disagreement, and a Follower which maximizes it. For a general class of finite contextual Stackelberg games, we establish equilibrium existence and, under regularization, uniqueness and a closed-form characterization. We introduce Stackelberg Mirror Descent-Ascent (SMIDA), a two-timescale algorithm with linear last-iterate convergence in the tabular setting, and develop a neural policy-gradient analogue. Experiments with a synthetic oracle support disagreement as an informative proxy for prediction error. On held out data the trained policies outperform reference policies and increase the fraction of comparisons that exceed a disagreement threshold and reverse the oracle’s preference.
[AI-413] Recurrent In-Context Learning for Tabular Foundation Models
链接: https://arxiv.org/abs/2610.14998
作者: Tejas Vaidhya,Gaurav Sahu,Irina Rish
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Tabular foundation models adapt to new datasets through in-context learning (ICL), and, like most transformers, they tie computational depth to the number of independently parameterized layers. We ask whether the ICL stage of such a model needs a distinct set of parameters for every layer, or only depth. We introduce Tab-Loop, which replaces the 12 untied ICL blocks of TabICLv2 with 4 tied blocks applied k times and changes nothing else: the encoders, decoder, synthetic prior and training recipe are shared with an untied baseline trained for the same compute. At matched pretraining compute (18.2 EFLOP) and matched inference compute (12 block applications), Tab-Loop matches the baseline with 62% fewer parameters (10.4M vs.\ 27.6M): held-out log loss is 0.4583 vs.\ 0.4588, mean accuracy on 200 TALENT datasets is 84.10% vs.\ 83.92%, and Elo on 38 TabArena datasets is 1409 vs.\ 1396. A 4-block untied control with Tab-Loop’s parameter count, which sees 32% more training tables for the same compute, trails both (83.46%; Elo 1343). Because weights are shared, depth becomes a choice made at inference. Accuracy rises with the number of loops through the trained range and holds beyond it, and continued training with deeper loops extends the useful range, reaching the baseline’s stage-2 scores at 4–5 \times its inference compute. At this scale, the ICL stage of a tabular foundation model needs depth more than it needs distinct parameters.
[AI-414] Certify the Decision Not Every Score: Exact Top-k Selection from Few Evaluations
链接: https://arxiv.org/abs/2610.14996
作者: Yuhe Sui,Yingzhi Tang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Choosing a few robust algorithms or schedules can require evaluating every candidate under many adverse scenarios, even though an agent only needs to identify the winners. A predictor may suggest those winners, but cannot rule out a surprise in an unevaluated scenario. We introduce Certified Adaptive Selection (CAS), which returns the exact top- k set as soon as remaining uncertainty cannot change it. CAS keeps guaranteed lower and upper bounds on Expected Shortfall (ES), the average loss in the worst scenarios. Its stopping test compares the worst admissible selected candidate with the best admissible excluded one; for information-disjoint Cartesian intervals this one-cut test is necessary and sufficient. Prediction has a separate role: selecting the next outcome to evaluate, never authorizing the decision. For empirical ES with exact unit-cost queries, we compute in polynomial time the fewest evaluations that could have proved the realized decision in hindsight. Our prediction-guided agent CAS-PG is exact for every predictor, including adversarial ones. On two public algorithm-selection matrices, CAS-PG certifies the exact top three after 630 of 3150 and 981 of 2200 replayed queries, or 1.14 and 1.18 times that hindsight minimum. Without historical outcomes, its transcript-only variant requires 1.28 and 1.37 times the minimum, outperforming the tested non-predictive certified policies. Separately, a prediction-free CAS policy avoids 93.0% of exact recourse solves in synthetic energy scheduling. One certificate thus supports several acquisition policies and separately proved risk/cost extensions: prediction changes the work, not the justified choice.
[AI-415] Mixing Regularization for Averag e-Reward Q-Learning and Q-Planning
链接: https://arxiv.org/abs/2610.14988
作者: Julien Roger,Sam Bodmer,Jian Qian,Benjamin Riviere
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Optimization and Control (math.OC); Machine Learning (stat.ML)
备注: 10 pages main body, 56 pages references and appendix, 8 figures. Under review
Abstract:The average-reward setting for reinforcement learning is challenging because its Bellman operator is known to contract in span only when the underlying Markov chain is mixing. Whereas existing work typically assumes a contraction property, e.g., through environment mixing, we consider the setting where we are given a reward-agnostic m -step mixing policy, and we control the contraction and exploration properties of the learned policy by regularizing toward the mixing policy. We provide finite-time analysis of Q -learning algorithms that can be instantiated with multiple choices of regularization, with guarantees measured with respect to the regularized fixed point. Our framework exposes novel regularizer-dependent objects that control contraction, exploration, and operator bias. In the one-step setting, our rates match the best known rates for Q -iteration and synchronous Q -learning, and for m -step mixing we derive new rates. We also study Q -planning variants, and show that planning acts as time-localized learning, where the planning horizon H trades off the bias from the boundary value against the sample complexity. Finally, we present numerical studies validating the analysis: (i) knowledge of a strong mixing policy improves convergence and exploration, and its advantage increases with bottleneck severity and, (ii) our theory predicts the convergence and bias rankings of regularizers from the \alpha -divergence family.
[AI-416] Fitting Every Hint Is Not Enough: Premature Floods in Recurrent Algorithmic Reason ers
链接: https://arxiv.org/abs/2610.14982
作者: Amin Najafgholizadeh,Mohsen Afsharchi
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 33 pages, 6 figures
Abstract:Recurrent algorithmic reasoners are trained with per-step hints on small graphs and run for many more steps on large ones. On breadth-first search in SALSA-CLRS, we find that a reasoner can reproduce every training hint and still fail on deep graphs: after some exact steps, it declares nodes reached before the search gets there. We call this a premature flood and trace two routes to it. In our GRU processor with a degree input, the flood is carried in the hidden state and set off when that input is in the tail of the training distribution. Setting the input higher at inference brings the flood sooner, and none of the six flooding models counts steps. Without the input, summing the neighbors’ states instead of averaging them also breaks the processor. In the benchmark’s own GIN and PGN, trained with its code, some seeds instead flood from a fixed step, on sparse and dense graphs alike. Clamping the degree input or adding 200 dense training graphs removes every failure of the GRU processor. Detaching the state between steps avoids the flood in that processor (0 of 23 seeds, against 6 under full backpropagation) but not with sums or a residual update, and two published remedies leave floods. None of the benchmark’s test sets is both deep and dense, but its deepest one shows the second route: on its Delaunay graphs at 1,600 nodes, one GIN seed floods from step 18. We therefore recommend auditing the range of every input and aggregate, which catches the first route, and testing on deep graphs, which catches both.
[AI-417] Single-Sample Decoding of a Two-Layer Sound-Word Canvas Model for Thai Klon Pad
链接: https://arxiv.org/abs/2610.14980
作者: Worasait Suwannik
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:We present a generator of Thai klon pad verse, one stanza from a short topic, and evaluate it against human verse on held-out topics. The generator is a masked language model initialised from WangchanBERTa over a fixed canvas with two layers: a sound layer of pronounced syllables, where rhyme, tone and syllable count are enforced as hard masks, and a word layer of whole words, so that no word is cut into written pieces. Decoding commits one cell at a time and samples at temperature 0.4; one sample is drawn per topic, with no selection, reranking or repair. On 78 held-out topics never used in development, judged by three independent runs of a frozen LLM judge in the same pools as the human stanzas from which the topics were written, 70.5% of the generated stanzas pass form, lexical cleanness of all four lines and topic alignment together, against 76.9% for the human stanzas (difference -6.4 points, 95% CI [-19.2, +6.4]). Cleanness is level with human (85.9% vs. 87.2%), and the stanzas pass form and cleanness together slightly more often than the human ones (+3.8 points, [-6.4, +14.1]); the difference lies in alignment (84.6% vs. 91.0%), on topics that were written from the human stanzas. The low temperature matters: at higher temperatures the same model passes clearly fewer of the same topics, because lexical errors lie in the model’s lower-probability choices; sharpening the sampling distribution removes most of them, at a cost in vocabulary (7.5% rare words against 16.5% in the human stanzas) and diversity. The pre-registered target, all-three within 5 points of human, is missed by 1.4 points. KlonSuphap-LM, the only public klon pad generator we know of, passes none of the 78 held-out topics when judged in separate pools together with the method’s and the human stanzas (form 19.2%, cleanness 3.8%).
[AI-418] FleetPilot: Fine-Grained Steering Empowers Real-World LLM Personalization
链接: https://arxiv.org/abs/2610.14970
作者: Kounianhua Du,Jianxing Liu,Kangning Zhang,Wenxiang Jiao,Yuan Lu,Jiarui Jin,Jianghao Lin,Weiwen Liu,Yong Yu,Weinan Zhang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:With the rapid evolution of large language models (LLMs), there is an increasing demand for effective personalization techniques that can adapt model behavior to individual user preferences. Personalized LLMs must infer a user’s query-dependent preferences from limited, heterogeneous, and evolving interaction histories. Existing approaches rely primarily on in-context personalization, parametric adaptation, or personalized reward modeling, but respectively incur recurring costs in context management, model optimization and deployment, or multi-candidate generation and scoring. These limitations become particularly restrictive when user observations are sparse and the preference evidence relevant to each query varies across contexts and over time. To address these challenges, we propose \modelname, a fine-grained and instance-tailored steering framework that dynamically generates sample-level interference vectors from user data and injects them into the model’s forward pass for personalized adaptation. Our approach introduces: match contrast that produces user-relevant signals, fine-grained steering component that captures nuanced signals by hooking activations from attention and MLP layers, and query-conditioned aggregation module that synthesizes these signals into contextually relevant enhancements. Extensive experiments across diverse scenarios, including short-to-long text generation, and web function calling, validate the effectiveness and compatibility of \modelname. Results show that our method consistently brings significant performance gains, and show extreme superiority in fast-shifting and data-sparse environments. Additionally, \modelname rids us of training cycles and incurs light overhead of storage or inference latency. These make \modelname a strong and real-world deployable method. Implementation is available \footnotethis https URL.
[AI-419] SkillHorizon: Constructing Long-Horizon Agent Tasks through Skill Composition
链接: https://arxiv.org/abs/2610.14966
作者: Yuxuan Liu,Zhaochen Su,Hangyu Guo,Shijue Huang,Zhixiong Zhang,Kawai Chung,Zhongwei Xie,Tianshi Zheng,Wei Fan,Qing Zong,Haochen Shi,Baixuan Xu,Haoran Li,Yangqiu Song
类目: Artificial Intelligence (cs.AI)
备注: 29 pages, 7 figures
Abstract:Long-horizon agents must complete workflows in which later activities depend on earlier outputs and changes to shared state. A key challenge for evaluation is to expand task requirements and their dependencies while keeping tasks executable and outcomes verifiable. We introduce SkillHorizon, a benchmark that constructs long-horizon tasks through skill composition in seven business environments. Its generators use domain-specific rules to turn reusable skill procedures into concrete business activities. These activities are connected through input-output dependencies, shared resources, and state updates, allowing workload and coordination demands to grow together. Successful reference executions validate task feasibility and measure interaction workload, while programmatic evaluators assess local completion and whole-task success. We evaluate five language models on 1,050 tasks with reference interaction workloads ranging from 64K to 1M tokens. Across these scales, mean local-goal completion falls from 90.4% to 46.8%, as completed work fails to keep pace with growing requirements. At larger workloads, agents more often submit unsuccessful final answers with substantial work unfinished and tool-call allowance remaining. Trace analysis also identifies cases where scripts execute successfully but their outputs fail downstream business checks. SkillHorizon provides a configurable testbed for studying sustained progress, dependency coordination, and complete task delivery in growing workflows. Our code is available at this https URL.
[AI-420] Humans vs. Machines: A Comparison With AI Agents in Software Reverse Engineering
链接: https://arxiv.org/abs/2610.14960
作者: Samuele Doria,Stefano De Rosa,Ruoyu Wang,Zion Leonahenahe Basque,Eleonora Losiouk,Simone Aonzo
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:Can an autonomous AI agent reverse engineer a binary as well as a skilled human analyst? Do they adopt the same strategy, and can agents scale to real-world programs? We compare human and LLM-based software reverse engineering (SRE): nine models tackle two compact static-analysis challenges previously solved by human analysts, scored with a common understanding rubric, and we record their complete interactions with IDA Pro. This lets us compare not only what humans and agents recover, but also how they investigate a binary. The best agents (GPT-5.5 Pro, Claude Opus 4.8, and Kimi K3) achieve statistically human-competitive understanding on both tasks, while weaker models remain far behind. Ablations show that the gap depends on code representation and workflow demands, not on model capability alone: open-weight models improve substantially with richer code representations, whereas frontier models show no significant change. Similar success rates hide different workflows. Humans repeatedly revisit functions and follow local program structure; strong agents jump across the binary, retrieve information broadly, and rarely return to inspected code, a behavior we call greedy fetching. This strategy is not fixed: given semantically richer code, agents fetch less without losing understanding. Because greedy fetching aggregates large amounts of code in the agent’s context, we measure context-window saturation across 4,000 real-world binaries: large windows accommodate many programs, but larger applications and lower-level representations quickly exhaust them. Autonomous reverse engineering is becoming competitive, with a large speed advantage, but scaling it will require better navigation, selective retrieval, and persistent analysis state, not merely larger context windows.
[AI-421] MCPDisCo: Seesaw-Mitigated Tuning for MCP-Native Agents
链接: https://arxiv.org/abs/2610.14956
作者: Kounianhua Du,Jianxing Liu,Shuai Shao,Ke Li,Xing Sun,Weinan Zhang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:The Model Context Protocol provides a shared interaction substrate for the agentic web, allowing heterogeneous services to expose capabilities through a standardized interface. Learning to use this ecosystem entails a hierarchy of capabilities: tool-format skills, server-level transfer, and cross-server composition. We present MCPDisCo, an ecosystem-scale reinforcement learning framework for MCP-native agents. Using a Thought-Markov Decision Process (Thought-MDP), the framework explicitly represents internal reasoning alongside routing, execution, and termination and trains on real-server, long-horizon interactions spanning 70 MCP servers and 527 tools across six domains. We study the reasoning–tool seesaw, a joint-learning failure mode in which improvement in one objective can accompany deterioration in the other. To mitigate potential interference, MCPDisCo combines token-masked reward attribution with bidirectional gradient surgery: the former separates direct supervision for reasoning and action spans, while the latter removes conflicting components from both objective gradients in shared parameters. MCPDisCo achieves higher success with fewer chain-of-thought tokens. On the out-of-distribution MCP-Atlas and the GAIA benchmark, MCPDisCo maintains monotone gains, confirming that the learned behavior generalizes beyond the training server distribution and broader agentic tasks. Model weight and implementation are available.
[AI-422] Counterfactual Rollouts for Offline Contrastive Goal-Conditioned Reinforcement Learning
链接: https://arxiv.org/abs/2610.14953
作者: Tingrui Huang,Ai Bo,Junzhe Zhang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Contrastive reinforcement learning (CRL) learns goal-conditioned values by associating state–action pairs with the states subsequently reached, assuming that no unobserved variable jointly affects the recorded action and its outcome. When offline demonstrations come from a demonstrator with information hidden from the learner, this assumption fails and the critic absorbs spurious action–outcome associations. We formalize this setting with confounded Markov decision processes and introduce counterfactual transitions that separate what the data identify from what they do not; minimizing over all compatible transitions yields a worst-case occupancy that lower-bounds the true one. Causal Contrastive Reinforcement Learning (CauCRL) learns a pessimistic counterfactual transition from positive and negative offline demonstrations and trains the critic on counterfactual rollouts, substantially improving success rates over standard CRL on confounded goal-reaching tasks.
[AI-423] Probabilistic Multi-Source Data Selection via Learnable Categorical Priors: Variance Reduction and Guaranteed Derandomization
链接: https://arxiv.org/abs/2610.14942
作者: Shizhan Cai,Qixin Zhang,Siyuan Ma,Siyuan Liang,Yongcheng Jing,Dacheng Tao
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Bilevel data selection is a fundamental task with numerous applications in machine learning and statistics. A common strategy for tackling this problem is to relax the original bilevel combinatorial optimization into a continuous one. However, existing methods, including 0-1 relaxation and probabilistic relaxation, still suffer from several limitations, such as unstable gradient estimation and infeasible rounding. To address these issues, this paper proposes a new probabilistic relaxation for the multi-source bilevel data selection problem. Our key idea is to introduce multiple learnable categorical priors to parameterize a probability distribution supported only on feasible data subsets, thus naturally yielding a lossless rounding scheme. We further show that our proposed relaxation admits an effective two-point gradient estimator, which can simultaneously avoid the inverse-Hessian computation required by existing 0-1 relaxation and the unbounded score term appearing in prior probabilistic relaxation. Furthermore, we develop a reducible representation of this estimator, which substantially reduces the number of lower-level retrainings required for gradient computation. Built upon this reducible two-point estimator, we then design a novel stochastic gradient algorithm, termed ReSGD, for multi-source bilevel data selection problem, and establish a rigorous approximation guarantee for the subset returned by ReSGD under weak submodularity. Finally, extensive experiments are conducted to validate the effectiveness of our proposed ReSGD.
[AI-424] VC-OPD: Visual Counterfactual On-Policy Distillation for Grounded Vision-Language Reasoning
链接: https://arxiv.org/abs/2610.14939
作者: Jiahe Feng
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:GRPO-style reinforcement learning has advanced vision-language reasoning, but its sequence-level rewards provide limited credit assignment for individual reasoning tokens. On-policy distillation (OPD) addresses this issue by providing dense supervision on student-generated trajectories, yet standard OPD assigns similar importance to all teacher-preferred tokens, including template phrases and visually irrelevant continuations. We introduce Visual Counterfactual On-Policy Distillation (VC-OPD), a grounding-aware distillation method that prioritizes tokens whose correctness depends on image-specific evidence. After a teacher-supervised warm start that stabilizes response formats and reduces noisy rollouts, VC-OPD performs on-policy distillation with two teacher evaluations for each student trajectory: one under the original image and one under a counterfactual image with corrupted instance-specific evidence. The original-image teacher distribution determines the distillation target, while the original–counterfactual teacher gap estimates token-level visual dependence and softly reweights the loss. To avoid unstable response-level update scales, VC-OPD further applies mass-preserving normalization to the token weights. This design preserves the dense supervision of OPD while shifting learning capacity toward visually grounded reasoning tokens. Experiments on multimodal reasoning benchmarks show that VC-OPD improves visual reasoning performance and produces more effective token-level distillation.
[AI-425] Bridging Risk Approximation Gaps in Model Predictive Task Sampling via In-Context Modeling
链接: https://arxiv.org/abs/2610.14913
作者: Jiarong Wen,Qi Tao,Kaiyu Zhang,Yun Qu,Lizhou Cai,Heming Zou,Yuhang Jiang,Qi Wang
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Reinforcement learning (RL) is essential for adaptive decision-making, e.g., robust Meta-RL and domain randomization (DR), and efficient post-training of large language models (LLMs), where tasks are commonly structured as Markov decision processes or prompts. A central bottleneck in all these settings is the cost of agent-environment interactions required for task selection and policy optimization. Model predictive task sampling (MPTS) has emerged as a promising approach to improve sample efficiency by querying informative tasks via a risk predictive model (RPM) trained on optimization history. However, existing RPMs suffer from two failure modes: (i) sparse optimization histories, which limit RPMs’ generalization across the task space, and (ii) a one-step lag in risk approximation, which causes biased difficulty estimation as the policy evolves. This work formalizes these failure modes, derives lagged-risk decompositions that expose the resulting temporal bias, and recasts risk prediction as an in-context sequence modeling problem. The developed RPM integrates a one-step look-ahead mechanism with a temporal-weighted sliding window to mitigate data non-stationarity and scarcity, without requiring additional rollouts. Empirically, our approach yields more reliable difficulty estimates and consistent performance gains across robust Meta-RL, DR, and prompt curriculum RL post-training of LLMs.
[AI-426] Probable but Not Valid: A Psychometric Audit of Decision Models as Synthetic Survey Respondents
链接: https://arxiv.org/abs/2610.14912
作者: Mantas Lukauskas,Viktorija Šarkauskaitė
类目: Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Applications (stat.AP)
备注: 36 pages, 6 figures, 16 tables. Code, raw model outputs and tables are available from the corresponding author on request
Abstract:Language models are increasingly used to create synthetic survey samples, also called silicon samples. Decision models are a new class of language models that return a probability for each point of a rating scale instead of text. We ask whether these models, or the newest generative LLMs, can stand in for human respondents on a psychometric survey. We test fourteen decision models from eleven providers and 47 generative LLMs on an open Lithuanian organisational-psychology survey (n=263, 68 items), with 100 real respondent profiles, the Psychometric Similarity Score (PSS) and statistical baselines. The best generative model, MiniMax M3, reaches PSS 0.813. This is level with a Gaussian copula fit only on other respondents (0.824) and below a copula fit in sample (0.927). A second sample of 100 real people scores below the best models, so at this size PSS cannot rank them against real respondents. In 5 of 10 product lines the newest generative release scores below an earlier release of the same line beyond resampling noise, and none gains beyond it. With the default rounded readout, 13 of 14 decision models score below the generative reference gpt-5.4-mini. Of the 47 generative models, 33 reproduce the pattern of item correlations better than any decision model. Averaged over respondents, the probability vectors are much closer to the human item marginals than rounded answers. They also rank respondents about as well as a profile regression fit on 163 real people. As forecasts of each person’s own answers, however, every decision model is worse than the answer frequencies of other respondents, mainly because its average vector differs from them. The profile regression improves on these frequencies by only 1.0%, so the profile fields say little about how a person answers. Synthetic respondents built on such profiles should therefore aim at sample-level fidelity.
[AI-427] Beyond Outcome Feedback: Trajectory-Quality-Guided Experience-based Learning
链接: https://arxiv.org/abs/2610.14877
作者: Bingjie Zhang,Yibo Yang,Qiantong Liu,Benshuo Fu,Yi Chang,Dandan Guo
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Experience-based learning enables large language model (LLM) agents to improve over time by reusing knowledge from past executions. However, existing methods typically rely on final task outcomes as feedback for experience learning, overlooking the execution process and limiting their ability to evaluate and optimize experience based on trajectory quality. To address this limitation, we propose TQG-Exp, a trajectory-quality-guided experience-based learning framework that introduces process feedback. Specifically, we introduce Trajectory Quality Loss (TQL), which jointly considers task completion and process quality to provide a learning signal beyond final task outcomes. Based on this signal, we employ paired rollouts with and without retrieved experience to guide experience optimization. Representative trajectories are selected based on their outcomes and trajectory quality to extract task-level and task-type-level experiences. Meanwhile, differences in trajectory quality between the paired groups are used to estimate the contribution of retrieved experience and update its utility. The learned utility is then combined with semantic relevance to guide subsequent experience retrieval. Experiments on ALFWorld, WebShop, and ScienceWorld with two backbone LLMs show that TQG-Exp outperforms existing experience-based methods, improving Avg.@3 over the strongest baseline by 4.50 and 8.55 percentage points on DeepSeek-V4-Flash and Qwen3.5-Flash, respectively.
[AI-428] WeatherBridge: Forecast-Aware Multi-Source Input Adaptation for Pretrained Weather Forecasting Models
链接: https://arxiv.org/abs/2610.14863
作者: Mutian Xi,Xinyu Su,Xiaohui Zhong,Hao Li
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Data-driven weather forecasting models are commonly trained on ERA5, yet operational deployment must use timely initial states from physics-based numerical weather prediction (NWP) systems whose distributions differ substantially from ERA5. Direct initialization with these states can degrade autoregressive forecasts, motivating input adaptation that keeps the pretrained forecasting model unchanged. Here, we introduce WeatherBridge, a unified, source-conditioned adapter that maps heterogeneous operational NWP initial states to ERA5 reference states without updating the pretrained forecasting model. WeatherBridge predicts the structured residual corrections with a Swin-UNet and injects source embeddings across multiple spatial scales, allowing the model to capture shared atmospheric structure and source-specific discrepancies. Its training objective combines latitude-weighted ERA5 reconstruction with One-step Rollout Alignment (ORA), which backpropagates one-step forecast error through a frozen FuXi model to optimize the adapted states for both initial state fidelity and downstream forecast compatibility. We evaluate WeatherBridge using daily GFS, HRES, and CMA initializations from 1 January to 22 November 2025 across six variables and lead times up to 240 hours. Relative to raw source initialization of FuXi, WeatherBridge reduces the mean RMSE of FuXi over the 6- 240-hour rollout for 17 of 18 source-variable combinations, with a median reduction of 10.2% and a maximum reduction of 46.3%. The adapter trained with frozen FuXi also improves FengWu forecasts on GFS inputs without retraining either the adapter or FengWu. These results demonstrate that forecast-aware input adaptation can improve forecasts from heterogeneous operational NWP inputs and transfer across pretrained ERA5-based forecasting models.
[AI-429] PEAM: On-device Communication Agent for Personalized Efficient and Autonomous Mobile Network Configuration
链接: https://arxiv.org/abs/2610.14855
作者: Haowen Wan,Shangzhuo Xie,Syue-Ju Syue,Xianyue Wu,Runhua Li,Qianqian Yang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:The growing deployment of compact large language models on smartphones is enabling autonomous on-device decision-making, such as emerging mobile GUI agents. However, many mobile decisions, such as network configuration when users switch apps or connectivity conditions change, must satisfy stringent latency constraints to preserve user experience. In this paper, we propose PEAM, an on-device LLM-based communication agent that autonomously generates personalized network-configuration rules executable within milliseconds. Specifically, PEAM first fine-tune the on-device LLM with a dataset constructed from manually annotated real-user data to integrate the communication knowledge into the local model. PEAM then operates in two phases. In the rule-generation phase, which is performed when the device is idle, such as overnight, PEAM analyzes historical network-configuration records in batches and reflects across batches to consolidate consistent evidence, reduce local bias, and derive robust configuration rules. To enable more accurate and stable rule generation over time, PEAM further constructs a user profile from high-confidence rules to capture the user’s network preferences and incorporates this profile into each rule-generation cycle. In the rule-execution phase, PEAM is invoked when the communication context changes, such as when users switch apps or connectivity conditions change, and searches the rule library to identify and execute the appropriate rule. Experiments demonstrate that PEAM, based on a fine-tuned 4B model, achieves 99.13% correct network configuration decisions, while 92.50% of the rules generated by PEAM are valid, outperforming the traditional expert-rule based method by 4.41% and 84.50%, respectively.
[AI-430] Pre-Registered Genesis Accounting for Recursive Self-Improvement: First Audited Operator-Genesis Events in a Physics-Discovery Runtime
链接: https://arxiv.org/abs/2610.14843
作者: Wenbin Mao,Pei Zou
类目: Artificial Intelligence (cs.AI)
备注: 7 pages, 1 figure
Abstract:Claims of recursive self-improvement are easy to make and hard to falsify: did improvement come from re-using designer-provided mechanisms, or from mechanisms the system authored itself? We introduce a pre-registered, two-tier accounting: M1 counts machine-generated content instances within a designer-provided vocabulary, while M2 counts machine-authored mechanism types (new operators registered into the runtime’s own vocabulary, with a held-out-validation certificate). We report the first audited M2 genesis events. In the primary event, depth-limited compositional search over elementary operators, judged by held-out R^2 on real NIST hydrogen Balmer measurements, produced the operator sq(inv(n)) = 1/n^2 – the Rydberg-level form – which no standard-library transform could express (held-out R^2: -0.546 best standard transform vs. 1.0000000 composed). A protocol flaw (independent variable built by inverting the very law under test) was caught in audit and repaired via an independent-label protocol; extrapolation to unseen levels (n=8,9 trained on n=7) attains relative error ~10^-6, four to six orders beyond the best standard operator. A second event independently recovered sqrt(inv(a)) – Kepler’s velocity law – from JPL ephemeris vectors. Also incorporated: an audited compounding channel (first domain-level measurement: 1.109x accuracy ratio on a real held-out domain) and blind, decurated genesis exams whose honest tally is zero. The Kepler event passes line audit; the hydrogen gap reproduces on He II; the certificate is threshold-stable. Against eleven prior systems, the contribution is the measurement and governance of machine-authored mechanisms, not the search algorithm or rediscovered physics. Limitations: exams were curated by agents who knew the textbook answer, triggers manual, grammar observer-injected; both events demonstrate capability to author, not yet autonomous genesis.
[AI-431] EvoCowork: A Benchmark for Real-World Office Tasks
链接: https://arxiv.org/abs/2610.14828
作者: Zhou Liu,Xingyan Chen,ZhaoYang Han,Zewei Pan,Zeli Su,Ningxin Pan,Xue Hu,Kaixiong Ting,Wentao Zhang
类目: Artificial Intelligence (cs.AI)
备注: 32 pages
Abstract:Office workflows are a demanding testbed for self-improving agents because their deliverables must be correct, natively editable, and mutually consistent. A central question is whether gains on feedback tasks reflect reusable procedures rather than task-specific fixes. We introduce EvoCowork, a 240-task benchmark for evaluating harness-level recursive self-improvement across four office modalities, with model weights held fixed. The benchmark covers presentations, documents, spreadsheets, and cross-office workflows, and scores submitted artifacts for content, native structure, formulas, cross-file consistency, and visual quality where needed. Its protocol separates feedback, revision selection, and final testing: methods learn from single-modality tasks, then are evaluated on unseen single-modality tasks and cross-office workflows under a shared initial harness and explicit budgets. In static evaluations of two models and four harnesses, the best configuration scores 42.25 overall and 30.62 on cross-office tasks. Starting from the same MSA harness with GLM 5.3, ACE and AHE improve in-domain scores by 4.90 and 2.64 points, respectively, but change cross-office scores by only +0.03 and -1.72. Thus, gains on the file types that provide feedback do not necessarily transfer to coordinated workflows. EvoCowork makes this distinction explicit while recording the capability, transfer, and cost of harness evolution. Code and task packages: this https URL.
[AI-432] AI-Native Product Management: A Vision
链接: https://arxiv.org/abs/2610.14823
作者: Ricardo Britto
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:Artificial intelligence is increasingly used to support product management tasks. At the same time, AI-native research and development (RD) may increase software realization capacity. Organizations may realize software faster than they can identify, qualify, and reconsider what should be realized. This paper asks how product management should be re-envisioned under these conditions. We synthesize research on software product management, product learning, human-AI collaboration, and the Product Management and RD boundary. We propose a vision in which AI becomes a persistent participant in product work. The vision covers sensing, exploration, decision-making, realization, and learning. We propose six principles and make the vision tangible through an operating model grounded in human professionals, AI personas, agents, skills, and a shared product context. We further argue that AI-native Product Management and AI-native RD can form a continuous product-learning process. Finally, we identify four AI-specific research directions that require empirical investigation.
[AI-433] A Reference Architecture for Integrating Agent ic AI into Enterprise Software: Responsibilities Interactions and Variability
链接: https://arxiv.org/abs/2610.14818
作者: Mingqin Yu,Fethi Rabhi,Liming Zhu,Qinghua Lu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Agentic AI moves generative AI applications toward systems that coordinate reasoning, enterprise context, model inference, and external actions. Integrating these capabilities into enterprise software introduces diverse architectural concerns, from orchestration, knowledge and memory, and model access to tools and actions, security, runtime infrastructure, and evaluation. Existing research and industry architectures address many of these concerns but differ in scope, abstraction, responsibility organization, interaction representation, and assumptions about which capabilities are required. This paper presents a technology-neutral reference architecture for enterprise agentic AI integration. We synthesize architectural capabilities from research publications, industry architecture guidance, and interoperability specifications by architectural purpose, organize them into ten responsibilities, and define boundary invariants, typed cross-responsibility interactions, and variation points that distinguish responsibilities from condition-dependent capabilities and replaceable realization mechanisms. We instantiate the architecture as an ESG portfolio-analysis and pre-trade compliance agent over an existing service-oriented ESG metric computation and reporting system, and complement the case with a mapping of an independently published agentic architecture. The case realizes seven of the ten responsibilities; different questions take different interaction paths, controls activate only under their conditions, and the probed boundary invariants hold. The independent architecture maps onto the proposed responsibilities without requiring an additional one. These results provide initial evidence for the architecture’s organization and applicability rather than its completeness. The deployment and recorded results are available as a replication package.
[AI-434] Seamless Deployment of Machine Learning Inference at the Far Edge: An Empirical AWS IoT Greengrass Architecture for Industrial Predictive Maintenance
链接: https://arxiv.org/abs/2610.14815
作者: Prince Kwabena Appiah Boadu,Jeffery Baafi Kyei,Enoch Kofi Koranteng,Griffth Selorm Klogo
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Artificial Intelligence (cs.AI)
备注: Conference paper draft in IEEE format, with code appendices. Already accepted in AiBThings - IEEE 4th International Conference on Artificial Intelligence, Blockchain and Internet of Things
Abstract:Industrial IoT sensors produce continuous, high-volume data streams that predictive maintenance needs to analyse in real time. Cloud-centric architectures send all raw telemetry to central servers, which costs latency, bandwidth and resilience. This paper presents a hybrid cloud-edge architecture for manufacturing predictive maintenance built on AWS IoT Core and AWS IoT Greengrass v2, with local machine learning inference at the edge and selective cloud synchronization. We deploy two configurations on real AWS infrastructure and compare them: a cloud-only baseline (Configuration A) and the proposed hybrid edge architecture (Configuration B). We measure end-to-end inference latency, bandwidth consumption and inference accuracy. Configuration B cuts cloud-bound bandwidth by 86.8% (63.6 KB to 7.2 KB per 200-reading window) and decision latency by 256x (23.1 ms cloud round trip to 0.09 ms local inference). At the edge, with no cloud involvement, it keeps an anomaly-class F1 score of 0.84 and 91% recall. The evaluation also instruments system uptime and Stream Manager queuing under cloud disconnection; the quantitative disconnection experiment is left as planned work. Our review of the literature points to four gaps that this work addresses: the lack of empirically validated Greengrass v2 deployments for predictive maintenance, unquantified bandwidth savings from selective edge synchronization, unmeasured resilience under cloud disconnection, and the open MLOps feedback loop at the edge.
[AI-435] ASIWorld: Benchmarking Autonomous Agents in a Persistent Survival Sandbox
链接: https://arxiv.org/abs/2610.14807
作者: Jianfeng Lin,Runze Wang,Haorong Hong,Yuxuan Xie,Zeno Zeng,Zhongyuan Liu
类目: Artificial Intelligence (cs.AI)
备注: 15 pages, 5 figures, 9 tables. Project website: this https URL
Abstract:Evaluating autonomous agents requires examining whether they can translate goals into coherent action across extended interactions. In persistent environments, this requires coordinating interdependent actions under incomplete observations, accounting for delayed effects, and adapting as earlier decisions shape subsequent opportunities. We introduce ASIWorld, a survival sandbox that combines persistent world and character state with character-scoped observations and a programmable asynchronous interface to study sustained goal execution. Agents pursue interdependent production and exploration tasks through direct actions or agent-authored programs that continue running between model calls. The evaluation assesses task success, the attainment of development achievements, and survival costs, complemented by an analysis of execution behavior. In an exploratory study of nine agent configurations using a common harness, eight complete the production task and six complete both tasks. Analyses of execution trajectories and performance across difficulty levels reveal different preparation, combat, and recovery behaviors among configurations with similar task outcomes. Incomplete preparation is associated with stalled progression, while configurations that complete the tasks differ in survival costs and the fulfillment of additional achievement criteria. These findings illustrate the value of assessing sustained goal execution through both task outcomes and the behaviors and costs associated with achieving them.
[AI-436] CESNET-TimeSeries26: A Multi-Year Network Traffic Time Series Dataset for Annual Seasonality Concept Drift and Long-Term Anomaly Detection
链接: https://arxiv.org/abs/2610.14799
作者: Josef Koumar,Karel Hynek,Kamil Jerabek,Pavel Siska
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Networking and Internet Architecture (cs.NI)
备注:
Abstract:Forecast-based models are fundamental to network traffic management and anomaly detection. However, it is essential to evaluate these forecasting models over extended periods of time in order to assess their adaptability to unexpected phenomena and shifts in underlying patterns. This kind of long-term evaluation requires multi-year data to account for yearly seasonality and behavioral drift. Existing public datasets are too short to support such long-horizon forecasting tasks. We introduce CESNET-TimeSeries26, comprising 995 days of continuous traffic time series from the CESNET3 network, together with a single subnet spanning approximately 4.5 years. The dataset captures 278,466 IP addresses, 285 institutions, and 566 institutional subnets at three aggregation intervals. The dataset provides a definitive longitudinal benchmark for evaluating forecasting models in both operational management and security monitoring. We demonstrate that annual seasonality is the dominant periodic component. Moreover, we show that a forecasting model can exploit the annual component only when its training window covers at least one complete annual cycle, which no sub-year dataset provides.
[AI-437] GREAT-ESDD: Graph Edge Attention Network for Effectively Steered Delivery Decisions
链接: https://arxiv.org/abs/2610.14792
作者: Attila Lischka,Ze Zhou,Xuegang(Jeff)Ban,Balázs Kulcsár
类目: Artificial Intelligence (cs.AI)
备注: 28 pages, 12 figures
Abstract:Online, same-day delivery requires operators to constantly update their routes, and the routing cost for deliveries increases with the number of customers. Out-of-home (OOH) delivery options can help companies handle increasing customer numbers while keeping delivery costs low. We develop a dynamic pricing-and-routing framework for this online ordering scenario with home and OOH delivery options. First, we develop a neural solver for the capacitated vehicle routing problem (CVRP) based on the Graph-Edge-Attention-Network (GREAT). We train it to solve CVRP with varying instance sizes and starting conditions, allowing it to reoptimize partially executed routes under changing conditions. It is fast while achieving solution quality competitive with traditional baselines. Our solver processes several CVRP instances in parallel in batches and we use it to solve the candidate routing problems that arise from an incoming customer’s choice of home or OOH delivery. This provides an estimate of the delivery cost increase per option, which we use to set dynamic prices. Further, it directly provides the updated route once the customer decides. We trade off delivery fees against routing costs and steer customers toward OOH delivery when home delivery would be costly. However, effective steering requires us to know the customer preferences. To learn those online, we use Thompson sampling with a Laplace-approximated posterior. We call the resulting framework GREAT-ESDD (GREAT for Effectively Steered Delivery Decisions). Compared to a traditional routing solver, it increases profit by around 11% under a similar pricing decision time budget. Moreover, we show that we can estimate customer preferences successfully, outperforming a static pricing strategy after only 5 days of simulated operations.
[AI-438] Counterfactual Regret for Reflective Agents with Lean-Verified Guarantees
链接: https://arxiv.org/abs/2610.14790
作者: Zhongnian Li,Shi Xu,Mengting Xu,Peng Ying,Xinzheng Xu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Reflective language agents reuse feedback and memory to adapt across tasks without updating model weights. However, existing guarantees for policy selection and action generation do not by themselves establish when local reflection yields long-term performance guarantees for the complete agent under resource constraints. In this paper, we address this gap by deriving sufficient conditions for sublinear counterfactual regret from local generation and deployment guarantees, while retaining the native controller. Specifically, we establish a resource-sensitive transfer that bounds full-state deployment discrepancies along actual execution paths and preserves the own-history laws of both learner and comparator. Besides, local generation reliability yields a cumulative generation opportunity bound, linking policy acquisition to long-term performance through an explicit quality, coverage, and hard-budget tradeoff. To verify these guarantees, we formalize the conditional proof chain for all thirteen numbered theoretical results in Lean, including a native instance with repeated policy growth and imperfect validation.
[AI-439] LYNX: Analytical Program Synthesis via Grounding and Closed-Loop Refinement
链接: https://arxiv.org/abs/2610.14777
作者: Ruichen Xia,Xiao Han,Dingyu Yang,Luwei Yang,Sai Wu,Dongxiang Zhang,Binhui Yao,Gang Chen
类目: Artificial Intelligence (cs.AI); Databases (cs.DB)
备注:
Abstract:LLM-powered data analytics aims to translate natural language instructions into executable analytical programs over underlying data. However, existing approaches, whether treating LLMs as intent-to-code translators or as standalone analytical engines, often fail to reliably align domain semantics with complex schemas and heterogeneous metadata, resulting in brittle execution and semantically inconsistent analytical outputs in real-world workloads. We present LYNX, a multi-agent framework that formulates LLM-driven analytics as a schema- and knowledge-constrained analysis problem rather than a single-pass generation task. LYNX organizes analytical synthesis into a closed-loop workflow of planning, generation, verification, and refinement through three role-specialized agents that collaboratively enforce schema grounding, semantic consistency, and execution correctness. To address domain ambiguity and schema misalignment, LYNX integrates Semantic-Structural Distillation and Stage-Aware Allocation (SSD-SAA). A delta-guided iterative revision continuously diagnoses and repairs semantic and execution-level failures, substantially improving synthesis robustness. We evaluate LYNX on two NL-to-SQL benchmarks, BIRD-SQL Mini-Dev and the more challenging Spider 2.0-Lite. Experimental results demonstrate strong performance and efficiency, achieving 10.66% to 59.54% execution accuracy across lightweight and commercial models, while reducing token consumption by 70.53% compared with the prior most efficient method. Our framework displays high generalization over different backbones, from open-source, lightweight models to the latest commercial models. Additional case studies on quantitative finance analytics and complex multi-relational SQL workloads further demonstrate the framework’s robustness, scalability, and practical applicability in real-world analytical scenarios.
[AI-440] raining a Compact Strong and Realistic Text-to-Speech Model
链接: https://arxiv.org/abs/2610.14776
作者: Runyuan Cai,Yu Lin,Yiming Wang,Xiaodong Zeng
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:
Abstract:Text-to-speech (TTS) systems are conventionally optimized for intelligibility and speaker similarity. However, neither objective captures the prosodic detail that distinguishes human performances from synthetic speech, leaving a realism gap that larger models and higher-quality data can narrow but not close. In the absence of an objective that directly targets this gap, realism remains a persistent challenge in TTS. We present Audio8-TTS, a compact 0.6B-parameter model for zero-shot voice cloning that achieves state-of-the-art performance in intelligibility and speaker similarity, with particularly strong results in perceptual naturalness. To align the training signal with human perception rather than signal-level fidelity, we construct a quality ladder from widely used open-source TTS systems and train a realism critic to assess generated speech against matched human recordings of the same sentences. An identity-erasure objective discourages the critic from relying on generator fingerprints. We train the model in three stages: multilingual supervised fine-tuning; group relative policy optimization (GRPO) with word error rate (WER) and speaker similarity (SIM) rewards; and GRPO guided by the realism critic. The final stage improves perceived prosodic naturalness and speaker similarity while preserving WER performance.
[AI-441] ProgAgent : Learning Tool Orchestration through Programmatic Training
链接: https://arxiv.org/abs/2610.14774
作者: Hao Li,Keqing He,Longtao Zheng,Junlong Li,Weihao Zeng,Qian Liu,Yuzhen Huang
类目: Artificial Intelligence (cs.AI)
备注: Project page: this https URL
Abstract:Large language models (LLMs) are increasingly expected to orchestrate diverse tools over long-horizon tasks. Existing agents typically use Direct Tool Calling (DTC), where the model invokes tools sequentially, and tool responses accumulate in the model context, which can lead to context rot and orchestration breakdown. In contrast, Programmatic Tool Calling (PTC) invokes tools through executable code and stores intermediate results in runtime variables rather than appending them to the context. Prior work has explored PTC mainly as an inference-time optimization, but its benefits for agentic training remain unclear, and suitable verifiable training tasks are scarce. In this work, we introduce ProgAgent, a full PTC training stack, comprising a pipeline for synthesizing verifiable long-horizon tasks, a lightweight PTC runtime, and a two-stage training recipe. Using only 215 tasks instantiated at four environment scales, ProgAgent yields higher pass rates than the evaluated baselines of comparable size across four benchmarks. On LOCA-bench, ProgAgent-14B achieves a 30.7% pass rate at a 64k-token environment state, where these baselines score zero. Cross-harness experiments show that PTC-trained models outperform DTC-trained models under both evaluation interfaces, suggesting that the benefits of PTC training extend beyond the interface.
[AI-442] Prior-Fitted and Frozen Scorers for Multimodal Contrastive Learning
链接: https://arxiv.org/abs/2610.14770
作者: Tillmann Rheude,Roland Eils,Benjamin Wild
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Multimodal contrastive learning typically relies on manually specified scoring functions, such as cosine similarity, to quantify the similarity between representations from different modalities. Several such functions have been recently proposed to extend contrastive learning beyond two modalities. In principle, however, any function that assigns high compatibility to matched samples and low compatibility to mismatched ones could serve as a scorer. Motivated by this broader view, we instantiate the scorer with a Transformer and find that even a randomly initialized model, when kept frozen, can outperform strong handcrafted multimodal contrastive baselines, whereas optimizing the scorer jointly with the encoders is less effective. Building on this observation, we introduce the Prior-Fitted Scorer (PFS): a Transformer scorer pre-trained on synthetic multimodal data generated from structural causal model (SCM) priors before transferring it to real-world data. The SCM exposes the scorer to diverse dependency structures, including shared, pairwise, and private information, providing a structured prior over higher-order cross-modal similarity. Across four real-world datasets, frozen Transformer scorers are competitive to contrastive baselines, and prior-fitting further improves the random initialization. More broadly, our results suggest that multimodal contrastive learning need not rely on manual similarity functions: Transformers can provide strong inductive biases, while causal pretraining can further shape these biases.
[AI-443] Last Secure Code Benchmark: Is the Code You Vibe Secure?
链接: https://arxiv.org/abs/2610.14766
作者: Tianyi Wu,Chao Ge,Peng Chen,Fengji Zhang,Yongxing Wu,Yuchong Xie,Jian Li,Chaoyuan Peng,Kaikai Zhang,Dong Huang,Mingzhe Du,Yukai Huang,Jiaheng Zhang,Terry Yue Zhuo,See-Kiong Ng
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 10 pages
Abstract:AI coding agents now write much of the code that ships, yet whether that code is secure remains unclear. We introduce the Last Secure Code Benchmark, a comprehensive evaluation of secure code generation by coding agents. Built from 150 real vulnerabilities, its 450 tasks span seven widely used languages (Java, C, C++, JavaScript, Python, PHP, and Go), 24 CWEs including 21 of the 2025 CWE Top 25 and 8 of the Top 10 KEV weaknesses, more than any benchmark we compare with, and three vibe-coding scenarios: filling in a function, writing a file, and building a repository’s source tree from scratch. Every task asks only for functionality, through a specification that never mentions security, and hidden functional and security tests grade the result. Across six frontier agents, 73.6% to 91.3% of tasks pass functionally, yet even the strongest, Claude Fable 5, produces code that is both functional and secure for only 25.1% of tasks, and the others for 16.4% to 24.7%. The best model only passes 12% of the vulnerability scenarios across all vibe coding setting, posing strong concern on the security of the code written by vibe coding.
[AI-444] GateSight: Recurrent Memory Anticipates Attention Memory for Query-Agnostic KV Eviction in Hybrid LLM s
链接: https://arxiv.org/abs/2610.14754
作者: Junjie Li,Jiong Lou,Jie Li
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Hybrid language models interleave Gated DeltaNet (GDN) layers, whose recurrent memory has a fixed size, with full-attention layers, whose key-value (KV) cache grows with the context and dominates serving memory. In this paper, we ask whether the recurrent memory already knows what the attention memory will need. We first show that which GDN heads of Qwen3.8-27B keep long memories is fixed by the weights rather than by the document. Next, we define the future utility of a KV entry by how much removing it changes the attention output over the model’s own continuations, and find that the write gate, decay and write magnitude that GDN layers emit during prefill anticipate this utility before any query is seen. Because the relation is spread across many GDN heads and layers, and a linear map recovers its trend but less of its high-utility tail than a nonlinear one, we propose GateSight, which learns it with lightweight per-head multilayer perceptrons that read these signals and no attention score. Their predicted utility decides which entries each attention layer keeps, and how much each layer keeps is set by the model’s native attention output gate, whose mean differs by more than an order of magnitude across layers. Finally, at serving time, GateSight scores the cache once after prefill and physically releases discarded KV entries. Experimental results on SCBench and RULER-4096 with Qwen3.8-27B and, with the same recipe, the sparse-attention model Qwen3.8-Flash-Next, up to the 262K-token limit of the models, show that with a tenth of the attention KV cache, GateSight is the most accurate of five query-agnostic methods on three of the four model-benchmark pairs and leads KVzip on all four, by 9.5 to 47.2 points. Notably, it scores a full-length context with negligible overhead relative to prefill.
[AI-445] Equal-information evaluation of credit assignment in modular financial reinforcement learning
链接: https://arxiv.org/abs/2610.14750
作者: Gennadiy Chistiakov,Yury Yanovich
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Credit assignment in modular financial reinforcement learning is difficult to evaluate because an attributed signal may improve a component loss without adding decision-relevant information or improving economic outcomes. We propose an evaluation framework combining equal-information reconstruction controls, fixed-utility comparisons and prespecified downstream guardrails. The framework is instantiated in an attribution-guided modular trading architecture and tested in three controlled studies covering update allocation, supervision-target representation and portfolio risk control. We find that residual-based allocation is exactly reproduced by an ordinary loss-budget control under the disclosed interface; direct shortfall supervision improves component prediction but fails decoded-outcome preservation; and component-wise critic supervision worsens risk-condition certainty equivalent and drawdown relative to scalar supervision under fixed utility. These results identify three distinct evaluation failures: redundant attribution signals, representation-induced prediction trade-offs, and misalignment between component supervision and downstream risk outcomes. A preserved historical engineering case further shows why module-level learning claims require run-level provenance. The contribution is a reusable methodology for evaluating credit assignment in expert systems operating under partial observability, delayed feedback and economic risk constraints.
[AI-446] SAGE: Pushing 2-Bit Vector-Quantized LLM s toward Real-World Usability
链接: https://arxiv.org/abs/2610.14745
作者: Xianglong Yan,Zenan Wu,Kewei Zhang,Kaicheng Yang,Yulun Zhang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Extreme low-bit weight-only quantization is an effective approach to reducing the deployment costs of large language models. Below 3 bits, scalar quantization often suffers severe accuracy loss, while vector quantization (VQ) better preserves accuracy through more expressive joint encoding. However, existing VQ methods still exhibit substantial degradation on real-world tasks, and two design challenges remain underexplored: (1) protecting sensitive weight channels within jointly encoded vectors; and (2) allocating a fixed bit budget among representation components, such as direction and radius, to suit different weight matrices. To address these challenges, we propose SAGE: Split Adaptation and Grouping by Energy for weight-only vector quantization. First, Energy-Stratified Grouping (ESG) couples channel grouping with activation-weighted fitting, grouping channels from different energy tiers within each subvector to prioritize sensitive channels. Second, Adaptive Bit Split (ABS) balances direction-codebook capacity and radius precision per matrix, selecting the split with the lowest activation-weighted reconstruction error under a fixed budget. Across benchmarks covering knowledge, mathematics, code generation, instruction following, and tool use, SAGE achieves state-of-the-art performance in the 2-bit quantization setting. On Qwen3-30B-A3B, it retains 95.3% of the BF16 model’s average benchmark score. With our custom-designed CUDA kernel, SAGE delivers up to a 2.55-fold non-streaming end-to-end speedup over FP16. The code is available at this https URL.
[AI-447] PTGPO: Prefix Tree Group Policy Optimization
链接: https://arxiv.org/abs/2610.14737
作者: Ruolin Chen,Yang Li,Jihang Wang,Zhuoya Zhao,Yuchen Liu,Jiaming Xu,Qian Zhang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Training language agents on long-horizon interactive tasks is challenging because sparse terminal rewards provide weak credit signals for intermediate actions. Existing group-based methods either assign a shared trajectory-level advantage or estimate step-level advantages from state groups, which may conflate the current action with subsequent decisions. We propose Prefix Tree Group Policy Optimization (PTGPO), a critic-free method that constructs prefix trees over trajectory suffixes within anchor-state groups. PTGPO assigns each tree edge a credit given by the difference between its child and parent empirical mean returns, capturing the conditional return change revealed by each additional action. These prefix-tree credits provide fine-grained learning signals without requiring a critic or additional rollouts. Experiments on interactive benchmarks ALFWorld, WebShop and Sokoban show that PTGPO consistently improves over RL baselines, demonstrating its effectiveness.
[AI-448] hink in Strategies Act with Tools: Traceable Agent ic Optimization for Combinatorial Problems
链接: https://arxiv.org/abs/2610.14736
作者: Zhongnian Li,Shi Xu,Peng Ying,Weikang Gong,Zhixiao Wang,Xinzheng Xu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Optimization by PROmpting (OPRO) directly leverages Large Language Models (LLMs) to generate candidate solutions through natural-language prompting. However, OPRO’s optimization process is difficult to track, which prevents LLMs from capturing useful information accumulated over previous iterations, thereby limiting solution quality. To address this limitation, we propose a Tool-Augmented Strategic Agent (TASA) which summarizes and exploits trajectory to improve solution quality. Specifically, TASA plans the strategy for each step based on the optimization trajectory, which decomposes the optimization process into traceable strategy selections. Moreover, we leverage LLM to construct problem-specific tools and introduce a strategy-guided adaptive tool planner, which efficiently utilizes the information of trajectory to update solutions. We establish local convergence under persistent exploration and evaluate the complete framework under a shared candidate-evaluation budget. The results support its solution quality and the traceability of strategy-guided search.
[AI-449] MHWA: Multi-timescale Hierarchical World-Action Model
链接: https://arxiv.org/abs/2610.14727
作者: Pengcheng Pan,Guoqing Ma,Yuhan Zhang,Yang Chen,Yichen Liu,Ziheng Li,Shan Yu
类目: Artificial Intelligence (cs.AI)
备注: 21 pages, 4 figures, 15 tables, including references and appendices
Abstract:Autoregressive world models have emerged as a powerful paradigm for sample efficient decision making, yet their open loop rollouts often exhibit increasing drift that leads to error accumulation even within a relatively small planning horizon. Hierarchical temporal abstraction is a natural candidate to mitigate such drift by reducing the effective prediction horizon; however, existing hierarchical designs can introduce added structural and conceptual complexity, sometimes making them difficult to control and limiting gains over flat baselines. To this end, we introduce MHWA (Multi-timescale Hierarchical World-Action model), which employs a strided high level context channel to reduce global context update frequency, effectively curbing the propagation of compounding drift. Crucially, via context aware routing, the gating network dynamically reconfigures a low level mixture of experts by using both global context and current observations, thereby enabling flexible high-low coordination, helping alleviate interference commonly observed in static hierarchies. Empirically, using RLiable’s seed-by-game aggregation, MHWA (77M parameters) achieves an IQM human normalized score (HNS) of 0.884 across three training seeds on a 14-game Atari subset using 10% subsampled offline data, outperforming a matched JOWA-70M baseline (0.529) and achieving numerically comparable performance to JOWA-150M while using nearly half the parameters and 70.88% fewer training FLOPs per sequence.
[AI-450] Integrating AIRL-Based Reward Estimation into a Control Method Using Fuzzy Inference and Policy Gradient
链接: https://arxiv.org/abs/2610.14723
作者: Seiji Ishihara
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:A method integrating policy gradient with fuzzy inference has been proposed. This method, hereafter referred to as the base model, has high flexibility with respect to the configuration of fuzzy inference rules and fuzzy representation. It can also automatically adjust the rule weights and the membership functions in the antecedent and consequent parts using the same framework for learning. In methods that integrate reinforcement learning with fuzzy inference, however, the reward function must be designed in advance. It is therefore useful to add learning-based reward estimation to the base model. This paper proposes a fuzzy control method that integrates a reward estimation mechanism based on the adversarial inverse reinforcement learning (AIRL) framework into the base model. The generative adversarial nets (GAN) framework adopted in AIRL, however, tends to suffer from unstable learning. To address this, the proposed method treats the policy of the base model as the generator and also uses the set of membership functions in the policy as the cost function of the discriminator. This configuration allows the discriminator update for reward estimation and the generator update for policy estimation to be performed simultaneously on the parameters shared by both, without estimating a value function. In simulations of an illustrative vehicle speed control task, the proposed method acquired policies with high generalization performance through stable and efficient learning.
[AI-451] Relational Semantic Grounding for Test-Time Discovery
链接: https://arxiv.org/abs/2610.14721
作者: Tianle Liu,Gongbo Cheng,Chenggong Ni,Minming Gu,Tao Zhou,Fuyuan Hu,Fan Lyu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Test-Time Discovery (TTD) aims to recognize known classes and detect novel ones from a continuous unlabeled test stream after deployment. Existing TTD methods mainly rely on visual clustering or prototype comparison, and can only assign anonymous labels to discovered classes, leaving them difficult to interpret in open-world applications. In this paper, we propose SEGA, a Semantic Grounding with Attributes framework for TTD. SEGA leverages large language models to generate attribute-rich descriptions for known classes and constructs an attribute-anchored visual-semantic space during training. At test time, we introduce Relational Semantic Grounding (RSG), which grounds unknown samples with similar and dissimilar known-class anchors, prompts the LLM to generate candidate concepts, and reranks them in the learned semantic space. The selected concept is further used for confidence-guided test-time semantic stabilization. By coupling attribute-level semantic alignment with relational concept grounding, SEGA transforms anonymous test-time clusters into interpretable semantic concepts. Experiments show that SEGA improves TTD baselines with the promise of semantic grounding for open-world.
[AI-452] DSABench: Benchmarking Source Discovery Coordination and Execution for Multi-Database Agents
链接: https://arxiv.org/abs/2610.14700
作者: Jinlong Luo,Haoran Wu,Bo Xu
类目: Artificial Intelligence (cs.AI)
备注: 39 pages
Abstract:Most text-to-SQL benchmarks provide the target database. In a large data catalog, however, choosing the sources is part of the problem: an agent must find every database needed for a request, determine how their results fit together, and execute the resulting workflow. We present DSABench to evaluate these capabilities jointly. Its 815 human-reviewed tasks draw on Spider, BIRD, and LiveSQLBench; 618 require several databases to contribute, and 105 allow more than one interpretation of the required sources. Executable contracts define source participation, cross-source dependencies, and source-local execution, while RouteEnv lets agents inspect public catalog evidence without revealing the gold annotations. In our routing-only study across seven model backends, the highest RA-Set score is 60.55%, compared with a highest RA-Graph score of 34.66% when coordination requirements are included. The gap becomes more pronounced when execution is required. Across four backends, median non-privileged Route+Exec-Graph success ranges from 6.69% to 7.98% without clarification and from 11.44% to 13.25% with one clarification; the best observed clarified configuration reaches 16.99%. On ambiguous tasks, clarification improves exact branch-source recovery by 39.8-43.2 percentage points. These findings illustrate why retrieving plausible databases or producing correct local SQL is not enough: an agent must carry the right information through the entire multi-database workflow.
[AI-453] Joint Discrete and SE(3) Diffusion for Protein Interface Design
链接: https://arxiv.org/abs/2610.14699
作者: Zhenqiao Song,Jianwen Xie,Lei Li
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:The design of high-affinity, target-specific protein binders is central to the development of modern therapeutics. However, existing protein-generation methods remain limited in this setting, as many generated designs fail to simultaneously achieve geometric and physicochemical complementarity at binding interfaces. To address this challenge, we introduce PIDiff, a physicochemically enhanced multimodal diffusion framework for protein-protein complex design. PIDiff jointly designs target-conditioned protein sequences and structures by coupling diffusion in discrete sequence space with diffusion on the continuous (SE(3)) structural manifold. At its core, PIDiff employs PIFormer, a novel architecture that interleaves geometry-aware invariant point attention with global sequence self-attention. We further curate PIBench, a protein-protein complex dataset with diverse target scaffolds derived from the Protein Data Bank (PDB). We first pretrain PIDiff on PIBench and subsequently fine-tune it for two therapeutically relevant applications: minibinder design and antibody design. PPFrame consistently outperforms baseline methods across sequence design success rate, structure binding energy, and sequence-structure co-designability on all tasks.
[AI-454] AlignAudio: A Unified Dual-Alignment Framework for Noise-Robust Speech-Conditioned Generation
链接: https://arxiv.org/abs/2610.14697
作者: Yixuan Li,Xuenan Xu,Zeyu Xie,Mengyue Wu
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:
Abstract:Speech-conditioned generation maps speech signals to target audio outputs. Prior end-to-end speech-conditioned generative models achieve high fidelity in clean acoustic conditions, but their performance degrades substantially under environmental noise. While prior noise-robust approaches have achieved significant success in feature-level purification, maintaining consistent generation under clean and noisy conditioning remains a key challenge. To address this, we propose AlignAudio, a noise-robust generative framework based on dual alignment: (i) aligning the representations of clean and noisy speech to preserve semantic cues, and (ii) enforcing generation-level consistency within a flow matching paradigm to improve conditioning fidelity. Experiments on speech-to-audio (STA) and voice conversion (VC) tasks demonstrate that AlignAudio maintains performance comparable to baselines on clean speech while significantly outperforming them across various noisy conditions. Furthermore, the proposed dual-alignment strategy is applicable to both single- and multi-stream speech conditioning and can be integrated into different flow matching-based generation frameworks and post-training settings. Audio demonstrations and qualitative comparisons: \hrefthis https URL\textcolorcyan\textitthis https URL.
[AI-455] AcrossVAM2.0: Conflict-Aware Point Supervision for Parameter-Efficient Video Diffusion Adaptation ICLR
链接: https://arxiv.org/abs/2610.14693
作者: Nan Wu,Yee Hung Hong
类目: Artificial Intelligence (cs.AI)
备注: ICLR conference
Abstract:Explicit point correspondence is attractive for adapting video generators: it describes how visible surfaces move, but can overwrite the appearance prior learned by a large pretrained model. We study this tension in a 5.01B-parameter text-to-video diffusion transformer. ACROSSVAM2.0 combines a confidence-weighted track-and-visibility head, zero-initialized residual feedback, and asymmetric PCGrad that removes only the auxiliary component opposing the primary flow-matching gradient on shared LoRA parameters. The design preserves the generator exactly at initialization and protects its primary update during adaptation. In a four-way ablation trained for 1,000 steps with three independent seeds, the full method lowers KLT forward-backward cycle error in all three seeds relative to a matched flow-only continuation (mean seed-level difference -0.00711, SD 0.00505), while robust sharpness remains positive in all three. An expanded 24-prompt evaluation (216 videos) yields 9.7% lower aggregate CoTracker3 cycle error than the matched control and near-matched performance on six VBench dimensions; however, the seed-level tracking effect is heterogeneous and I3D-FVD regresses from 0.1824 to 0.2022. These results support a bounded claim: conflict-aware adaptation improves correspondence consistency without broad perceptual degradation, but does not uniformly improve video quality. We release equations, pseudocode, matched outputs, source tables, and artifact provenance.
[AI-456] DexCast: Learning Real-World Dexterous Control from Human Videos
链接: https://arxiv.org/abs/2610.14691
作者: Zihan Gao,Yuhong Zhang,Mengzhe Wang,Kaisheng Liu,Shengpeng Li,Siyao Xiao,Runqing Cheng,Xiao Lin,Zhifang Liu,Haoqian Wang
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Recent advances in 3D hand-object interaction (HOI) perception have opened new opportunities for learning dexterous robotic skills from human videos. However, existing dexterous imitation learning methods often rely on costly and manually teleoperated robot demonstrations, requiring substantial human effort. Therefore, we build DEXCAST, a framework that learns dexterous robotic hand control directly from monocular RGB-D videos by leveraging joint HOI reconstructions, without requiring any teleoperation. Furthermore, how to model high-dimensional robotic hand motion and extract temporal context to inform future action generation remains a key challenge. To this end, we propose a DexHand Motion Perception Module (DMPM) to capture inter-joint coordination and temporal dependencies in compact latent representations that condition our diffusion policy. Extensive real-world experiments demonstrate that DEXCAST achieves exceptional performance across diverse dexterous manipulation tasks.
[AI-457] RASSG: Repair-Aware Semantic Survival Gating for Streaming Tool Execution
链接: https://arxiv.org/abs/2610.14659
作者: Daijin Hu,Xi Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Streaming speech assistants can launch tools before users finish speaking, overlapping execution with incoming speech to reduce user waiting time. However, subsequent self-corrections can invalidate apparently stable arguments, causing stale execution whose irreversible effects persist even after the final call is corrected. We propose Repair-Aware Semantic Survival Gating (RASSG) to jointly control tool launch and cancellation during streaming speech. RASSG launches a tool when three consecutive speech recognition results arrive with identical nonempty arguments, then combines calibrated argument invalidity risk with execution costs to decide cancellation at the last feasible opportunity before effects become irreversible. Evaluation on ReCAST, FDB-v3, and Audio2Tool uses a controlled execution clock with fixed tool runtimes and cancellation parameters. On ReCAST’s prospective-repair subset, RASSG reduces the stale execution rate from LocalAgreement-2’s 50.00% to 3.15%, with a mean delay of 3.676s across successful trajectories. The stale reduction also holds against every early-execution baseline on both public speech datasets. Ablations with identical initial launches support active cancellation, showing the value of controlling ongoing calls before their effects become irreversible. Code and data are available at this https URL.
[AI-458] Shaping Latent Mixture-of-Experts: From Expert Geometry to Layerwise Capacity Allocation
链接: https://arxiv.org/abs/2610.14644
作者: Yaoxiang Wang,Kai Zhang,Ruobing Xie,Lanrui Wang,Guoliang Zhao,Junhui Li,Jinsong Su
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Fine-grained mixture-of-experts (MoE) models commonly increase expert count by narrowing each expert, treating expert shape as a by-product of routing granularity. LatentMoE moves expert computation into a lower-dimensional space, exposing the input dimension and intermediate width of each expert as explicit design choices. We show that an expert’s local Jacobian rank is jointly bounded by these two dimensions, and that balancing them maximizes the rank ceiling at fixed per-expert parameters. Controlled shape sweeps across model scales and routing granularities consistently favor near-square experts. We further use layerwise intrinsic dimension (ID) to guide capacity allocation across depth, co-varying the number and shape of experts while keeping per-layer parameter count, activated FLOPs, and communication volume constant. Compared with layer-homogeneous baselines, our ID-guided allocation achieves lower validation loss and stronger downstream performance. These results establish expert shape and layerwise allocation as practical design axes for MoEs.
[AI-459] AegisVox: Factorized Privacy and Clinical Preservation for Long-Form Speech
链接: https://arxiv.org/abs/2610.14643
作者: Praphul Singh,Akshat Agarwal,Ratul Ghosh,Aashna Kanuga,Devashish Khatwani,Anurag Dwarakanath,Sri Gadde,Krishnaram Kenthapadi
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Clinical speech reveals identity through both spoken identifiers and voice, yet its pauses, delivery, and speaker roles can also carry useful clinical evidence. Protecting that speech therefore requires more than preserving its words. We present AegisVox, a factorized framework that assigns consistent surrogate voices and edits identifier-bearing turns using document-consistent replacements. Across 390 public simulated or non-patient conversations totaling 63.6 hours, a matched comparison of two speech renderers separates replacement realization from retained identifiers, speaker linkage, and clinical preservation. Text-to-speech rendering realizes 535/601 planned replacements versus 453/601 for the codec-based speech editor. However, similar aggregate clinical preservation conceals more model-labeled contradictions, and conversation-wide speaker scores conceal greater edited-speech linkage under a score-direction sensitivity analysis. Fresh transcription of original audio provides a reference for interpreting the clinical measurement chain; an independent judge tests sensitivity to model-based assessment. Better replacement realization can therefore coexist with worse clinical or speaker-linkage measurements. AegisVox supports use-specific evaluation of those tradeoffs, not a guarantee of identifier removal or clinical safety.
[AI-460] PEARL: Process-Experience Aware Reinforcement Learning for Multi-Turn Agents
链接: https://arxiv.org/abs/2610.14636
作者: Jie Yang,Mianqiu Huang,Yiwei Zhao,Bo Wang,Yang Liu,Yuxin Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Terminal rewards make multi-turn agents trainable without step-level annotations, but group-relative advantages provide little guidance about which intermediate decisions to adjust. Failed trajectories also contain reasonable actions, while unfavorable decision patterns can recur as training proceeds. We introduce PEARL (Process-Experience Aware Reinforcement Learning), which improves local credit assignment without an additional critic. Given two continuations from the same history with different outcomes, PEARL locates their first divergence and builds a feature representation of each branch’s local decision. It clusters these decision features in recent failure and success memories to identify recurring local patterns. Using this pattern memory, PEARL selectively strengthens negative advantages at lower-reward decisions, while leaving positive advantages and the base policy optimizer unchanged. On AppWorld, PEARL improves over ARPO on Normal and Challenge tasks at both 4B and 8B scales. On ALFWorld, it achieves 97.9% mean success at 4B, 2.7 points above the strongest baseline, while remaining competitive at 8B. Ablations of shaping location, success-bank filtering, and memory window support the value of selective, recurring-pattern-based updates.
[AI-461] From Existing Methods to New Research Ideas: Two Perspectives on Recursive Self-Improvement
链接: https://arxiv.org/abs/2610.14625
作者: Leijun Zhou,Yanke Yu,Zhihao Liu,Qu Xiang,Tiancheng Hu,Yiru Dong,Yihan Wu,Yangxi Xiang,Xuejun Wu,Xi Chen,Buyue Qian,Yaowei Zheng,Junhao Hu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Recursive self-improvement (RSI) offers a route to AI systems that participate in their own continued development. Recent methods allow language agents to retain changes to their parameters, memory, skills, and code for subsequent execution and improvement. We examine how these methods can inform new research ideas through two complementary perspectives. First, we introduce a unified taxonomy with nine design dimensions organized into four groups: update targets, update organization, update schedule and scope, feedback and selection. This taxonomy places existing methods in a shared design space, where examining individual dimensions and their interactions can inspire new research ideas. Second, we examine how existing RSI methods draw on traditional deep learning, for example through optimization, regularization, data augmentation, continual learning, and adversarial learning. We develop further ideas by examining more specific learning mechanisms and drawing on evolutionary computation. We also review how existing benchmarks assess self-improvement under online, offline, and offline-to-online settings. Together, these perspectives help researchers develop new RSI research ideas from existing designs and learning mechanisms. RSI papers and resources are available at this https URL.
[AI-462] Limitations on Autoregressive Token Generation via Sequential Sampling ICLR2027
链接: https://arxiv.org/abs/2610.14615
作者: Gal Arnon,Andrej Bogdanov,Paul Lou,Giulio Malavolta,Alon Rosen,Lorenzo Taschin
类目: Information Theory (cs.IT); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注: In submission to ICLR 2027
Abstract:We study autoregressive token generation through the notion of sequential samplers: A circuit that samples the next symbol only given the previous ones and possibly a small amount of state. We show that (1) a sequential sampler with k bits of state can be simulated by a stateless one at a roughly O(2^k) size overhead; (2) efficient insertional sampling (Stern, Chan, Kiros, Uszkoreit, ICML 2019) is more powerful than efficient sequential sampling in any fixed order; (3) sequential samplers with small state cannot establish certain forms of covert cryptographic channels. Result (1), together with the pseudoentropy vs. sequential sampleability duality of Vadhan and Zheng (STOC 2012), characterizes efficient sequential sampleability with small state. Result (3) provides a potential safety mechanism against subliminal coordination among malicious LLM agents.
[AI-463] CB-Agent : Concept-Based In-Context Learning for Sequential Action Selection
链接: https://arxiv.org/abs/2610.14574
作者: Qingyue Zhang,Huaze Tang,Minyu Zhang,Yuxiang Yan,Wenbo Ding,Shao-Lun Huang
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:In-context learning (ICL) allows large language models to use demonstrations without updating their parameters. For interactive agents, demonstrations are trajectories of reasoning, actions, and observations that illustrate how to complete a task. Reusing demonstrations in new situations is challenging: directly inputting full trajectories consumes limited prompt space and constrains the overall effectiveness, while natural-language summaries require the agent to translate general guidance into specific action choices. Therefore, we study whether expert action choices can directly supervise a reusable scoring rule. We propose CB-Agent, which extends concept-based ICL to sequential action selection. We extract environment-action decisions from expert trajectories and fit a concept vector by maximum likelihood to model the conditional distribution over candidate actions given the context. This model combines the concept with frozen language-model features and priors. During ReAct interaction, the agent uses this distribution to select environment actions without including demonstrations in the action-scoring context. Across TextWorld, ALFWorld, and ScienceWorld, CB-Agent achieves the highest cross-benchmark average among the evaluated methods at all three tested model sizes and ranks first on TextWorld and ScienceWorld at each size. ALFWorld experiments support the learned concept’s contribution beyond prior-only scoring and show further gains when combined with Reflexion. Code is available at this https URL.
[AI-464] Learning after Execution with Action-Specific Credit
链接: https://arxiv.org/abs/2610.14573
作者: Yufeng Tian,Jiahao Tang,Minghao Wu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Execution is often used as terminal feedback: it scores a generated program, but the resulting state is not used for another learning action. Learning after execution instead uses the execution result as an intermediate state that supports an additional learning interaction for the same policy. This creates a credit-assignment problem because the two actions share the same final objective but play different roles in the trajectory. Accordingly, action-specific credit assigns the initial action credit from the final trajectory outcome and the corrective action credit for improvement over the intermediate state. Image-to-TikZ serves as an executable visual testbed for this setting. On 945 held-out examples, learning after execution strengthens the shared policy under single-call inference and also produces a post-execution corrective capability. Single-call render success reaches 92.70%, compared with 88.25% for continued single-stage GRPO after the same number of additional training iterations. This corrective capability is not reproduced by continued single-stage training and depends on the intermediate state. Alternative credit definitions also induce different group-relative preferences over the same trajectories. Together, these results suggest that executable tasks contain useful learning structure beyond a terminal reward. Code is available at this https URL.
[AI-465] PsyMemBehav: Psychology-Informed and Memory-Grounded Two-Level Identity-to-Action Reasoning for Social-Media Behavior Simulation
链接: https://arxiv.org/abs/2610.14570
作者: Wenji Zhang,Xilong Cheng,Yunxiao Qin,Mai Tu,Tongye Zhu,Yuhao Zhou,JingBo Ji,Yuting Tan,Duoyin Li,Hongjiang Xiao,Yuan Zhang
类目: Artificial Intelligence (cs.AI); Social and Information Networks (cs.SI)
备注: 20 pages, 7 figures, including appendices
Abstract:Large language models (LLMs) support social simulation, but faithfully reproducing real users requires grounding both structured platform decisions and language in sparse, noisy behavioral evidence. We introduce PsyMemBehav, which maps psychology-informed profiles and retrieved memories to target-first platform behavior through two-level identity-to-action reasoning. To support this approach with comprehensive behavioral traces, we construct a dataset PsyMemBehav-bench from 5.57M observed behaviors of 5,862 real users (approximately 950 per user) across Reddit, Bluesky, and Twitter. Each recorded engagement is modeled as target selection followed by Comment, Like, or Repost and optional response generation. PsyMemBehav combines 41-dimensional user descriptors, top-5 BM25 memories, reverse-CoT supervision, and rubric-guided DPO over complete reasoning-action trajectories. Experiments show that PsyMemBehav 27B ranks first in 11 of 12 reported individual-level metrics, improving average response quality over GPT-5.4 from 54.28 to 59.68 (9.95% relative), and obtains the lowest group-level MAE (0.081) and MSE (0.009) among simulated systems. These results support evidence-grounded user modeling for observed-thread continuation; the benchmark remains conditional on recorded engagement and does not model no-action decisions.
[AI-466] LASER: Defeasible Certificate Search for Long-Range Protocol Dependencies
链接: https://arxiv.org/abs/2610.14552
作者: Zihan Tang,Huibiao Zhu
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: 11 pages, 6 figures, 10 tables
Abstract:Protocol specifications rarely justify executable behavior in one place: a transition may inherit its source state from a remote control block, name its events only by number, and remain subject to a later correction. Recovering the nearby clause is therefore insufficient; the system must assemble a source-linked certificate and determine whether any stronger rule applies. We introduce LASER, a proof-carrying search method for sparse, defeasible protocol evidence. LASER represents missing support and unresolved corrections as explicit obligations, searches typed evidence routes, and lets an occurrence-aware symbolic checker alone decide whether a certificate warrants, refutes, or leaves a behavior unresolved. On 91 BehaviorCheck families under a matched 2K budget, LASER improves Joint Edge+Proof over the strongest local reader on two model families ( \Delta=+.132 , p=.017 ; \Delta=+.198 , p.001 ) while cutting false discoveries by 2–3 \times . Controlled replays attribute the gain and its limits: neither reference routing nor deterministic citation expansion alone exceeds the local baseline; whole-document reading (29–65K tokens), two-round iterative retrieval, and even oracle minimal-certificate evidence all fail to reach LASER, localizing the bottleneck in typed obligation closure rather than evidence access. An authority-corruption replay shows source-link errors only cause abstention, while the only unsafe-warrant mechanism is mis-recording a Held erratum as Verified, confined to that erratum’s family. LASER thus improves long-range dependency recovery while converting incomplete evidence and uncertain authority into explicit residual obligations rather than unsupported behavior.
[AI-467] Active Neural Matter: A Finite-Window Response Framework for Physical Systems and Distributed AI Workflows
链接: https://arxiv.org/abs/2610.14549
作者: Tianchi Chen,Shihua Yu
类目: Machine Learning (cs.LG); Statistical Mechanics (cond-mat.stat-mech); Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC); Applied Physics (physics.app-ph)
备注:
Abstract:Physical systems and distributed AI workflows can retain past information, receive delayed feedback and combine conflicting signals. These features pose a common modeling problem: determining what information about past interactions a state must retain to account for future responses. Active Neural Matter (ANM) addresses this problem through a finite-window response framework. Forward analysis uses each system’s dynamics to predict state and output responses to perturbations, separating propagation from output sensitivity. Inverse analysis tests whether a candidate state retains enough information for these responses. With the system and future inputs or interaction rules fixed, distinguishable responses from histories assigned the same state identify distinctions the representation must preserve. Their distributional separation gives a lower bound on the worst-case error of approximating behavior from that state. Uncertainty bounds distinguish rejection, support within a fixed tolerance on tested comparisons, and unresolved cases. Delayed-consensus simulations check analytical predictions of response amplitude and phase over a narrow range of delays; flocking illustrates dependence on the chosen observable. An exact delayed-return construction establishes a sufficient state for a specified program. The AI studies identify distinctions in interaction history and evidence order that a representation must preserve to reproduce observed response distributions. Tests on new histories in a controlled feedback task provide bounded support for a state specified by the task instructions; the instructed rule predicts every observed action. ANM connects response prediction to measurable information requirements, separating behavioral fidelity from task correctness and initial-state adequacy from a closed evolution law for that state.
[AI-468] A Self-Defining Operator for Distributed Applications
链接: https://arxiv.org/abs/2610.14547
作者: Shihang Li,Thomas Anderson,Ratul Mahajan,Simon Peter
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Artificial Intelligence (cs.AI)
备注:
Abstract:To operate distributed applications with high availability, operators must rapidly diagnose and mitigate any incident. LLM-based agents can help, but unlike experienced human operators they do not improve as they handle more incidents. Each investigation starts from scratch, re-deriving likely causes and re-testing hypotheses even when a similar fault was resolved in the past. We present a self-defining operator (SDO) that improves with time by updating its operational behavior and memory. When SDO resolves an incident, it distills that experience into playbooks, which describe reusable diagnostic instructions for future incidents with similar faults, and detectors, scripts that take application telemetry as input and output relevant playbook(s). SDO continuously runs the detectors and invokes LLMs only when one fires, so its token cost scales with incident count rather than with operational time. Across 11 open-source microservice applications, SDO cuts resolution time by 52% on recurring faults while keeping per-incident cost low. SDO is open source at this https URL.
[AI-469] Are Multimodal Benchmarks Really Useful? Item-Level Multimodal Benchmark Diagnosis via Structure-Response Co-Calibration NEURIPS2026
链接: https://arxiv.org/abs/2610.14542
作者: Shiqi Zhang,Weixin Zeng,Ziheng Zhang,Jiuyang Tang,Xiang Zhao
类目: Artificial Intelligence (cs.AI)
备注: Accepted at NeurIPS 2026. 34 pages, including appendices
Abstract:Multimodal benchmarks are central to evaluating whether models can reason over information from multiple modalities. However, existing literature has revealed that some benchmarks may contain shortcut items that can be answered from a single modality or from answer-option regularities alone, thus failing to reflect models’ true cross-modal inference ability. Existing methods for assessing benchmarks typically rely either on intrinsic item structure or extrinsic model response behavior, which are not adequate for a comprehensive diagnosis. To fill in this gap, we propose SRCoD, a structure-response co-calibrated diagnosis framework for item-level diagnosis of modality dependence in multimodal benchmarks. SRCoD estimates sample-wise multimodal information structure from benchmark content and target answers, and uses this signal as a soft structural anchor for modality-conditioned item response modeling. The learned model produces a calibrated item-level diagnostic profile that reflects how strongly an item depends on different forms of modality evidence, enabling fine-grained benchmark analysis beyond aggregate accuracy. SRCoD also provides interpretable attribution of the diagnosis results. Extensive experiments under two complementary protocols, i.e., benchmark-refinement oriented validation and human-labeled item validation, demonstrate that SRCoD outperforms state-of-the-art benchmark diagnosis methods.
[AI-470] GG-Nav: Goal-oriented Grounding Chain-of-Thought Elicits Reasoning in Navigation Foundation Models
链接: https://arxiv.org/abs/2610.14541
作者: Shaohao Zhu,Wensi Huang,Siqi Zhang,Yuqiang Yang,Meng Wei,Chenyang Wan,Tianle Liu,Tai Wang,Jinming Xu
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 22 pages. Project page: this https URL
Abstract:Emerging strong Vision-Language Models (VLMs) have advanced navigation foundation models. However, these models still have unsatisfactory performance on goal-oriented navigation settings. In this paper, we first conduct a motivating study revealing that the key issue lies in the insufficient utilization of streaming observations and imprecise target localization. To tackle these challenges, we propose GG-Nav, a simple yet effective navigation framework that learns a structured chain-of-thought (CoT) mechanism via supervised fine-tuning to explicitly localize goal-relevant instances, accumulate grounding evidence, and support more reliable decision-making. Specifically, the structured CoT is designed with two key stages: (i) Observation-to-Evidence, which reasons over all intermediate frames between adjacent decision steps to retain useful goal-related evidence; and (ii) Evidence-to-Decision, which integrates the current observation with accumulated evidence to ground the target and support the generation of navigation actions or clarification questions. Extensive experiments show that GG-Nav outperforms the state-of-the-art methods in terms of success rates on the HM3D ObjectNav, VL-LN, and CoIN benchmarks by 3.1%, 16.6%, and 5.3%, respectively. Ablation studies further show that accumulating evidence from intermediate observations reduces evidence-missing failures by 11.6%, while explicit grounding reduces stopping errors by 49.4%. We also deploy GG-Nav on a Unitree Go2 robot, demonstrating its zero-shot generalization capability in unseen real-world environments.
[AI-471] Who to Trust in Multi-Robot Systems: Joint Robot and Object Hypothesis Trust Inference via Neural-Symbolic Cross-Validation
链接: https://arxiv.org/abs/2610.14530
作者: Ruiyang Wang,Hao-Lun Hsu,Haocheng Meng,Songqiao Xie,Miroslav Pajic
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Multi-robot systems (MRS) offer significant advantages over single-robot platforms by enabling information sharing and coordination for different tasks in complex environments. However, reliable cooperation critically depends on inferring trust over both information sources and their reported hypotheses, particularly under partial observability and potential adversarial interference. Existing approaches frequently depend on pre-established trust relationships or side-channel information, which may not be available or reliable in realistic MRS deployments. This work introduces NeST-Bayes, a neural-symbolic framework that jointly infers robot and hypothesis trust using only inter-robot observation consistency, without requiring preassigned robot trust, reputation information, or side-channel identity signals. NeST-Bayes infers instantaneous trust evidence scores for robots and object hypotheses by reasoning over ego-centric relational graphs constructed from cross-robot observation consistency. These evidence scores are accumulated over time through Bayesian updates of Beta-distributed trust beliefs to maintain temporally consistent trust estimates. Extensive evaluations show that NeST-Bayes achieves the highest mean object precision and robot recall across all benchmark settings, and the highest robot precision and object recall in most settings, compared to trust-based methods. Under aggressive optimized attacks, it achieves 98.8% robot precision with 93.8% recall and 98.3% object precision with 95.4% recall. In an occlusion-rich Webots warehouse, it maintains 92.5% robot precision with 92.0% recall and 97.3% object precision with 85.0% recall. These results demonstrate joint improvements in the reliability and coverage of retained information.
[AI-472] Energy Ring: Stabilizing Feature Energy for Reliable Test-Time Discovery
链接: https://arxiv.org/abs/2610.14525
作者: Tianle Liu,Fan Lyu,Tingliang Feng,Chenggong Ni,Tianqi Wang,Zhang Zhang,Fuyuan Hu,Rui Yao,Tao Zhou
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Test-Time Discovery (TTD) aims to recognize known classes and detect novel ones while the model is already deployed, operating directly on a continual unlabeled test-time stream. Existing TTD methods are overly conservative, making known and unknown features collapse together and difficult to separate during test-time adaptation. In this paper, we revisit TTD from an energy perspective and find that the natural energy gap between known and unknown classes is unstable and easily collapses. We therefore propose Energy Ring to stabilize this gap by anchoring the two types of classes on separate energy rings. A lightweight adapter enlarges the norm-based gap while preserving known-class directions, and a perturbation consistency module forms compact clusters for unknown classes. During testing, energy-guided confidence weighting improves prediction, and a stability-based rectification mechanism corrects unreliable updates. Experimental results show that Energy Ring significantly improves TTD performance against existing TTD methods.
[AI-473] MontageFlow: Physically Valid High Fidelity EEG Generation
链接: https://arxiv.org/abs/2610.14507
作者: Sudarshan Regmi,Nabaraj Subedi,Arvind Pillai,Yu Yvonne Wu,Yuliang Chen,Michael V. Heinz,Lisa Marsch,Nicholas C. Jacobson,Andrew Campbell
类目: Artificial Intelligence (cs.AI)
备注: Code: this https URL
Abstract:Generative modeling offers a promising solution to the scarcity of labeled clinical electroencephalography (EEG) data under strict privacy constraints. Although recent generative models report substantial improvements in generation fidelity, we show that these gains conceal a categorical failure. EEG is not an arbitrary multichannel signal: each channel records the voltage difference between two scalp electrodes. So, real recordings satisfy exact linear identities among their channels. Current generators invariably violate these identities, yet this has gone unnoticed. We therefore propose MontageFlow, a flow-matching model that enforces these identities by restricting its output to signals that satisfy them. We find that ensuring validity improves fidelity too. Furthermore, to retain the EEG characteristics that oversmoothing washes out, we also condition the model on a physiological state vector encoded from the whole EEG window. For generation, this vector is drawn from a prior fitted in latent space. Across 3 large-scale benchmarks, our model improves TS-FID by 1.6-3.6 \times over the closest rival while ensuring validity.
[AI-474] Rethinking Activation Functions: Reducing Bias Can Improve Learning
链接: https://arxiv.org/abs/2610.14500
作者: Zirui Peng,Heyi Zheng,Ziming Liu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 26 pages, 21 figures, 2 tables
Abstract:Many widely used activation functions, including ReLU, GELU, Swish, and Sigmoid, produce predominantly positive outputs, resulting in a nonzero post-activation mean that we term activation bias. Despite its ubiquity, the effect of activation bias on learning dynamics remains poorly understood. Using controlled models, we show that activation bias accelerates initial loss reduction but is accompanied by overshooting, non-smooth trajectories, and a steeper loss landscape. This early advantage also coincides with the model’s use of task-relevant information encoded in the activation bias component. In contrast, reducing activation bias slows initial learning but ultimately yields a lower final loss. Experiments on larger vision and language models exhibit a similar pattern: smaller activation bias is associated with better late-stage performance. Thus, this seemingly minor property can substantially redirect learning and warrants explicit consideration in future activation function design.
[AI-475] Reels-RSI: Program-Level Self-Improvement for Code-Rendered Explainer Videos
链接: https://arxiv.org/abs/2610.14495
作者: Wei Lin
类目: Artificial Intelligence (cs.AI); Graphics (cs.GR); Multimedia (cs.MM); Software Engineering (cs.SE)
备注: 19 pages, 4 figures, 9 tables. Code, rules, lessons, data and evolve reports: this https URL
Abstract:Explainer videos – a worked geometry problem, why the sky is blue, how a dynasty fell – are usually one more target for pixel video generation. We argue they are a different kind of artifact: a program (script, storyboard and per-frame animation code) that a deterministic renderer turns into frames. A program can be read, diffed, tested and audited; pixels can only be watched. Reels-RSI is an open-source reel director for any agent harness; its self-improvement happens at the program level rather than in weights or prompts. Every revision pass is logged with a code snapshot; a retrospective pairs each check finding with the diff that fixed it and proposes lessons; a lesson that a static check can catch is compiled into a rule with a must-fail and a must-pass example, so the memory is unit-tested; and a rule learned today runs over everything made before it. A bounded, benchmark-gated step lets an agent rewrite the director’s playbook, kept only when blind pairwise judgments on held-out topics prefer it; nothing enters memory without a human decision. Across 111 reels – a field study, two playbook evolutions, a 2x2 memory ablation, a weaker director, a 48-reel learning curve and a judge-reliability study – we find that the rule learned in the field caught four latent defects in frames authored before it existed and none after; that rules are an audit for a strong director (eight unguarded reels, no violation) and a guard for a weaker one (thirty violations unguarded versus five); that lessons shorten revision loops but do not compound or move the judge; and that the first rewrite’s 11–1 verdict was a position bias in our vote design (judges pick the second-shown reel 62–81% of the time), masking scripts that traded thirteen facts for craft. We document four failure modes of the preference signal, which check caught each, and the fixes. Code, rules, lessons and data are public.
[AI-476] HiPPT: Physiology-Grounded Wearable PPG Representation Learning via Multi-Scope Text Alignment
链接: https://arxiv.org/abs/2610.14489
作者: Jiaqi Wang,Wei Chen,Xiaoyu Li,Lexian Xia,ShuaiLong Tang,Chenran Zhang,Donglin Xie,Guangpu Zhu,Yelei Li,Yi Zhou
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Reusable wearable PPG representations should capture local pulse morphology and broader physiological regularities. Existing pretraining objectives, however, usually apply supervision at the recording or pair level, overlooking that different concepts become reliable over different scopes of waveform evidence. We propose HiPPT, a physiology-grounded framework that matches each supervision source to its appropriate evidence scope. We construct a PPG-Text corpus of over 4 million wrist-worn recordings from 202K users, pairing recordings with waveform-derived morphology and metadata-derived context descriptions. HiPPT combines record-level morphology alignment, which grounds individual recordings in pulse characteristics, with cross-recording context alignment, which aggregates recordings sharing the same metadata tuple to capture context-consistent regularities. Waveform-guided text modeling further recovers masked contextual concepts from PPG patch representations, providing fine-grained waveform-text supervision. Across six downstream tasks spanning blood pressure, vascular stiffness, physiological attributes, and cross-dataset heart-rate estimation, HiPPT achieves the best result on the primary linear-probing metrics. The aligned space also supports zero-shot profiling of age group, sex, and BMI category, individually and jointly, with performance improving as more recordings are aggregated. These results show that evidence-scope matching yields transferable and semantically accessible wearable PPG representations. Code and pretrained weights will be released.
[AI-477] AlignPoison: Conditional Targeted Poisoning of Generative Recommenders via Decoding-Aligned Prefixes
链接: https://arxiv.org/abs/2610.14487
作者: Zhuoxuan Zhang,Jinhao Duan,Xinyu Zhao,Jinduo Guo,Shuai Shao,Rui Li,Luke Simon,Zhun Deng,Tianlong Chen,Bo Sun
类目: Artificial Intelligence (cs.AI)
备注: 23 pages, 10 figures, including appendices
Abstract:Generative recommendation models based on autoregressive decoding have been adopted as alternatives to classical ranking-based systems. While their modeling flexibility has been widely studied, their resilience to data poisoning remains largely unexplored. In this work, we ask: how susceptible are generative recommenders to conditional targeted poisoning? Under a realistic setting where an adversary can inject limited malicious interactions and query a recommendation API as a black box, our empirical study shows that generative recommenders are vulnerable: injecting fewer than 1% synthetic interactions significantly increases the likelihood of promoting malicious targets under the trigger. To characterize this vulnerability, we analyze the synthesis of malicious interactions from two perspectives: (i) decoding alignment and (ii) prefix popularity. We observe that short, context-aligned prefixes that are both common under the model’s induced transition dynamics and strongly linked to a frequent trigger consistently amplify targeted promotion. Motivated by this analysis, we develop AlignPoison, a black-box framework that discovers and samples context-aligned prefixes from API outputs. We conduct experiments on benchmarks in unimodal and multimodal generative recommendation settings. Results show that these prefixes outperform competitive baselines, improving ASR lift by up to 16.3 percentage points and highlighting a decoding-dependent, context-sensitive poisoning vulnerability in generative recommendation models.
[AI-478] Diffusion-Based Auctions for Multi-Winner Banner Ads
链接: https://arxiv.org/abs/2610.14473
作者: Lilian Sun,Henry Huang,Fucheng Warren Zhu,Giannis Daras,Constantinos Daskalakis
类目: Computer Science and Game Theory (cs.GT); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Appeared in EC 2026
Abstract:We propose a new framework for banner ad auctions, leveraging diffusion models to allow for multiple bidders to co-win a single banner ad opportunity, and in particular influence the ad shown on the banner depending on the strength of their bids, enabling a far richer outcome space compared to standard banner ad auctions, where there is a single winner who shows their ad. In our framework, the outcomes are images (what ad to show) and each bidder expresses their preferences over outcomes by declaring their value per unit of alignment between the image chosen by the mechanism and a textual description of their desired image. Our proposed mechanism combines the advertisers’ bids and textual descriptions into a score that drives the reverse process of a diffusion model. This is used to sample a set of candidate images from which the welfare maximizing one is eventually chosen. Extensive experimentation evaluates our proposed mechanism around the following axes: (1) efficiency improvements of our allocation rule relative to the single-winner allocation rule; (2) monotonicity of our allocation rule; and (3) approximate incentive compatibility when combining our allocation rule with Myerson lemma-based or Clarke rule-based payments. Improvements along all these axes are shown with the number of Monte Carlo samples. Meanwhile, these benefits are attained while preserving the image quality.
[AI-479] KV4-RL: Unlocking Faster Asynchronous Rollouts for RLVR with INT4 KV Cache Quantization
链接: https://arxiv.org/abs/2610.14448
作者: Sihwan Park,Sung-Yub Kim,Doohyuk Jang,Eunho Yang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 32 pages, 15 figures, 6 tables
Abstract:Asynchronous rollout improves the efficiency of reinforcement learning with verifiable rewards (RLVR) by reducing under-utilization during long-tail generation. Sustained concurrency also concentrates context processing into fewer decode steps, increasing the KV workload per step. We characterize this \emphKV concentration and find that KV cache quantization delivers substantially greater acceleration than weight-only and weight-activation quantization in our matched-workload comparisons. However, INT4 KV rollout can degrade learning even with token-level truncated importance sampling (TIS). We propose \emphKV4-RL, which combines calibration-free INT4 KV quantization with selective INT8 protection of attention sinks and recent context. Protecting a bounded number of positions reduces sampler mismatch while retaining most of the speedup from INT4. Across 400 iterations of Qwen3-8B-Base GRPO training, KV4-RL achieves 1.80\times cumulative rollout speedup and 1.27\times cumulative full-iteration speedup over BF16 rollout, reaching 1.97\times rollout speedup over the final 50 iterations. Under a common BF16 evaluator, its final six-benchmark mean accuracy is within 0.5 percentage points of BF16-rollout training for both Qwen3-4B-Base and Qwen3-8B-Base.
[AI-480] Specification-Based Testing of Large Language Models
链接: https://arxiv.org/abs/2610.14443
作者: Kausar Moshood,Jace Bolante,Roshan Aditya Murali,Manish Motwani
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: 21 pages, 1 figure
Abstract:Model cards, system cards, and safety reports increasingly document LLM safety behaviors, yet safety evaluations typically start from predefined benchmarks rather than asking which documented claims those benchmarks actually test. We introduce LEST (LLM Evaluation through Specification-driven Testing), a framework that treats documented safety claims as specifications for black-box testing. In LEST, each claim determines which benchmark prompt tests it and whether a Refuse-to-Comply transition, where a model refuses a prompt but complies with its mutated version, counts as evidence against it. LEST mutates these prompts to search for counterexamples, manually validates each candidate, and traces validated counterexamples back to the claims they contradict. We evaluate LEST using the system cards of Claude Opus 4.8 and GPT-5.5 with 97 seed prompts from SORRY-Bench across Mental Health and Self-Harm, Harmful and Violent Content, and Fairness and Bias. Of 56 identified safety claims, 29 (51.8%) are prompt-testable, but only 17 (58.6%) are exercised by an applicable benchmark prompt. Both models refuse 95% of benchmark prompts. Evaluating 1,836 mutated prompts on both models reveals 312 candidate Refuse-to-Comply transitions flagged by automated response classifiers. However, human validation confirms only 19 (6.1%) as both intent-preserving and genuinely compliant, showing automated evaluation substantially overestimates behavioral failures. Tracing these 19 transitions back to the claims reveals counterexamples for 6 of 9 covered claims for which refusal is an applicable oracle. Our results show that benchmark prompts may not be sufficient to validate all documented LLM safety behavior, and automated failure detection requires manual validation. LEST provides end-to-end traceability from natural-language safety documentation to executable tests and human-validated counterexamples.
[AI-481] arget-ESS Tempered Policy Optimization (TETPO): Smooth Variance Control via Bridge Importance Sampling
链接: https://arxiv.org/abs/2610.14440
作者: Shayan Mohajer Hamidi,Kayhan Behdin,Wenhui Zhu,Aida Rahmattalabi,Jincheng Cao,Jelena Markovic-Voronov,Alborz Geramifard
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Asynchronous reinforcement-learning post-training of large language models is computationally efficient but inherently off-policy: the learner updates on rollouts generated by a stale policy, while correcting this mismatch requires importance weights that can be extremely heavy-tailed. A small number of trajectories may then dominate an update, collapsing the effective sample size (ESS) and increasing the variance and instability of the resulting gradient estimates. Existing stabilizers apply fixed interventions regardless of the realized severity of weight concentration, either discarding selected samples or reweighting them without adapting the correction strength. We introduce \emphTarget–ESS Tempered Policy Optimization (TETPO), which replaces each importance weight \rho by \rho^\alpha for a data-adaptive exponent \alpha\in[0,1] : \alpha=1 preserves the original importance weights, whereas \alpha=0 assigns uniform coefficients. For each prompt group, TETPO selects the largest \alpha whose tempered weights satisfy a target ESS fraction \kappa , applying little tempering to healthy groups and stronger tempering when weight degeneracy is severe. At the population level, the tempered coefficients are exact importance weights for a geometric bridge between the sampler and learner policies. The exponent is efficiently found by one-dimensional bisection, while the resulting weights satisfy deterministic bounds on normalized-weight dispersion and weighted-gradient norm that are independent of policy drift, together with a fixed- \alpha bound on bridge-target discrepancy. Power tempering also introduces no explicit zero-weight mask. On mathematical-reasoning benchmarks under high staleness, TETPO outperforms strong on- and off-policy baselines.
[AI-482] Frustratingly Easy Agent ic RAG via Portable Skills: A Case Study in Customer Support EMNLP2026
链接: https://arxiv.org/abs/2610.14428
作者: Istok Bako,Bonan Min,Tiffany Shao,Claire Cheng,Yashar Mehdad
类目: Artificial Intelligence (cs.AI)
备注: Accepted to EMNLP 2026 Industry Track. 15 pages, 1 figure, 5 tables
Abstract:Enterprise workflow automation typically requires retrieving a complete and accurate set of relevant information from large operational document corpora (SOPs, policy documents) given a specific scenario or request. In customer support (CS), for example, human specialists must determine the correct sequence of resolution steps for each case, a task that demands precise grounding in policy documentation. Meeting this bar in production typically requires complex document chunking, indexing, and retrieval pipelines, often layered with learned re-rankers, query rewriters, and bespoke agentic RAG infrastructure. We report a deployment case study of a simpler alternative: a portable skill consisting of a natural-language instruction, lightweight tools (a retrieval API, format-checking scripts), and a corpus enriched with an LLM-generated navigation layer, which is a structured “Wikipedia-style” index over the underlying documents. This agentic RAG approach is straightforward to implement (1-2 engineer days), portable across capable agentic harnesses (e.g., Claude Code), and performs strongly on the challenging task of CS resolution workflow generation, achieving 99.2% procedural equivalence to human-approved ground truth and 83.3% evidence sufficiency. It is also roughly an order of magnitude slower per case than our deployed production assistant (~90 s vs. ~8-9 s), so we claim no latency advantage. Our claim is scoped to that setting: a mature, well-curated corpus, a latency-tolerant workflow, and human review of every case. There, the skill did the work of a complex RAG stack, and maintenance became a matter of editing text and data rather than retraining models. Comments: Accepted to EMNLP 2026 Industry Track. 15 pages, 1 figure, 5 tables Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2610.14428 [cs.AI] (or arXiv:2610.14428v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2610.14428 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-483] Coarse-Graining Recursive Expert Computation: Continuation-Relative Interfaces and Failure Modes
链接: https://arxiv.org/abs/2610.14413
作者: Shihua Yu,Yanzhi Wang,Tianchi Chen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computational Complexity (cs.CC); Computational Engineering, Finance, and Science (cs.CE)
备注:
Abstract:Computational coarse-graining aims to replace a sequence of updates with a cheaper module while preserving task performance. We present a continuation-relative evaluation framework that assesses replacements through the computation that follows them, using downstream task quality and total execution cost as joint criteria. The framework distinguishes boundary representations, summaries of skipped computation, executed operations, and downstream loss. It makes interfaces and permitted deployment information explicit, combines fine-path references with directly trained controls using the same shortened architecture, and accounts for selection overhead. An exact loss decomposition separates replacement effects at a fixed downstream action from effects of changing that action. A finite construction shows how retained information can reduce the stipulated cost of zero-error prediction under fixed information and controller restrictions. Experiments in two frozen pretrained language models show that representation fidelity and task loss can favor different fitting objectives. In a prespecified comparison on a 1.7-billion-parameter model, next-token-loss training lowers in-distribution negative log-likelihood by 0.1901 nats per scored token compared with reconstructing the skipped blocks’ output, using the same optimizer and update count. This ranking reverses in all nine paired comparisons under distribution shift. In a recursive expert prototype, the jointly adapted candidate lowers task error relative to frozen-bank and router-only baselines, but trails directly trained controls, and the tested input-only selectors fail the joint quality-cost criteria. Together, the protocol and studies provide explicit criteria and diagnostics for distinguishing useful computational replacement from local approximation gains within specified interfaces and evaluation settings.
[AI-484] Memory Without Curation: Learning Multi-Stream Latent Memory for Robot Policies
链接: https://arxiv.org/abs/2610.14406
作者: Yilin Wu,Rutav Shah,Shuaiyi Huang,Sumedh A. Sontakke,Homanga Bharadhwaj,Andrea Bajcsy,Max Simchowitz,Guanya Shi
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:It is widely appreciated that robotic policies require long-term memory to perform complex tasks. However, directly conditioning policies on all past observations leaves policy learning susceptible to learning spurious shortcuts, and introduces large latency at policy deployment. As such, today’s memory-augmented policies curate what to remember, either by retaining only recent frames, selected keyframes, or textual summaries, or by learning latent representations of the visual stream alone. In this work, we propose an alternative approach IMPRINT, a policy-agnostic latent memory framework that augments vision-language-action (VLA) models with multi-stream memory without explicitly constraining what the policy should remember. IMPRINT compresses different streams into a latent memory that is updated incrementally at inference, enabling low latency. To avoid shortcut learning, IMPRINT adopts a two-stage training procedure, first supervising memory to reconstruct multiple streams (observations, actions, proprioception) of past information, as well as estimate current task progress, followed by end-to-end finetuning, jointly with the policy, on action prediction. Across both simulation and real-world tasks demanding memory, IMPRINT improves average success over the base VLA by at least 26%, outperforming popular alternatives, and maintaining lowest inference latency among memory-augmented VLAs. Through careful ablations, we demonstrate the utility of incorporating diverse information streams into robotic memory, validating the advantages of foregoing history curation. Project website: this https URL.
[AI-485] SAT-RL: Training Language Models for Logical Reasoning with Symbolic Solvers
链接: https://arxiv.org/abs/2610.14402
作者: Jialiang Sun,Kuldeep Meel
类目: Artificial Intelligence (cs.AI); Symbolic Computation (cs.SC)
备注:
Abstract:Large language models (LLMs) have achieved strong performance in mathematical and logical reasoning, yet they still suffer from the curse of complexity and struggle as problem difficulty increases with the number of constraint interactions and the size of the search space. In contrast, satisfiability solvers such as Z3 provide scalable and verifiable reasoning through constraint solving under formal theories, but lack the flexibility to interpret natural-language problem descriptions. In this work, we propose SAT-RL, a reinforcement-learning (RL) framework that trains LLMs to interact with symbolic solvers through multi-turn tool execution. SAT-RL leverages synthetic data generation with verifiable solutions, supervised warm-up on solver-augmented traces, and solver-grounded reinforcement learning with tool-aware rewards. Experiments show that SAT-RL substantially improves symbolic reasoning with tools on both in-domain and out-of-domain logic tasks. Our code is available on this https URL.
[AI-486] Measuring the Inference Energy Efficiency of Open-Weight LLM s across Benchmarks Reasoning Modes and Hardware
链接: https://arxiv.org/abs/2610.14389
作者: Zeyu Yang,Wes Armour
类目: Artificial Intelligence (cs.AI); Performance (cs.PF)
备注:
Abstract:Benchmark scores tell us what an LLM can do, but not how much energy is required to deliver that capability. We directly measure this trade-off for 86 open-weight model checkpoints, evaluated on seven benchmarks and six hardware platforms ranging from a consumer GPU to an eight-GPU server. We find large differences in energy efficiency even among models with similar capability. Most configurations lie away from the score–energy Pareto frontier: they use a median of 2.6–4.8 \times more energy than another configuration that scores at least as well, with differences reaching 27 \times . Reasoning can improve capability, but enabling thinking increases energy by about 3.7 \times on average, and the benefit depends strongly on the model family and output budget. Generated-token count is also an imperfect proxy for energy, misordering 18–24% of model pairs. In contrast, FP8 reduces energy by roughly a quarter with little change in score, while changing hardware largely rescales energy without changing the relative ranking of models. These results show that capability alone gives an incomplete picture of efficiency. We argue that benchmark scores should be reported together with measured energy consumption under a specified serving regime and output budget.
[AI-487] Deep learning from a control vantage point
链接: https://arxiv.org/abs/2610.14385
作者: Álvaro Rodríguez Abella,João Pedro Silvestre,Bahman Gharesifard,Paulo Tabuada
类目: ystems and Control (eess.SY); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Optimization and Control (math.OC)
备注:
Abstract:This article offers an overview of recent developments in deep learning from a control vantage point. It reviews several classes of models including feedforward and residual neural networks, transformers, large-language models, and selective state space models while explaining how control theory plays an important role in understanding their training, properties, or architecture design. The article explains how residual neural networks can be interpreted as the discretization of a continuous-time control system thereby enabling the use of control theoretic techniques for its analysis. To illustrate the power of this bridge between control and deep learning, two technical results are discussed: 1) the ability of a residual neural network to approximate arbitrary continuous functions is a non-standard controllability problem that can be solved using geometric nonlinear control and 2) transformers, the key architectural component of most large language-models, can be seen the discretization of a consensus algorithm and thus large language-models exhibit consensus. Another point of contact between control theory and deep learning is the connection between backpropagation, the algorithm used to train neural networks, and optimal control. The article addresses how the older connection between these problems, established in the 80’s using discrete-time models can be reconciled with the more recent approach using continuous-time models. Several open problems and directions for future research are discussed, including selective state-space models, a recent class of models inspired by the models used in control.
[AI-488] ACED-Bench: Evaluating How LLM Agents Acquire and Act on Causal Evidence
链接: https://arxiv.org/abs/2610.14381
作者: Yiwei Chen,Amila Weerasinghe,Siddharth Biswal,Utkrisht Rajkumar,Olcay Boz,Taylor Berg-Kirkpatrick,Rose Yu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models are increasingly used as scientific assistants, yet we still lack reliable ways to evaluate whether they can investigate a system rather than merely answer a prompt. A useful evaluation must let agents choose experiments, interpret finite evidence, and make causal decisions, while keeping the environment coherent across repeated observations and interventions and preserving objective ground truth. We introduce ACED-BENCH (Active Causal Evidence and Decision Benchmark), a neuro-symbolic benchmark built around this requirement. The key idea is to separate the readable scientific surface from the formal causal world: LLMs generate natural-language scenarios and task text, while hidden probabilistic graphical models define the causal graph, data-generating process, intervention semantics, validation checks, and oracle answers. This yields scientific worlds that are natural for agents to investigate but exact enough for scalable, programmatic evaluation. The main leaderboard ACED-CORE contains 180 tasks: 60 set-valued causal ancestry/descendancy questions and 120 constrained causal-decision tasks. We also report ACED-QUAL, a 150-question yes/no qualification check using the same hidden-PGM interface. We evaluate seven contemporary LLMs and two earlier-generation models across three active agent designs. ACED-QUAL checks whether agents can use the interface to draw conclusions from evidence, while ACED-CORE tests whether they can decide what evidence to collect and when to commit to an answer. Evidence-ledger analysis localizes failures within the investigation: agents test the wrong intervention, lose relevant evidence, or continue after sufficient evidence has appeared. These results suggest that reliable AI co-scientists need explicit support for experiment design, evidence retention, answer commitment, and stopping.
[AI-489] DexFLEX: Contact-Aware Foundation Controller for Command-Guided Dexterity
链接: https://arxiv.org/abs/2610.14372
作者: Weikang Wan,Zeyu Lai,Jianyu Hou,Yuchen Zhou,Jiawei Fu,Henrik I Christensen,Hao Su
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Dexterous robot hands can receive useful motion intent from teleoperation or learned policies, but successful execution also depends on contact decisions that these commands rarely specify. We introduce DexFLEX, a contact-aware foundation controller that turns upstream fingertip-motion drafts into contact-consistent joint commands. Instead of treating a draft as a trajectory to copy, DexFLEX treats it as evidence about intent: from the current tactile-proprioceptive state, it proposes feasible short-horizon motion chunks, predicts their contact consequences, and selects the candidate that best follows the command while preserving future contact stability. At inference time, a simple Draft-Dream-Select procedure combines pure-prior proposals for recovery with draft-seeded proposals for responsiveness, then decodes the selected motion into executable joint-space targets. Across degraded-command simulation, real-world shared control, and visuomotor policy learning, the same trained controller improves robustness without retraining, increasing real-world teleoperation success from 29.2% to 78.3% and reaching 46.7% policy-learning success, 3.5 times direct joint-action prediction, and 1.56 times prior-only correction. Videos are available at this http URL.
[AI-490] When Pleasing the User Compromises Care: AI Sycophancy in Caregiver-Mediated Healthcare
链接: https://arxiv.org/abs/2610.14360
作者: Priyadarsi Mishra,Arjun Beniwal,Ziyu Yi,Yu Ding,Sheng Liu
类目: Artificial Intelligence (cs.AI)
备注: 58 pages, 17 figures
Abstract:Large language model (LLM) alignment often assumes that the user interacting with a model is also the primary beneficiary of its behavior. Yet in many consequential settings, one person directs the system while another bears the consequences. We study this problem in caregiver-mediated healthcare, where parents, pet owners, and caregivers make decisions for children, animals, and older adults. We introduce an evidence-grounded benchmark spanning veterinary, pediatric, and older-adult care, and test whether AI advisors maintain appropriate care when users express clinically inferior preferences. Holding patient facts fixed, we vary stated care preferences and communication styles and evaluate both initial recommendations and multi-turn follow-up. Across frontier and open-weight models, inferior preferences substantially increase switching from appropriate to inappropriate care beyond neutral conversational drift and increase agreement with the requested action, while communication style alone has a smaller effect. These failures accumulate across turns, including on cases models initially answer correctly, and often carry moderate or severe potential harm. In matched eldercare cases, the effect of an inferior preference is similar whether it is expressed by the patient or by a caregiver. Together with models’ limited tendency to elicit patient preferences in caregiver conversations, this suggests that current models are highly sensitive to expressed user preferences but less sensitive to whose preferences are represented and who bears the consequences. Our results expose a broader limitation of user-centric alignment in settings with multiple stakeholders.
[AI-491] Actionable Meta-Requirements for AI Robustness Auditing
链接: https://arxiv.org/abs/2610.14356
作者: Joshua Owotogbe,Roya Nasiri,Indika Kumara,Damian Andrew Tamburri,Willem-Jan van den Heuvel
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:
Abstract:AI-enabled systems must remain dependable when the conditions under which they operate change (e.g., due to input perturbations, distribution shifts, or changes in external dependencies). This makes robustness auditing of such systems important for judging whether the available evidence is sufficient to support a defined robustness conclusion. However, designing an AI robustness auditing framework is challenging. A key unresolved issue is determining the requirements such a framework should support; these requirements have not yet been consolidated. We address this gap through meta-requirements: higher-order requirements specifying what a class of auditing frameworks should support. In design science research, literature and domain experts are common sources of meta-requirements; accordingly, we derive the initial set from literature and validate and refine it with practitioners. Our systematic literature review retained 74 of 2,890 records and derived 43 meta-requirements across 16 categories. We evaluated them with 128 practitioners, obtaining 5,504 ratings and 586 substantive comments. Overall, 81.20% of ratings indicated agreement or strong agreement. In particular, 41 of the 43 literature-derived meta-requirements were classified as strongly supported. Practitioner feedback refined the requirements and identified an additional requirement concerning external components and supply chain dependencies. The final set comprises 44 meta-requirements across 16 categories, of which 43 received quantitative validation. These meta-requirements can help organizations implement AI robustness auditing frameworks, assess existing frameworks for completeness, and refine them.
[AI-492] Natural Code Transformation Confuses Agents in Vulnerability Analysis
链接: https://arxiv.org/abs/2610.14350
作者: Jie Zhu,Ronik Bhaskar,Heather Zheng,Ben Y. Zhao,Kexin Pei
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Having demonstrated a strong potential for automating vulnerability discovery, AI agents can also be used to attack software systems at scale. Today’s software systems need mechanisms to level the playing field by slowing down potential attacks, giving them time to patch vulnerabilities and strengthen software. Along this line, our study of agent trajectories in vulnerability analysis shows a surprising finding: agents produce dramatically different results across real-world code commits that carry the same functionality and the same vulnerability. This phenomenon appears consistently in multiple code domains and vulnerability categories. It motivates us to study natural and functionality-preserving code transformations derived from real-world software development, and their potential as a defense against attacker agents. We build a corpus of natural transformations by mining recurring development patterns from real-world code commits, and apply them to evaluate impact on agentic vulnerability analysis. Our results show that even a single form of transformation can effectively “confuse” frontier agents, e.g., Codex with GPT-6 Sol, reducing successes in vulnerability discovery. We also study different agent failures, whether transformations can be composed to provide stronger protection, and adaptive attackers who seek to reverse them. Overall, our study of natural code transformation significantly expands the research space of using code transformations to defend against malicious agents. Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) Cite as: arXiv:2610.14350 [cs.CR] (or arXiv:2610.14350v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2610.14350 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-493] he RNN Homunculus: Learning to Gate Long-Term Memory
链接: https://arxiv.org/abs/2610.14279
作者: Zeyu Fan,Alexandru-Dragos Manolache,Jan van Gemert
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 20 pages, including references and appendices
Abstract:Learning long-range dependencies with recurrent neural networks (RNNs) is difficult because of the vanishing gradient problem. Gated recurrent neural networks like GRUs and LSTMs address this problem and are commonly explained by their ability to create additive copy paths through time. This explanation is correct, but incomplete: useful RNN gate values must themselves be learned, which is typically done with another RNN. We study this gate-learning problem through carefully controlled sequence classification tasks. Our analysis shows that gated architectures struggle when all training samples require long-range memory and thus equally suffer from the vanishing gradient problem. We intervene by adding short-dependency samples which let gated models learn solutions that capture long-dependency samples, but also generalize across different class labels. When properly learned, gated RNNs match or even exceed the test accuracy of modern architectures like Mamba and xLSTM. Overall, our results highlight gate learning as a central part of learning long-range memory and show how training data can support this process by exposing the same task over shorter dependencies.
[AI-494] Echo: Error-Compensated 4-Bit Tensor-Parallel Communication for LLM Training and Inference
链接: https://arxiv.org/abs/2610.14277
作者: Xingyu Xie,Xiaoyong Xie,Zhouchen Lin,Kim-Chuan Toh
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Tensor parallelism ends every layer with an all-reduce that sits between two dependent matrix multiplications, so its cost cannot be hidden and grows as compute outpaces interconnect bandwidth. Sending 4-bit instead of 16-bit values would cut this cost by nearly 4x, but the all-reduce must add the values it carries, and rounding the sum injects an error at every call. Model quality depends on how these errors accumulate over layers, yet existing methods treat each call on its own. We present Echo, a 4-bit all-reduce for training and inference that controls the accumulated error. Echo moves data in 4 bits and sums in high precision, so each value is quantized at most twice per call for any number of GPUs. The summation reveals each injected error, and Echo feeds it back into later calls, so old errors cancel instead of accumulating. We prove that the accumulated error stays bounded independently of depth. On A100 GPUs, Echo is 2.09x faster than the 16-bit NCCL all-reduce on 1 GB messages. It speeds up the time to first token by up to 1.16x and the training step by up to 1.09x on an idle interconnect, and by up to 1.44x and 1.38x as background traffic grows. It keeps the average accuracy over seven benchmarks within 0.4 points of 16-bit inference on Qwen3-8B. In 10B-token pretraining, its final loss is less than 0.9% above that of 16-bit training. Echo needs no FP8 or FP4 hardware.
[AI-495] Cortex: Measurement-Gated Evolution and Pruning of Coding-Agent Skills and Rules
链接: https://arxiv.org/abs/2610.14216
作者: Samuele Giampieri
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: 46 pages, 22 figures, 8 tables. Abstract abridged. Code and data: this https URL
Abstract:Coding agents such as Claude Code also follow text files that developers write for them: a project file, skills the agent may choose to open, and rules that load automatically when the agent reads matching files. We call these files the agent’s harness. Developers add to it whenever they correct the agent, but one success proves little: the same agent can pass a task once and fail it the next time. We present Cortex, a tool that tests changes to the harness instead of trusting them. It saves fixes from the developer’s sessions as replayable tasks, finds the corrections that recur, writes one skill or rule at a time, and keeps it only if repeated runs with and without it show a net gain and no task made noticeably worse. Existing text is tested the same way before it is deleted, and Cortex neither adds nor deletes text the agent never used. We tested Cortex with Claude Haiku 4.5 on a repository built for the purpose, with conventions the model tends to break, one area with nothing to learn, and a script playing the developer. On held-out tasks, the harnesses it built made the agent follow the conventions far more often than no skills or rules did, left the comparison area unchanged, and added no text read on every turn. Rules, which load automatically, did most of the work; skills helped only when offered with the files a task touches, and those offered on every turn were rarely opened. Cortex kept none of the useless or harmful suggestions we planted, but the agent never opened the useless ones or the harmful skills, and it reached no verdict on the harmful rules, so the test shows that Cortex keeps no unused text, not that it detects damage. Under the stronger Claude Sonnet 5 a single run’s harness added little, and on a real open-source project, where the agent was rarely corrected, Cortex found nothing to learn. We release the tool, the lab and its data. Comments: 46 pages, 22 figures, 8 tables. Abstract abridged. Code and data: this https URL Subjects: Software Engineering (cs.SE); Artificial Intelligence (cs.AI) Cite as: arXiv:2610.14216 [cs.SE] (or arXiv:2610.14216v1 [cs.SE] for this version) https://doi.org/10.48550/arXiv.2610.14216 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-496] Should SLMs be in the Drivers Seat? A Study of Local Models as Interpreters in Cloud-Written Agent Programs NEURIPS2026
链接: https://arxiv.org/abs/2610.14209
作者: Boris Radovič,Marco Canini,Veljko Pejović
类目: Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC)
备注: Accepted at SLM-Agents: 1st Workshop on SLMs for Agentic Systems, NeurIPS 2026
Abstract:The most accurate agents executing user-provided tasks are driven by large, cloud-hosted models, which solve them by reading the private data of the user. Agents powered by models small enough to run on edge devices keep that data local, but yield far worse results. A third option is to have cloud models write a program that devices execute: the data stays local, but as the program is written before any data is seen, it fails whenever a step depends on interpreting content that appears only at run time. We explore an intermediate architecture, in which the cloud-written program defers bounded questions over user data to a small local model through an ordinary function call. On AppWorld this improves the success rate of a code-only agent in all six combinations of backbone model and test partition we evaluate, by up to 14.9pp, recovering a substantial part of the gap to a cloud-hosted ReAct loop while keeping user data on the device.
[AI-497] Cross-Modal Latent Influence: Patch-Level Training Sample Attribution for Vision-Language Models
链接: https://arxiv.org/abs/2610.14190
作者: Laura Yao,Shixing Yu,Jie Peng,Promit Ghosal,Kyra Gan,Tianlong Chen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Explaining a vision–language model (VLM) prediction requires tracing training-example influence to the multimodal evidence that mediates it, yet existing attribution methods largely operate only at the sample level. Recent latent-mediated influence methods provide feature-level attribution in language models, but lack the multimodal and spatial structure needed to ground influence in VLMs. We introduce MOSAIC (Multimodal Orthogonalized Sparse Autoencoders for Influence Characterization). It learns a shared sparse image–text latent space whose concepts are 1) sparse within each modality, 2) supported on contiguous image regions, 3) aligned to the phrases that describe them, and 4) approximately decorrelated. It then applies EK-FAC to decompose training-example influence through these concepts and localize their contributions to specific text phrases and image regions. We evaluate MOSAIC on Qwen and Gemma across captioning, chest X-ray understanding, and visual question answering. The learned features are more localized and coherent without degrading task behavior, and ablating influence-ranked latents changes model outputs more than ablating latents ranked by activation or frequency. We find that visual evidence can substantially reorganize training-data attribution while preserving more of the shared latent structure. MOSAIC thus connects where model behavior comes from with what multimodal evidence carries that influence.
[AI-498] Visualizing the World Model in an Agents Value Function
链接: https://arxiv.org/abs/2610.14186
作者: Shawheen Naderi,Fatemeh Akherati,Yubei Chen
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 20 pages, 15 figues, 3 tables
Abstract:Could model-free reinforcement learning algorithms actually be learning a model of the environment? In this work, we apply counterfactual interventions which hold everything about a state fixed, besides a targeted change, to explore if the value function of a Proximal Policy Optimization agent forms a world model of the Crafter environment. Despite only being trained to optimize reward, we discover that the value function learns concepts such as the rules governing when a diamond is collectible, water and food having an impact on health, heuristics regarding inventory, and more. Finally, we discuss the safety implications of our visualizations and some shortcomings of current reinforcement learning agents that will prevent us from building truly advanced machine intelligence.
[AI-499] VETNET: Assembling Benchmark Datasets for AI-Driven Veterinary Innovation
链接: https://arxiv.org/abs/2610.14149
作者: Katherine J. Koebel,Renata Ivanek,Parminder Basran,Jennifer Sun,Casey Cazer,Gayatri Anil,Tommaso Banzato,Liz Barton,Dottie Cimino Brown,Leszek Bukowski,Marta Castelhano,Simon Coghlan,Christopher De Sa,Nikki Dutta,Hala Ghali,Julio Giordano,Joshua Glass,David C. Hogg,Nicholas Hollingshead,Adam Kiehl,Ziang Liu,Monica Meduri,Param Mohapatra,Kim Nayyer,Amelia Nelson,Peter-John Mantyla Noble,Christopher Pinard,Beryl Rabindran,Scott Ross,Ron Seccia,Moira Sheehan,Fei Wang,Tracy L. Webb,Kilian Q. Weinberger,Heather M. Whitney,Chen Xin
类目: Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注:
Abstract:VETNET represents an ecosystem of live benchmark datasets and evaluation frameworks designed to accelerate AI-driven innovation in animal science and veterinary medicine through standardized, high-quality datasets and multi-institutional collaboration across academic and non-academic sectors. This framework addresses the gap between AI advances in human medicine and the fragmented state of veterinary AI development.
[AI-500] Long-Horizon-Terminal-Bench 2.0: Automating Diverse Task Construction and Measuring Sustained Agent Progress
链接: https://arxiv.org/abs/2610.14146
作者: Zongxia Li,Zhongzhi Li,Yucheng Shi,Ruhan Wang,Junyao Yang,Chengsong Huang,Zhichao Liu,Xiyang Wu,Anhao Li,Yue Yu,Ergan Shang,Fuxiao Liu,Tong Zheng,Ninghao Liu,Lichao Sun,Haitao Mi,Leowei Liang
类目: Artificial Intelligence (cs.AI)
备注: 21 pages, 11 figures, 3 tables. Includes appendix
Abstract:Long-horizon terminal benchmarks measure whether an agent can sustain progress through extended workflows with process rewards, not only whether its final state passes a binary check. We present Long-Horizon-Terminal-Bench 2.0 (LHTB 2.0), an extension of Long-Horizon-Terminal-Bench that preserves its 46 original tasks while correcting task and evaluation-infrastructure defects and extending 31 tasks. LHTB 2.0 contains 77 tasks across 13 domains and eight task types. The extended tasks are produced through a systematic reconstruction-task pipeline. We start from a working scientific or engineering artifact, then use a parameterized reference implementation to generate multiple scenarios, reveal six as behavioral evidence, hold out three for evaluation, and use dense progress ladders to reward partial progress while preventing memorization of public outputs. We also extend the evaluation horizon from 90 minutes to three hours and record verifier checkpoints every 30 minutes, enabling direct measurement of sustained progress within a run. Across 1,145 scored model-task runs from 15 frontier models, GPT-6-astra leads with dense mean reward 0.69 and 37 tasks at R at least 0.95, followed by GPT-5.6-sol with dense mean reward 0.67 and 32 tasks at the same threshold. On the 30 reconstruction tasks, binary verifier scores average 0.221, whereas dense ladder scores average 0.653, showing that outcome-only evaluation discards substantial intermediate progress. LHTB 2.0 provides a broader, hardened, and process-aware testbed for studying reliable agent execution over hours.
[AI-501] Correction Training for Diffusion Models
链接: https://arxiv.org/abs/2610.14136
作者: Donghoon Ahn,Runqian Wang,Alexei A. Efros,Taesung Park,Yash Belhe
类目: Artificial Intelligence (cs.AI)
备注: Project page: this https URL
Abstract:Iterative sampling in diffusion models can lead to error accumulation due to exposure bias. The denoiser is trained on real images interpolated with noise, but at inference its inputs are constructed from its own imperfect predictions from the earlier steps, something it has never seen at training time. This might be one reason diffusion models often struggle without additional help, such as CFG. We introduce a correction step for diffusion models which attempts to reduce this training-inference gap. The corrector takes the base model’s prediction as input and is trained against the same ground-truth target. This makes correction an explicit learning task, with training inputs provided by the base model. During sampling, at each denoising step, we iteratively refine the denoiser’s clean-sample prediction. This allows us to correct predictions at intermediate denoising steps, preventing early artifacts from persisting into the final sample. Although the corrector is trained only on predictions from the base denoiser, it can be applied repeatedly to its own outputs at inference for better results. This simple framework leads to substantial improvement in sample quality while largely preserving diversity. We evaluate our method across DiT, SiT, and JiT. Our method substantially improves upon previously reported FID scores for unguided generation and, notably, unconditional generation, where producing high-quality samples without guidance has been particularly challenging. Our method achieves performance competitive with classifier-free guidance (CFG), and combining correction with CFG further improves diversity and reduces oversaturation typical of CFG. Finally, we show that the additional inference cost of correction can be amortized by distilling the refined predictions back into the base diffusion model, retaining the quality gains without additional inference-time overhead.
[AI-502] he Silent Failures in Agent ic Data Science
链接: https://arxiv.org/abs/2610.14115
作者: Xue Li,Hongqian Xia,Daniel Gomm,Aécio Santos,Cornelius Wolff,Madelon Hulsebos
类目: Artificial Intelligence (cs.AI)
备注: preprint
Abstract:Large language models have significantly lowered the barrier to writing code. Naturally, this has also led to the development of data science agents that promise automating the design and execution of data science from task formulation to model development. Existing evaluations of agentic data science (agentic DS) assess downstream predictive performance and code correctness, providing preliminary insight in agentic DS capabilities. We argue that these evaluations overlook the fundamental principles of data science: methodological data science- and domain knowledge. As such, only assessing downstream performance masks methodological failures such as data leakage and inappropriate validation, limiting the development of robust agentic DS systems. In this paper, we evaluate agentic DS beyond accuracy from three different angles: 1) a large-scale analysis of failure modes in common agentic DS frameworks and models across 220 data science tasks, 2) a reusable and extensible test suite with curated data science tasks that test for six specific failures in a controlled manner, and 3) a qualitative assessment of agentic DS pipelines against expert solutions on Kaggle competitions. We find that the majority of agentic solutions contain at least one failure mode despite the complexity of agentic harness or model capacity. Moreover, we identify a positive correlation between the occurrence of data leakage failures and predictive performance, confirming the necessity to evaluate agents beyond accuracy. Our qualitative analysis shows that DS agents generally lack problem solving creativity and domain knowledge, compared to competition-winning data scientists. We believe that principled failure mode analysis, as presented here, should be integral part of agentic DS evaluations, and that the insights we surface will be key in developing trustworthy agentic DS pipelines.
[AI-503] Embodied Zero-Shot Task Planning via Hardware-Aware LLM -VLM Integration for Grounded Robotic Manipulation ICML
链接: https://arxiv.org/abs/2610.14106
作者: Lonnie Tankersley,Noorbakhsh Amiri Golilarz,Soroush Korivand
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 6 pages, 5 figures, 1 table. Accepted and presented at the 25th IEEE International Conference on Machine Learning and Applications (ICMLA 2026), Oct 5-7, 2026
Abstract:Large language models can translate natural-language instructions into plausible manipulation plans, but robotic execution requires physically grounded decisions that respect robot embodiment, workspace geometry, reachability, collision constraints, and grasp feasibility. This paper presents a hardware-aware large language model-vision-language model (LLM-VLM) manipulation framework for a simulated Franka Research 3 robot in NVIDIA Isaac Sim. The framework separates symbolic reasoning from metric execution: an API-hosted LLM planner receives a structured scene summary and generates only symbolic task information, including the object to manipulate, target relation, approach preference, and objects to avoid. A local Qwen3-VL 8B VLM verifier supports semantic confirmation of ambiguous objects and open-container surrogates. Both the LLM planner and the VLM verifier are prohibited from generating robot coordinates, grasp poses, keepout dimensions, gripper commands, trajectory waypoints, or success labels. Instead, these decisions are handled by deterministic grounding and execution modules that use the robot model, scene geometry, MoveIt 2 collision checking, inverse kinematics, grasp validation, release verification, and final visual assessment. Execution feedback is returned only as symbolic context for replanning. The framework is evaluated through simulated digital-twin trials involving pickup, occluded-source access, keepout reasoning, relation placement, and sequential manipulation tasks.
[AI-504] SAGE: Efficient Protected Vision-Language Inference on Smart Glasses
链接: https://arxiv.org/abs/2610.14089
作者: Zhimin Li,Pan Wang,Jingxian Chen,Yuantao Tang,Qian Lou,Jingtong Hu
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE)
备注: 18 pages, 9 figures
Abstract:Smart glasses are a promising platform for vision-language assistants, but private split inference remains expensive: protecting visual representations before transmission does not remove the cost of on-device vision encoding, wide protected tokens, or hardware-unaware execution. We present SAGE (Smart-glasses-Aware Guarded Encoding), a hardware-aware framework for efficient protected VLM inference on smart glasses. SAGE keeps raw images and unprotected visual tokens on the trusted device, reduces visual input cost, executes shape-specialized INT8 graphs, and compresses protected representations into compact 256-dimensional INT8 packets before transmission. Unlike efficiency methods optimized only for nominal model size, SAGE selects its operating point using downstream answer quality together with measured latency, memory, and thermal behavior on real Mentra Live glasses. On a matched Qwen workload, SAGE reduces glasses inference latency by 39.18%, device-sensor temperature rise by 21.91%, and protected packet size by 96.56%, while retaining 85.62% of the protected reference’s relaxed utility and comparable prompted containment. Direct glasses-to-phone tests deliver all 192 packets and reduce median packet-plus-acknowledgment time from 71.3 to 10.0 ms. The compact protected interface also transfers across four additional VLM families. These results show that private wearable VLM inference must be optimized jointly across representation protection, communication, and real-device execution rather than through model compression alone.
[AI-505] wo-Stage Multi-modal Learning for Dynamic Audio-Visual Navigation: From Physical Alignment to Scene Aggregation
链接: https://arxiv.org/abs/2610.14077
作者: Jia Li,Yinfeng Yu
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Robotics (cs.RO); Audio and Speech Processing (eess.AS); Signal Processing (eess.SP)
备注: Main paper (10 pages). Accepted for publication by IEEE Transactions on Audio, Speech and Language Processing(TASLP 2026)
Abstract:Embodied Audio-Visual Navigation (AVN) requires an agent to locate sound targets by jointly leveraging visual and auditory cues. While prior work has achieved notable success with static sound sources, real-world scenarios often involve moving targets, giving rise to the more challenging problem of \textbfDynamic Audio-Visual Navigation (D-AVN). Existing approaches largely rely on semantic auditory features, which limits generalization to unheard sound categories and hinders fine-grained spatial reasoning. To address these challenges, we propose a novel \textbfTwo-Stage Multi-modal Learning framework for D-AVN. In the Physical Alignment stage, the proposed \textbfIntensity-Calibrated Geometry Encoder (ICGE) performs physics-informed alignment of sound intensity with visual geometry to capture category-agnostic spatial correspondence. In the Scene Aggregation stage, a Transformer-based \textbfCross-Modal Scene Aggregator (CMSA) dynamically integrates multi-modal features to construct a robust state representation for policy learning. Extensive experiments on Replica and Matterport3D demonstrate that our method achieves state-of-the-art success rates, while significantly improving generalization and navigation efficiency in complex environments and under unheard sound scenarios.
[AI-506] One Model Multiple Compute Budgets: Multi-Ratio ConceptMoE for Flexible Compute Across Turns
链接: https://arxiv.org/abs/2610.14074
作者: Yiyang Fu,Zihao Huang,Dongchen Han,Shukai Gong,Zhen Dong,Qiyang Min,Daquan Zhou
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models typically support different inference budgets through separate model sizes or fixed compute configurations. In multi-turn interactions, however, compute demand can vary substantially across turns. Using a high-compute model for every turn wastes computation, while routing among independent models may require rebuilding model-specific history states after switches. We therefore ask whether a single shared model can support multiple compute budgets that can be flexibly assigned across turns. We introduce Multi-Ratio ConceptMoE, which jointly supports token-to-concept compression ratios of 2, 4, and 8 (CR2/CR4/CR8) within a shared backbone. Lightweight full-token modules reduce compression-independent cost, ratio-specific boundary routers support distinct compression levels, and cross-ratio training enables mixed-ratio processing within the same context. Across PTBench and SFTBench, the three fixed modes outperform their compute-matched MoE baselines in all 6 comparisons while remaining competitive with independently trained fixed-ratio specialists. On CodeChat and CoSQL, mixed-ratio inference improves the accuracy–compute trade-off, outperforming the Large MoE by 5.87 and 2.68 percentage points while reducing FLOPs by 41.1% and 12.3%, respectively. Under the same turn-wise budget schedule, Multi-Ratio ConceptMoE further reduces FLOPs by 78.4% and 46.9% relative to routing among independent models. These results demonstrate that a single shared model can provide multiple compute budgets for flexible compute allocation across turns.
[AI-507] Auditing LLM Program Search for Bermudan Stopping with Primal–Dual Certificates NEURIPS2026
链接: https://arxiv.org/abs/2610.14063
作者: Yuhua Qian
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted at the MATH-AI Workshop at NeurIPS 2026
Abstract:LLM program-search agents are usually judged by the score that drives their search. We study LLM-driven search over Python exercise policies for Bermudan options, an optimal stopping problem where a trusted evaluator owns the simulation and a dual upper bound~ U . For any policy, 0 \le V^\star - L_\pi \le U - L_\pi ; we report Monte Carlo estimates of both bounds with their sampling error. We audit 24 generation attempts from two pipelines, a six-pair feedback-representation ablation, and a holdout evaluated after selection was frozen. No selected generated program improves on Ridge LSMC consistently or by more than Monte Carlo uncertainty; holdout gap scores lie within 0.002 of it. On six one-dimensional puts, a high-precision binomial reference places Ridge within Monte Carlo noise of optimal and attributes 99% of the estimated gap to dual-bound looseness, so this dual audits validity but cannot certify progress at the search’s scale. Structured memory showed no benefit over score-only memory in this ablation. Whether search ``worked’’ depends on reporting rules: retaining the initial reference, freezing selection before holdout, and counting failures.
[AI-508] Reinforcement Learning from Real-Robot Interaction without High-Quality State Estimates ICRA2027
链接: https://arxiv.org/abs/2610.14042
作者: Ryan Gao,Ashvin Anilkumar,Nicholas Corrado,William Cong,Josiah Hanna
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 9 pages, 7 figures, submitted to ICRA 2027
Abstract:A longstanding goal for robot reinforcement learning (RL) is that robots learn from their own physical experience, as humans and animals do. A key obstacle to learning directly from physical experience, especially outside instrumented lab environments, is that the state of the world is usually unobservable to the learner and must be estimated from on-board sensing. Errors in that estimate corrupt both the policy input and potentially the reward signal. We study this challenge using the contact-rich task of a bipedal robot learning to score soccer goals. We estimate all key state variables on-board with an unscented Kalman filter (UKF) and no external instrumentation (e.g., motion capture or overhead cameras). Because the UKF may incorrectly estimate state when the ball is unobserved, we learn a new state representation with a recurrent state-space model (RSSM) that, critically, only corrects the RSSM’s latent state when the ball is observed. Building on RLPD, we introduce RSSM-RLPD and apply it to the task of a NAO robot learning to dribble a soccer ball into a goal. We show that the robot learns to score from multiple starting positions after 30 minutes of real-world interaction, reaching 100% success against 0% for a feed-forward policy under the same training budget. We complement our physical robot experiments with a simulation study and show that RSSM-RLPD learns policies that achieve a higher success rate than behavior cloning, a feed-forward, and GRU baseline. Taken together, our results show that the addition of a learned state representation to existing on-robot learning methods enables RL from real-robot data even with a high degree of partial observability.
[AI-509] A Geometric Perspective on Reward Function Updates in Inverse Reinforcement Learning NEURIPS2026
链接: https://arxiv.org/abs/2610.14041
作者: Anish Diwan,Jan Peters,Oleg Arenz
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: Accepted as a conference paper at the The Fortieth Annual Conference on Neural Information Processing Systems (NeurIPS 2026)
Abstract:Inverse Reinforcement Learning (IRL) is classically solved using Lagrangian optimisation, by jointly optimising a primal-dual problem with gradient descent to obtain the optimal reward function parameters and corresponding optimal policy. While this algorithmic view of IRL has been highly influential, gradient-based reward updates only paint a partial picture, largely ignoring the geometry of the problem in occupancy space. In this paper, we present an occupancy-space view of reward updates in IRL. We leverage the insight that both the optimal solution and its geometry in occupancy space are known, and show that the geometry along the e-geodesic (straight line in occupancy space) toward the optimal occupancy directly affects the gradient of the IRL dual problem in reward parameter space. This perspective naturally suggests preconditioning the dual gradient with the average Fisher information along the occupancy trajectory. We show that this preconditioned gradient is just a convex interpolation of the old reward and a new target reward, yielding efficient natural-gradient-style updates with little additional computational overhead. We then analyse these occupancy-space reward updates from a Mirror Descent (MD) perspective, and show that policy search over the resulting rewards induces approximate entropic MD on the policy. Finally, we conclude with convergence results of such geometry-aware IRL algorithms.
[AI-510] Rotate the Loop: Blockwise Conformality for Looped Transformers NEURIPS2026
链接: https://arxiv.org/abs/2610.14030
作者: Alexey Zemtsov,Albina Klepach,Vadim Lomshakov,Andrei Polubarov,Nikita Lyubaykin,Alexander Derevyagin,Vladislav Kurenkov,Alexander Nikulin
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注: Accepted at NeurIPS 2026 Workshop “Transitioning from Pre-Training to Post-Training”
Abstract:Looped transformers reuse one shared block across many recurrent steps, so depth becomes a quantity that can be spent at runtime rather than a stack of distinct layers. Those extra steps help only if a write made on one recurrent step is still readable on later ones, and that depends on the linear map applied to the residual between recurrent steps. We show that whether the loop learns to track state is a geometric question about the form of this map, and we propose blockwise conformal injections: the residual splits into coordinate-aligned groups and rotates inside each one, so every group has its own timescale while the geometry stays exact within it. Our experiments show a performance increase on a demanding form of state tracking - last-token accuracy when the rest of the prefix is already correct - and on language modelling, where a signal written earlier has to be retrieved at the end of the sequence.
[AI-511] Can We Trust This Evaluation? Instance-Level Reliability of LLM -as-a-Judge
链接: https://arxiv.org/abs/2610.14029
作者: Yang Ba,Michelle V. Mancenido,Rong Pan
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注:
Abstract:LLM-as-a-judge evaluation commonly aggregates multiple judges’ decisions into a majority vote, yet this panel decision process does not guarantee that the evaluation of an individual item is reliable. We propose Structured Cross-fitted Uncertainty Risk (SCUR), a label-free approach for estimating instance-level evaluation reliability based on the judge panel’s behavior. A latent measurement model serves as the backbone for measurement system analysis, from which SCUR derives three complementary signals capturing panel uncertainty, measurement cross-entropy, and perturbation sensitivity. These three sources of evidence are combined into a risk score. We evaluate the SCUR approach on five real-world benchmarks using majority-vote error ranking, selective trust, and risk stratification. Furthermore, we show our approach to be more effective at detecting high-consensus risk items, which are the blind spots for many other LLM-as-a-judge evaluation methods.
[AI-512] Pick the Mixture Not the Samples: Combinatorial Data Selection for LLM s
链接: https://arxiv.org/abs/2610.14028
作者: Luyang Zhang,Jialu Wang,Bebei Li
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Data selection for LLM fine-tuning faces two facts about real training: data pools draw from many distinct domains, and evaluation spans diverse tasks. As a result, the same sample helps different tasks differently, and samples interact during training so that a subset’s value is not the sum of its parts. Current data-selection methods leave these realities largely unaddressed. We formalize them as Combinatorial Data Selection (CDS), a framework that separates attribution (how samples are scored) from selection (which subset to pick) and models the non-additive interactions among samples. On a real instruction-tuning data pool, we find substantial cross-domain complementarity (joint utility exceeds the sum of singletons), alongside redundancy within domains. Inspired by this, we propose Mixture-then-Select, a two-stage algorithm that first decides how many samples to take from each domain, then ranks samples within each domain by attribution score. Theoretically, the gap to the optimal subset is linear in the selection size and independent of data pool size. Empirically, Mixture-then-Select outperforms strong baselines by 20 to 54%, is robust across attribution methods, and works across different models and data pool sizes. The framework also extends to the data purchasing problem under a budget constraint when sample prices vary. Our findings suggest that allocating selection across domains offers a promising path toward more efficient and effective data selection for modern LLMs.
[AI-513] OPMD: On-Policy Mechanism Distillation Beyond Output Supervision
链接: https://arxiv.org/abs/2610.14022
作者: Yingdong Shi,Zhihong Li,Changming Li,Jingyi Yu,Quan Li,Kan Ren
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 21 pages, 6 figures
Abstract:On-policy distillation (OPD) improves language-model by supervising a student on prefixes sampled from its own policy, but its supervision remains confined to the output space. Consequently, OPD transfers what a teacher predicts without supervising the intermediate computation that produces those predictions. Simply distilling intermediate representations like representation matching is insufficient since teacher and student representations may differ geometrically, and even well-aligned hidden states need not be usable by the student’s subsequent computation. We introduce On-Policy Mechanism Distillation (OPMD), which transfers teacher-derived intermediate computation while explicitly accounting for its compatibility with the student. OPMD first constructs a student-compatible teacher that transforms teacher representations into intermediate states usable by the student’s downstream computation, then it internalizes these states from on-policy trajectories. OPMD achieves the highest average accuracy among the evaluated distillation baselines in all three teacher–student settings, and generalizes to multi-teacher distillation and cross-family supervised fine-tuning. Further analyses show that mechanism supervision yields more effective reasoning and meaningfully alters the student’s intermediate computation and solution strategies, supporting that it transfers information beyond output supervision alone. Together, these results establish mechanism-level supervision as an effective way to move on-policy distillation beyond output supervision, enabling students to learn inner mechanisms they can use and internalize.
[AI-514] LabInstruct: Benchmarking Situated Instructional Video Generation for Lab Procedures
链接: https://arxiv.org/abs/2610.14020
作者: Yuming Fu,Weijia Wu,Jing Chen,Jiahao Tang,Feifei Chen,Hongyu Zhu,Xin Jin,Alex Jinpeng Wang
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Self-driving laboratories (SDLs) aim to automate the full experimental loop, from scientific decision-making to physical execution. Ideally, AI-generated plans could be carried out directly by robotic systems, but reliable automation remains difficult in complex, open-world laboratory environments, where experiments often involve fine-grained manipulation, long-horizon procedures, and substantial variation across tasks and setups. Humans therefore remain an important execution interface between AI-generated plans and physical experiments, creating a need for clear and effective human-facing experimental guidance. Because laboratory procedures are inherently visual, spatial, and dynamic, video is particularly well suited to communicating apparatus configurations, manipulation actions, temporal dependencies, and state changes. Recent advances in video generation now make it possible to synthesize experimental demonstrations directly from an initial workspace image and a natural-language instruction. However, whether such models can reliably communicate real laboratory procedures has not been systematically studied. We introduce LabInstruct, a benchmark for situated instructional video generation in real laboratories. LabInstruct contains 204 tasks across 5 scientific disciplines, with real reference executions and structured annotations of objects, actions, contacts, and state transitions. Evaluating 8 frontier image-to-video models, we find that visually plausible generations frequently remain procedurally incorrect, revealing a substantial gap between visual realism and the reliability required for experimental instruction.
[AI-515] RoboHarness: A Simple Harness Outperforms VLA and World Actions Models
链接: https://arxiv.org/abs/2610.14018
作者: Bince Qu,Wei Chen,Bo Zhang
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 18 pages, including references and appendices
Abstract:The rise of large language models (LLMs) has opened a path toward generalist embodied AI, raising the question of how LLMs can act in the physical world. One straightforward approach is training the model to generate actions: collect robot-action data and video demonstration, refine LLM architecture with an action head, and train a vision-language-action (VLA) model at scale. However, we argue that coding agents have the potential to realize generalist embodied manipulation. All they need is a simple yet effective robot harness. We present RoboHarness: a robotic harness that gives LLM agents a visual-geometric control panel so that they can directly understand and invoke embodied tasks. With RoboHarness, Qwen3.8-Flash-Next outperforms state-of-the-art VLA models on BEHAVIOR Challenge 2025 tasks. The implementation will be released as open source.
[AI-516] Bias Robustness of Genetic Multi-Criteria Sorting versus Learned Rankers under Proxy Skew and Label Contamination
链接: https://arxiv.org/abs/2610.14015
作者: Vladyslav Shevchuk,Carlos Uziel Pérez Malla,Philipp Wissgott
类目: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 12 pages, 4 figures
Abstract:Learning-to-rank (LTR) models can align rankings with undesignated signals when those signals appear as features or leak into training labels. We compare label-free genetic multi-criteria sorting (open Pikaia strategy variants), which reranks fixed candidate pools using an explicit criteria vector C, to label-trained LTR (LightGBM, XGBoost, RankSVM) under two controlled protocols on MSLR-WEB10K, Baidu-ULTR, WANDS, and Expedia. In the main proxy-skew study, labels remain clean while a calibrated proxy P is available as a feature: Genetic AI rankings show lower Spearman rank-proxy correlation than the LTR peers (e.g. about 0.15-0.16 vs. 0.59-0.87 on WANDS), with both families evaluated jointly on clean nDCG@10 on a bias-utility frontier. In a supporting label-contamination study, training labels are mixed with contaminant B: because Genetic AI methods do not use labels, their bias and utility curves stay flat, whereas LightGBM and XGBoost increase rank-B correlation as contamination rises (OLS bias_slope about 0.58-0.65 on WANDS/Baidu). The results support genetic sorting as an architectural option for optimizing declared attributes on a fixed pool when undesignated proxies or label contaminants must be refused.
[AI-517] Physics-Informed State Space Models for Day-Ahead Photovoltaic Power Forecasting
链接: https://arxiv.org/abs/2610.14014
作者: Phil Aupke,Muhammad Hassan,Khalid Ali,Seema,Andreas Kassler,Andreas Theocharis
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Accurate day-ahead photovoltaic (PV) power forecasting requires both strong temporal modeling and adherence to physical PV system constraints. Purely data-driven models often ignore the underlying PV conversion process and may produce predictions that are outside the operational range of a PV system. In this paper, we propose a physics-informed state space framework that combines a differentiable PV conversion model with a selective state space backbone (PowerMamba). Our approach follows a physics-residual decomposition. The physics module provides a structured baseline based on solar geometry, irradiance, temperature effects, and inverter limits, while the state space model learns residual corrections from numerical weather prediction (NWP) inputs over a 24-hour horizon. Experiments on one year of utility-scale PV data show that the proposed method significantly improves forecasting accuracy, reducing RMSE by 38.5% and MAE by 45.8% compared to a physics-only baseline, and outperforming a purely data-driven model. Results further show that incorporating system-level physics provides the largest performance gains, while additional refinements offer only marginal improvements. Beyond accuracy, the model improves robustness by constraining predictions toward physical feasibility, reducing unrealistic predictions under corrupted measurements.
[AI-518] When Log Parsing Removes What Matters: Detecting Failures in Distributed GPU Workloads
链接: https://arxiv.org/abs/2610.14009
作者: Boyuan Cui,Yuyang Zeng,Yifang Tian,Hans-Arno Jacobsen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC); Software Engineering (cs.SE)
备注:
Abstract:Large AI models increasingly rely on distributed GPU execution, yet it remains unclear whether established failure analysis methods transfer effectively to this setting. We introduce GF-Bench, a PyTorch-based framework and dataset for generating diverse failures in distributed GPU workloads. The framework supports extensible workload and fault-injection logic; based on this design, the current release targets failures visible in multi-GPU communication logs and provides 220 labeled two-GPU executions covering normal runs, CUDA out-of-memory fail-stop failures, and persistent straggler fail-slow failures. Using GF-Bench, we evaluate three sequence-based methods (DeepLog, LogAnomaly, and LogBERT) and the LLM-based LogPrompt on parsed logs, and additionally evaluate LogPrompt on raw logs. On parsed inputs, no method reliably detects both failure classes, and the highest overall F1 score is 0.667. Raw inputs raise the overall F1 score of in-context LogPrompt to 0.983, at 9.0x its average input-token usage and 8.7x its per-run cost. These results expose a substantial trade-off between effectiveness and efficiency: generic log parsing lowers operational cost but discards the distributed execution context required for accurate failure detection. GF-Bench provides a reproducible basis for studying cost-efficient analysis methods and can be extended to additional GPU workloads and failure types.
[AI-519] What Can Language Models Inherit from Solomonoff Induction? Limits Attainable Properties and What Served Conditionals Reveal
链接: https://arxiv.org/abs/2610.14007
作者: Luan Ozelim,Abicumaran Uthamacumaran,Hector Zenil
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Information Theory (cs.IT)
备注:
Abstract:A familiar line of reasoning runs as follows: language models are flexible approximators trained under a strictly proper loss, so scale brings them closer to Solomonoff induction, and their predictive success is evidence of that approach. We apply known results from algorithmic information theory, universal prediction and Bayesian consistency. In its uniform form the conclusion holds for no computable model: each has cheap environments it never learns, and every uniformly computable family of total predictors shares one on which each member fails beyond a horizon growing with its index, so ``closer’’ can hold only pointwise, one environment at a time. Nor could the offered routes certify even that: finite evaluation cannot certify agreement on rarely reached contexts, perfect corpus fit selects the corpus law as prior, and asymptotic criteria cannot see the constants that carry Solomonoff’s guarantee. Our lens is the Kraft inequality read for predictors. Every model has at most the same budget of predictive success, so Solomonoff’s predictor, which leaves part of it unspent, differs from a language model not in having more of it but in where it spends it; on mutually singular families this is exact, and only full-mass Bayes mixtures reach the frontier. We then discuss what is inheritable, including in-context learning as amortised Bayesian inference; what survives within a context window; which Solomonoff-like properties a deployed system should have or avoid; and what served conditionals determine. A closing table states what common practices certify and what to report. Tests are developed in a companion paper.
[AI-520] Coding LLM s Can Be Harnessed to Learn Interpretable Models from Data
链接: https://arxiv.org/abs/2610.14006
作者: Aditya Kumar,Cassandra A. Cohen,William W. Cohen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Main Text: 10 pages, 4 figures
Abstract:Since compact, interpretable classifiers correspond to simple programs, it is natural to ask if LLM-based coding agents can be used to learn interpretable models from data. In this paper we present a simple harness that repeatedly prompts a coding LLM to produce the next rule in a rule set, given a sample of labeled examples. We show that this approach is competitive with, and often superior to, state-of-the-art rule-learning methods on tabular data, especially with small training sets, and wins by large margins on text, including against another LLM-based rule learner; that this approach outperforms standard harnesses, and that the approach immediately leads to new kinds of interpretable models, since coding agents are not restricted to search a highly restricted hypothesis space. In addition to the advances made in learning interpretable models, these results are of interest because they introduce a new class of coding benchmarks - construction of interpretable classifiers - which includes quantitative measurements of output quality (i.e., test-set error) and existing mature ``specialist coding algorithms’’ (i.e., existing learning methods).
[AI-521] PreSO: Privileged Representation Isolation for On-Policy Self-Distillation
链接: https://arxiv.org/abs/2610.14001
作者: Yuhang Yang,Mingxuan Xia,Chao Ye,Shuai Zhu,Shouqing Yang,Haobo Wang,Shenzhi Yang,Zhengqing Zang,Chenglong Wang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:On-policy self-distillation (OPSD) uses a self-teacher conditioned on privileged information to turn the sparse outcome signal of reinforcement learning with verifiable rewards (RLVR) into dense token-level supervision. Yet the teacher’s advantage may depend on information unavailable to the student at inference, raising a central question: which privileged changes are worth distilling? We use cross-problem recurrence as an empirical cue for potentially reusable guidance. Comparing the representation shifts induced by full solutions and compact hints, we find that hint-induced shifts are more concentrated and directionally consistent across problems, and yield more sustained downstream gains despite revealing less information. Motivated by this association, we propose Privileged Representation Isolation for On-Policy Self-Distillation (PreSO). At the source, PreSO uses compact hints to limit the solution-specific detail entering the privileged shift. At the target, it estimates a leading low-rank subspace from shifts pooled across related problems, projects the privilege-induced correction onto this subspace, and adds it to the student’s unprivileged representation to form the representation-level distillation target. The selected subspace serves as an operational proxy for recurring structure, rather than a guaranteed separation of transferable and instance-specific information. On five competition-level mathematics benchmarks, PreSO consistently outperforms strong baselines across Qwen3-1.7B, 4B, and 8B, improving average accuracy over the base model by 6.8, 9.0, and 10.0 points. Beyond peak accuracy, PreSO attains the highest final-checkpoint average at every model scale, with improvements that persist through the end of training.
[AI-522] VH-GPO: Verified Handoff for LLM Agent Reinforcement Learning
链接: https://arxiv.org/abs/2610.14000
作者: Haoze Li,Peng Zhou,Peiyuan Wu,Weijie Zhao,Yinan Xin
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Group-relative reinforcement learning methods such as GRPO train large language model (LLM) agents by comparing rollouts for the same task. When every trajectory in a group fails with the same reward, relative advantages collapse toward zero, leaving no useful learning signal. A frozen teacher can recover from a student-visited state, but letting it complete the task makes the recovered suffix entirely teacher-generated. We ask: after repairing a student failure, when should control return to the student? We propose outcome-verified teacher-student handoff, instantiated in Verified Handoff for Group Policy Optimization (VH-GPO). The teacher provides enough support to reach a state from which the student can succeed, then returns control so that successful experience is again student-generated. Given an environment-verified teacher recovery, VH-GPO selects the earliest independently confirmed handoff point found within a bounded search budget. With Qwen2.5-1.5B, VH-GPO increases success rates over GRPO from 59.38% to 94.79% on ALFWorld and from 71.62% to 77.34% on WebShop. Improvements also hold with the 7B student, for which additional teacher and student output tokens amount to 1.77% and 2.61% of regular student rollout output tokens on ALFWorld and WebShop, respectively. Ablation studies demonstrate the effectiveness of the key components. These results support using verified teacher assistance to repair failures while reinforcing the student’s ability to complete tasks independently.
[AI-523] Enhanced Trading Systems with Intelligent Market Analysis
链接: https://arxiv.org/abs/2610.13998
作者: Vedanth Padmaraman,Shreen Anand,Vamshi Vishruth J,Vineet V Pai,Arvind Srinath K.T,Subramanyam Natarajan,Shylaja S S
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 9 pages, 1 figure, 2 algorithms, 1 table. Accepted at ICAIA 2026 (proceedings pending)
Abstract:This paper proposes a hybrid reinforcement learning trading framework integrating intraday market data with transformer-based financial sentiment signals. The system extracts sentiment dynamics and semantic event clusters from financial news and incorporates them into a PPO-based trading agent. Experiments on intraday NSE data demonstrate improved risk-adjusted returns compared to price-only baselines.
[AI-524] Splicing Hints into Reasoning Traces: Learning from Demonstrator Failures
链接: https://arxiv.org/abs/2610.13996
作者: Hamed Mahdavi(Pennsylvania State University),Pouria Mahdavinia(Pennsylvania State University),Majid Farhadi(Google),Vasant Honavar(Pennsylvania State University)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 39 pages, 3 figures
Abstract:Reinforcement learning with verifiable rewards improves mathematical reasoning when a final answer can be checked, but a correct answer does not certify the derivation behind it, and proof problems offer no answer to check. Replacing the check with a reasoning model as an in-loop verifier faces a fundamental obstacle: the policy must already produce occasional correct attempts for any reward to exist, which it may never do at olympiad level. We instead move verification and repair out of the training loop and into offline data generation with SPLICE, a workflow in which a second model, given only a reference solution, edits a reasoning model’s traces as they are generated: it verifies progress, truncates a failed trace at its last verified step, and splices in an adaptively strengthened hint until progress resumes. The resulting dataset includes repaired examples whose sampled unaided trajectory failed, with failed continuations removed and progress repeatedly checked against the reference solution. A model finetuned on SPLICE traces outperforms the same model distilled from the reasoning model’s unedited traces on AIME and, under a grading protocol validated against human grades, on olympiad proofs. The trained model’s own traces show a corresponding behavioral signature: at difficult points it announces a branch in the style of the spliced hints and takes riskier steps, with predictive entropy rising there and just after, a pattern absent from the model trained on unedited traces.
[AI-525] Predict Before You Modify: Learning Improvement Dynamics for Self-Evolving Agents
链接: https://arxiv.org/abs/2610.13991
作者: Zixuan Xie,Chang Xu,Yuhua Jiang,Li Zhao
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Self-evolving agents improve through repeated cycles of modification and evaluation. However, existing methods receive feedback on modifications only after costly evaluations, and aggregate scores may obscure distinctions between failures that require different repairs. We introduce predictive self-evolution through an editable Improvement Dynamics Model (IDM). Before evaluation, the IDM predicts task-level outcomes of a proposed modification and guides an optional revision. After evaluation, past modifications and their observed outcomes are used to improve the IDM, allowing predictive knowledge to accumulate across generations. The system can thus improve how it solves tasks, proposes modifications, and predicts their effects. Building on HyperAgents, we evaluate this approach across multilingual coding, paper review, robotics reward design, and Olympiad-level mathematical grading. Our approach outperforms HyperAgents across all four domains, with up to 46% higher average performance over generations. Gains also persist under matched token budgets that include both agent evolution and task evaluation, and extend to held-out tasks. Forecast accuracy improves as the IDM evolves, and the learned predictive knowledge transfers to improve evolution in a new domain.
[AI-526] MRFlow: Mechanism-Structured Residual Flow Matching for Virtual Cell Perturbation Prediction
链接: https://arxiv.org/abs/2610.13966
作者: Haojie Yang,Ran Su
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Quantitative Methods (q-bio.QM)
备注:
Abstract:Predicting cellular responses to chemical and genetic perturbations is a central test for AI virtual cells, yet many single-cell models reduce each condition to an average expression shift. This target is misaligned with perturbation biology: rare states, resistant subpopulations, and context-dependent effect sizes may coexist under the same intervention. MRFlow treats virtual-cell prediction as conditional transport between empirical single-cell measures. Control and perturbed cells are embedded in a variational latent state space, greedy low-cost supervision is constructed over detached minibatch states, and a conditional flow-matching vector field transports control populations toward perturbed populations. The field decomposes into a mechanism-structured affine dictionary, which recombines shared latent response operators with a low-rank condition-specific correction, and a spectrally controlled residual field for nonlinear cell-level variation. Genetic perturbations are encoded through a regulatory-graph GCN, chemical perturbations through Morgan fingerprints, and both enter the same conditional dynamics interface. Across eight gene and drug perturbation benchmarks, MRFlow attains the highest composite score while preserving competitive raw performance on differential-expression and distributional metrics. The results support latent distribution transport as a principled and empirically effective formulation for perturbation-aware virtual cell modeling.
[AI-527] Dependency Compression and Synergy: A Unified Information-Theoretic View of Multimodal Learning
链接: https://arxiv.org/abs/2609.14421
作者: Liangjian Wen,Linjie Li,Jiang Duan,Yong Dai,Jianzhuang Liu,Zhao Kang
类目: Information Theory (cs.IT); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Recent advances in multimodal foundation models have intensified the need to understand how different modalities share, preserve, and complement information. Mutual Information (MI), the Information Bottleneck (IB), and Partial Information Decomposition (PID) provide complementary perspectives, yet existing studies often treat them as isolated tools. This survey presents an information-theoretic perspective connecting these principles as progressively refined views of multimodal information processing: MI characterizes inter-modal dependency, IB explains task-oriented information preservation under compression, and PID decomposes preserved information into redundancy, uniqueness, and synergy. We review 170 recent studies (2018–2026) and 12 foundational works, organizing multimodal learning around four challenges: cross-modal alignment, information-efficient fusion, interaction-type characterization, and scaling to multimodal foundation models. Rather than using application domains as primary taxonomy axes, we interpret healthcare, robotics, recommendation systems, affective computing, and wireless communications as empirical validations of these information principles. Beyond taxonomy, we organize existing multimodal paradigms within a single information-theoretic coordinate system – the Generalized Multimodal Information Lagrangian – in which they occupy exact or approximate parameter corners, and whose unoccupied regions name candidate method families the literature has not yet built. We further discuss how emerging multimodal foundation models instantiate these principles at scale and identify open challenges including scalable information estimation in high-dimensional settings, standardized evaluation across information-theoretic methods, combinatorial complexity of multimodal PID, and the transition from post-hoc information analysis toward information-aware multimodal learning.
[AI-528] Latent Algorithmic Structure Precedes Grokking: A Mechanistic Study of ReLU MLPs on Modular Arithmetic
链接: https://arxiv.org/abs/2603.23784
作者: Anand Swaroop
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 9 pages, 5 figures
Abstract:Grokking-the phenomenon where validation accuracy of neural networks on modular addition of two integers rises long after training data has been memorized-has been characterized in previous works as producing sinusoidal input weight distributions in transformers and multi-layer perceptrons (MLPs). We find empirically that ReLU MLPs in our experimental setting instead learn near-binary square wave input weights, where intermediate-valued weights appear exclusively near sign-change boundaries, alongside output weight distributions whose dominant Fourier phases satisfy a phase-sum relation \phi_\mathrmout = \phi_a + \phi_b ; this relation holds even when the model is trained on noisy data and fails to grok. We extract the frequency and phase of each neuron’s weights via DFT and construct an idealized MLP: Input weights are replaced by perfect binary square waves and output weights by cosines, both parametrized by the frequencies, phases, and amplitudes extracted from the dominant Fourier components of the real model weights. This idealized model achieves 95.5% accuracy when the frequencies and phases are extracted from the weights of a model trained on noisy data that itself achieves only 0.23% accuracy. This suggests that grokking does not discover the correct algorithm, but rather sharpens an algorithm substantially encoded during memorization, progressively binarizing the input weights into cleaner square waves and aligning the output weights, until generalization becomes possible.
[AI-529] Falsification-proof non-market allocation mechanisms
链接: https://arxiv.org/abs/2610.18625
作者: Eduardo Perez-Richet,Vasiliki Skreta
类目: Theoretical Economics (econ.TH); Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT)
备注:
Abstract:Relying on manipulable characteristics to target the allocation of resources is fraught with difficulties. Naively optimal rules lead to falsification and allocative inefficiency. Optimizing the rule for allocative efficiency while accounting for falsification typically induces falsification, at a heavy cost to the agents. Moreover, falsification generates negative externalities. We characterize optimal falsification-proof rules under a broad range of falsification technologies. We also provide an honesty condition under which they solve the unconstrained problem of a planner who maximizes a weighted sum of allocative surplus and agents’ welfare.
[AI-530] ask-evoked eye movements as candidate digital biomarkers of depression severity in adolescents and young adults
链接: https://arxiv.org/abs/2610.18494
作者: Youxiang Sun,Yujie Nie,Zhuoru Wu,Mengyuan Yu,Wenkong Wang,Xiang Li,Yumei Wang,Xin Ma,Lizhou Fan
类目: ignal Processing (eess.SP); Artificial Intelligence (cs.AI)
备注:
Abstract:Objective digital measures of depression severity could complement conventional symptom assessments and facilitate more timely clinical evaluation. Here, we investigated whether task-evoked eye movements provide interpretable behavioral markers of depression severity in adolescents and young adults. In this single-center, cross-sectional study, 155 participants aged 14–20 years completed emotional-spatial free-viewing and memory-guided recall tasks and were stratified by 17-item Hamilton Depression Rating Scale scores. We examined 19 prespecified eye-movement features across six task-informed domains using covariate-adjusted ordinal-trend models. Greater depression severity was associated with lower gaze-allocation entropy (standardized \beta=-0.599 ), fewer inter-area switches ( \beta=-0.526 ), longer first target-entry latency ( \beta=0.572 ), and lower target dwell proportion ( \beta=-0.678 ; all false-discovery-rate-adjusted q0.001 ). Emotion-specific attentional-bias measures were not associated with severity, whereas associations involving fixation dispersion weakened after adjustment for anxiety symptoms. In a complementary machine-learning analysis incorporating 44 candidate eye-movement features and demographic covariates, a three-stage cascade classifier achieved the highest four-class stratification performance (accuracy, 0.665; macro-F1, 0.670; area under the receiver operating characteristic curve, 0.869), outperforming single-stage and two-stage approaches. Together, these findings identify altered visual exploration and memory-guided gaze control as candidate digital biomarkers of depression severity, rather than generalized negative attentional bias. These results support further prospective evaluation of task-evoked eye tracking as a complementary approach for timely severity assessment and clinical decision support.
[AI-531] STAMP as a topology-aware molecular language for frag ment-level learning and editing
链接: https://arxiv.org/abs/2610.18438
作者: Yuxuan Ren,Jiatong Li,Ziqiao Meng,Yatao Bian
类目: Quantitative Methods (q-bio.QM); Artificial Intelligence (cs.AI)
备注:
Abstract:Molecular fragments provide chemically meaningful units, but existing molecular languages rarely make their locations and connections explicit for editing. Here we introduce STAMP (Sequential Tree Attachment-based Motif-Preserving representation), a topology-aware molecular language that connects fragment-level representation learning with precise structural editing. STAMP separately encodes motif identity, attachment sites and tree topology, making fragments reusable across contexts and subtrees directly accessible. Learned representations capture motif similarity and recurring transformations, and lower activity-cliff prediction error on 23 of 30 targets versus a matched SMILES encoder. Compatible subtree operations support scaffold-preserving optimization, ranking first among six methods on 15 of 23 benchmark objectives. A deterministic compiler enables frozen language models to act on indexed fragments, improving molecular editing, optimization, transformation transfer and change identification by 9 to 56 percentage points over plain SMILES. Together, these results establish an executable molecular language in which learned chemical units remain directly operable during design.
[AI-532] How Well Do LLM s Perform in Peer Review? Evidence from Finance and Accounting
链接: https://arxiv.org/abs/2610.18435
作者: Hai Lu,Yichen Luo,Luping Yu
类目: General Finance (q-fin.GN); Artificial Intelligence (cs.AI)
备注:
Abstract:Referees increasingly use large language models (LLMs), while finance and accounting journals disagree about whether to allow it. We ask which part of a referee’s job the models can do. GPT-4o, whose training data end before the 2024 conference season, referees the papers presented at the 2024 AFA and NBER finance meetings, and we compare its verdicts with publication in the top-three finance journals over the next two years. Papers the model scores one standard deviation higher are 44% more likely to be published, net of author reputation. The result is not driven by look-ahead bias, as the model does not recall the outcomes. The information comes from its judgements of correctness and exposition. Its judgements of originality and significance carry none, and a referee guide that stresses importance does not help. The split between correctness and contribution holds for accounting papers and for a newer model, GPT-5. LLMs can check whether research is right, but not whether it matters. The technical check can go to the model, while the judgement of contribution stays with people.
[AI-533] Lifting Finite-Sample Gaussian Approximation of Wide Neural Networks to Functional Convergence
链接: https://arxiv.org/abs/2610.18166
作者: Luis Wiederhold,Nil Ayday,Debarghya Ghoshdastidar
类目: Machine Learning (stat.ML); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 19 pages, 1 table, 1 figure, preprint
Abstract:Randomly initialised neural networks of width m converge to Gaussian processes as m \to \infty , with two types of quantitative bounds characterising this convergence. Finite-sample bounds compare the network with its Gaussian limit at a fixed finite sample set, on an arbitrary domain, in the quadratic Wasserstein distance \mathcal W_2 , achieving a convergence rate of m^-1/2 with explicit constants. However, these constants grow with the number of samples, so the bound becomes uninformative as the sample set is refined. Functional bounds remove that dependence by comparing the two as random functions, at the price of spherical input, metrics weaker than \mathcal W_2 , or slower rates. In this paper, we therefore lift a finite-sample bound to a functional one. The lifting refines the sample set until it is dense in the domain, so the bound must first be made independent of the number of samples. We achieve this by measuring the error per sample, which makes the constant uniform. For a polynomial activation and any compact input set, this gives a rate of m^-1/2 in the supremum norm, with an explicit constant.
[AI-534] High-fidelity Three-Dimensional Flow Field Reconstruction Driven by Turbulent Fluid Dynamics
链接: https://arxiv.org/abs/2610.18045
作者: Hanlin Kong,Jianan Wei,Yuntian Chen,Yujia Wang,Wenguan Wang
类目: Fluid Dynamics (physics.flu-dyn); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 66 pages, 9 main-text figures and 1 main-text table; includes Supplementary Information with 14 figures and 22 tables
Abstract:High-fidelity three-dimensional turbulent flow fields are essential for scientific and engineering analyses, yet obtaining them through direct numerical simulation or high-resolution measurements remains expensive. Super-resolution reconstruction from coarse-grained data therefore offers a promising alternative, but remains fundamentally challenging because fine-scale structures are intrinsically three-dimensional and dynamically coupled across scales. Existing data-driven approaches still struggle to recover multiscale volumetric structures and their coherent temporal evolution. To address these limitations, we propose a turbulence-aware framework for time-resolved three-dimensional flow reconstruction. The framework couples a multiscale neural operator, designed to recover volumetric structures, with a state space model to promote temporal consistency in the reconstructed dynamics. In addition, Navier-Stokes residuals are incorporated as soft constraints to regularize the reconstruction toward physically plausible dynamics. A mixture-of-experts mechanism is also introduced to support unsupervised adaptation to unknown coarse-graining. Across evaluations on canonical turbulent systems and atmospheric reanalysis data, the proposed framework reduces reconstruction errors relative to learning-based and classical baselines, while better preserving multiscale spatial structure and coherent temporal evolution. These results validate the effectiveness of the proposed framework in reconstructing high-fidelity three-dimensional flow fields from coarse-grained data.
[AI-535] Can Machine Learning Potentials Be More Accurate Than the Methods They Are Trained On?
链接: https://arxiv.org/abs/2610.18027
作者: Pavlo O. Dral,Yi-Fan Hou,Yuxinxin Chen,Hassan Nawaz,Mateusz Witkowski,Szymon Śmiga,Ireneusz Grabowski
类目: Chemical Physics (physics.chem-ph); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
备注:
Abstract:Although machine learning potentials (MLPs) generalize across configurational and chemical compound spaces, their accuracy is commonly believed to be limited by their training methods, with improvements beyond this limit achieved so far only by refitting to experimental observables. In this Perspective, we argue that MLPs can also be made more accurate than their training methods non-empirically by generalizing across a third space, that of quantum-chemical methods. We propose routes to method-transferable MLPs that predict at unseen methods, with working implementations in both density-functional and wavefunction-based theories showing that they can be more accurate than any of their training methods. Aiming beyond the accuracy of the training methods could also make universal MLPs more accurate than single-method models, even when the method-transferable MLPs remain less accurate than those QM methods. Method-transferable MLPs will shape the construction of training datasets and the evaluation of models, pushing the accuracy limits of chemical simulations.
[AI-536] A Tutorial on Quantum Amplitude Embedding with Data Re-Uploading
链接: https://arxiv.org/abs/2610.17836
作者: Hyojun Ahn,Joongheon Kim
类目: Quantum Physics (quant-ph); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 6 pages, 6 figures, 3 tables. Submitted to ICEIC 2027
Abstract:Amplitude embedding lets a variational quantum model read many classical features with few qubits, but the model output is then only a quadratic form in the normalized input. Data re-uploading, which encodes the input repeatedly between trainable blocks, can be used to enrich such a model. Because every additional upload of an amplitude embedding repeats a costly state preparation, the gain that an upload buys has to be known before the gates are spent. This tutorial thus addresses the problem of explaining what data re-uploading adds to an amplitude-embedded model and how large the gain is. We express sequential re-uploading and parallel re-uploading with two copies in one notation and show that each upload raises the reachable polynomial order by two. Under one trainable circuit family and one budget of trainable parameters, simulation results show that re-uploading raises the test coefficient of determination from 0.689 to between 0.883 and 0.975 on targets of order four and from 0.337 to between 0.642 and 0.783 on targets of order six, on all ten seeds. On a quadratic target and on handwritten digits, one upload remains the best choice, and every additional upload lowers the gate error that the classifier tolerates. These results help designers match the number of uploads to the polynomial order of the task at hand.
[AI-537] Learning from fracture: Teaching language models to design through physical experience
链接: https://arxiv.org/abs/2610.17804
作者: Bo Ni
类目: Materials Science (cond-mat.mtrl-sci); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 21 pages, 7 figures, 3 tables. Includes Supplementary Materials
Abstract:How can language models learn to optimize physical designs through experience? We introduce FracLearn, an experience-abstraction-generalization framework evaluated through a deliberately unfamiliar, Tetris-inspired fracture game. The model first proposes architectures and examines simulated fracture responses to accumulate experience. Abstraction organizes these observations into conditional principles linking reinforcement placement, crack evolution and load transfer, with supporting evidence and failure conditions. Generalization challenges the model to combine these principles in new architectures and surpass the earlier design record. Campaigns demonstrate progressive refinement and recovery through hypothesis revision, yielding a best composite score 11.0% above the experience-stage record. Fracture analyses connect improved performance to crack deflection and temporary trapping as toughening mechanisms, alongside greater residual load capacity. Learning occurs through external knowledge and decisions, without model-weight updates. This approach offers a route to scientific agents that turn computational experiments into reusable physical knowledge for materials discovery and engineering design.
[AI-538] Convex-Analytic View of Query Selection for Safety Verification of Markov Decision Processes
链接: https://arxiv.org/abs/2610.17541
作者: Han Bao,Shinsaku Sakaue,Kazuki Watanabe
类目: Optimization and Control (math.OC); Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO)
备注:
Abstract:Safety verification of a Markov decision process (MDP) asks whether the probabilities of reaching several failure states jointly lie in a convex safe region under every scheduler. Watanabe and Unno (2026) prove safety by refining an outer approximation of the failure probabilities, where each query bounds a nonnegative weighted sum of the probabilities. Their rule selects weights by the largest Euclidean violation. We present a unified framework to measure violation in an explicitly chosen error unit: directions are scored by the support-function gap between the outer approximation and the safe region, normalized by the support function of the unit. We show that the maximal score, when positive, equals the smallest enlargement of the safe region in multiples of that unit that contains the outer approximation. The Euclidean unit recovers the earlier rule; two new rules are generated by the unit cube and the safe region itself. We experimentally evaluate how the choice of measurement rule affects the performance of safety verification for MDPs.
[AI-539] Event-Centric Earthquake Monitoring with a Multi-Station Transformer
链接: https://arxiv.org/abs/2610.17404
作者: Haoyu Wang,Weiqiang Zhu
类目: Geophysics (physics.geo-ph); Artificial Intelligence (cs.AI)
备注: 17 pages and 7 figures in main body; 16 figures and 4 tables in supplementary
Abstract:Earthquake catalogs underpin studies such as fault structure, earthquake processes and seismic hazard, and their value depends on being both complete and reliable. Machine learning has improved individual tasks of earthquake monitoring, but processing stations independently discards the consistency of arrivals across the network, and running tasks in sequence fragments the information describing each earthquake. We present EQNet2, a geometry-aware multi-station Transformer that jointly predicts phase arrivals, first-motion polarities, source location and magnitude, and the association of arrivals into events. On a Northern California test dataset and on continuous waveforms from Ridgecrest and the San Francisco Bay Area, EQNet2 recovers more weak arrivals, produces fewer false detections, and associates more picks into located events. Its learned attention links information across stations along time–distance patterns consistent with seismic moveout. As seismic networks grow denser, exploiting multi-station coherence and event-centric prediction offers a path to more complete and reliable catalogs.
[AI-540] What the Machine Invented: Mining New Mathematical Instruments from OpenAI s October 2026 Release of AI-Generated Mathematics
链接: https://arxiv.org/abs/2610.17090
作者: Ying Xie
类目: History and Overview (math.HO); Artificial Intelligence (cs.AI)
备注:
Abstract:On 6 October 2026 OpenAI published an initial 722 AI-written mathematical manuscripts, organized into 372 problem families, in the public repository this http URL. The release includes claimed solutions to many prominent open problems. We approach this collection as a treasure mine of mathematical ideas. Our aim is to extract new lemmas, constructions, inequalities and proof mechanisms introduced in the source arguments, and to distinguish them from established tools and new combinations of those tools. We call these mathematical instruments. Extraction comes first; their validity, novelty and possible reuse then become subjects for further study. The resulting descriptions also help readers understand the selected papers without having to reconstruct every argument at the outset. We examine 38 families in two groups of 17 and 21, spanning number theory, geometry, analysis, combinatorics, complexity, algebra, operator algebras, topology and mathematical physics. Each entry records the claim, extracted instruments, antecedents and caveats. Eleven cross-cutting patterns connect the entries, and a scoreboard records formalization status and dependencies. Novelty assessments are provisional and sometimes mixed. Selected source comparisons correct mathematical statements, hypotheses and counts; they do not independently establish priority or certify the underlying proofs. No Lean development was compiled or kernel-checked for this survey. We also briefly describe a subsequent application of the signed restriction lemma from family 112 to circuit lower bounds at every fixed depth. Subjects: History and Overview (math.HO); Artificial Intelligence (cs.AI) Cite as: arXiv:2610.17090 [math.HO] (or arXiv:2610.17090v1 [math.HO] for this version) https://doi.org/10.48550/arXiv.2610.17090 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Related DOI: https://doi.org/10.5281/zenodo.23319145 Focus to learn more DOI(s) linking to related resources
[AI-541] RINO: Renormalization Group Distillation with No Labels
链接: https://arxiv.org/abs/2610.15919
作者: Zichun Hao,Raghav Kansal,Abhijith Gandrakota,Chang Sun,Haoyang Li,Zihan Zhao,Ngadiuba Jennifer,Javier Duarte,Maria Spiropulu
类目: High Energy Physics - Experiment (hep-ex); Artificial Intelligence (cs.AI)
备注:
Abstract:Supervised machine learning in high-energy physics (HEP) requires training on labeled simulations of particle collisions, which are known to be unfaithful to the underlying physics. This leads to significant inaccuracies and biases in the transfer of trained models to real data, limiting the power of deep learning methods. Self-supervised learning (SSL) is a promising alternative for pretraining directly on real, unlabeled data. However, existing techniques largely operate with inductive biases inspired by domains such as images and natural language. We propose RINO (Renormalization Group DIstillation with NO Labels), which, for the first time, exploits the fundamentally multi-scale renormalization group structure of high-energy jet formation as a physically grounded source of SSL views. We define teacher-student representations across energy scales using different levels of hierarchical clustering, aligned via DINO-style self-distillation. To mirror the background-dominated nature of real collider data, RINO pretrains exclusively on jets from quantum chromodynamics (QCD) processes and is evaluated on out-of-distribution (OOD) transfer to rare signal classes. In our controlled data-efficiency study, RINO matches fully supervised training on JetNet after finetuning on only 1% of the labeled data. With the full labeled dataset, RINO improves OOD classification accuracy on top-quark jets by 5.9% and, without retraining, improves AUC on the unseen Higgs boson decay channel H \to 4q by 4.1% over the strongest SSL baseline. Scale-structure diagnostics reveal that DINO-style self-distillation permits our backbone to retain scale-dependent structure. Ablation studies show that this accounts for an 8.6% OOD accuracy gap with respect to alternative SSL paradigms such as SimCLR. Code is available at this https URL.
[AI-542] Framework for an AI-Ready ARPES Community Database
链接: https://arxiv.org/abs/2610.15236
作者: Zhi-Xun Shen,Jin Gene Wong,Karen Lei,Leon Garcia,Felix Baumberger,Sergey Borisenko,Aaron Bostwick,Ting Cao,Tanny Chavez,Yulin Chen,Monika Choudhary,Andrea Damascelli,Dan Dessau,Thomas P. Devereaux,Sandy A. Ekahana,Alexei Fedorov,Atsushi Fujimori,Yucheng Guo,M. Zahid Hasan,Makoto Hashimoto,Yu He,Philip Hofmann,Zahid Hussain,Harold Y. Hwang,Zhurun Ji,Chunjing Jia,Timur K. Kim,Phil D. C. King,Takeshi Kondo,Alessandra Lanzara,Chun Lin,Donghui Lu,Matteo Michiardi,Sung-Kwan Mo,Brian Moritz,Laurenz Rettig,Eli Rotenberg,Takafumi Sato,Kyle M. Shen,Jonathan A. Sobota,Inna Vishik,Shuolong Yang,Ming Yi,Timothy Zaklama
类目: Materials Science (cond-mat.mtrl-sci); Strongly Correlated Electrons (cond-mat.str-el); Superconductivity (cond-mat.supr-con); Artificial Intelligence (cs.AI)
备注: Preprint under review
Abstract:Condensed matter stands to gain profoundly from widespread adoption of artificial intelligence (AI). However, its current means of data organization impedes this endeavor. As a premier measure of electronic structure underlying material properties, Angle-Resolved Photoemission Spectroscopy (ARPES) has accumulated over four decades of spectra, yet the community’s publication-centered workflow preserves only a highly selected fraction of its experimental record. A shared data ecosystem could transform these fragmented measurements and their contexts into a persistent, collective scientific memory, creating immediate value for reproducibility and reuse while laying the foundation for AI-enabled discovery. In this Perspective, we report outcomes of a recent community workshop and propose a practical path toward such an ecosystem. Its benefits unfold along three timescales. In the near term, standardized, searchable, and citable datasets strengthen analysis pipelines independently of AI. In the intermediate future, community-validated tools operating across laboratories could enable broader reuse and increasingly closed-loop automated experiments. In the long term, a diverse corpus of measured electronic spectra could support foundation models for quantum matter. Looking forward, a diverse corpus of measured electronic spectra could support foundation models for quantum matter and new discoveries. We discuss lessons from existing scientific databases, the current state of AI in ARPES, as well as areas of community convergence on data organization, access, credit, governance, and a staged pilot program.
[AI-543] oo Late to Prune: Generative AI the Screen and Team Creative Decisions
链接: https://arxiv.org/abs/2610.14984
作者: Difang Huang,Shumin Yu,Fadong Chen,Weihua Zhou
类目: General Economics (econ.GN); Artificial Intelligence (cs.AI)
备注:
Abstract:Generative AI makes options cheap and shifts the bottleneck of creative work from generating options to screening them. We argue that AI’s value depends on whether its options arrive before or after the team’s screen. In our model, early options compete for a place in the portfolio; late options arrive after that competition and are kept unscreened. In a preregistered experiment, we randomly assigned 148 three-person teams to five schedules of access to the same language model during a product-design task with timed search and selection stages. Moving the same 15 minutes of access from search to selection lowered expert-rated creativity by 0.62 standard deviations, although teams used the model at similar rates. Platform logs locate the difference at the screen. Early access enlarged the pool of ideas fivefold, and teams pruned it. Late-access teams reached the screen with the same small pool as teams without AI and then searched again, devoting 29% of their selection-stage queries to generating new options, compared with 11% for teams with access throughout. Late access added breadth, not quality: these teams submitted 73% more proposals than did teams without AI. Language-model judges detected the benefit of access but not the cost of late timing.
[AI-544] Cross-Lingual TTS for Indian Languages: A Benchmark and Inference-Time Duration Estimation Strategies AACL
链接: https://arxiv.org/abs/2610.14826
作者: Adarsh Arigala,Arjun Gangwar,Srinivasan Umesh
类目: Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI); Sound (cs.SD)
备注: 9 pages, 3 figures, 5 tables. Accepted at the AACL-IJCNLP 2026 Workshop Multi-LLL. The dataset and code recipes are available at: this https URL
Abstract:Recent alignment-free Text-to-Speech (TTS) models achieve high synthesis quality by removing explicit alignment modules and simplifying training. However, they often struggle in cross-lingual scenarios due to differences in character durations across languages, leading to duration mismatch and degraded speech quality. Existing approaches typically address this issue using explicit duration predictors, increasing model complexity. Systematic evaluation of TTS performance for Indian languages in cross-lingual settings remains under-explored due to the lack of standardized evaluation benchmarks. In this work, we construct a cross-lingual evaluation benchmark covering English and nine Indian languages and propose a language-aware inference-time duration estimation strategy for alignment-free TTS that incorporates language-specific duration-per character statistics without retraining or architectural modifications. We further develop heuristic formulations that refine duration estimation based on language-dependent duration variations. Experiments on bilingual and multilingual A-DMA models demonstrate improved speech naturalness, intelligibility, and speaker similarity while reducing duration mismatch.
[AI-545] Depth-progressive 3D seismic velocity model building via 2D generative diffusion models
链接: https://arxiv.org/abs/2610.14812
作者: Shijun Cheng,Randy Harsuko,Mohammad H. Taufik,Tariq Alkhalifah
类目: Geophysics (physics.geo-ph); Artificial Intelligence (cs.AI)
备注:
Abstract:Three-dimensional (3D) velocity model building (VMB) remains computationally challenging with conventional approaches and native 3D neural networks, largely due to the prohibitive memory and training cost of full 3D formulations. Therefore, we introduce a depth-progressive framework that achieves 3D VMB using 2D generative diffusion models through a depth-as-channel formulation, in which the depth dimension of each 3D velocity patch is mapped to the channel dimension of a 2D U-Net. This allows the model to synthesize overlapping depth patches sequentially from shallow to deep while conditioning the prediction of the deeper patch on the shallower one, as well as on any available geophysical constraints, including well logs and structural attributes. During training, paired shallow-deep patch pairs with randomized overlaps are used to teach the network how deeper velocity structures relate to shallower information, and a classifier-free guidance strategy enables the model to handle missing constraints gracefully. At inference, starting from a shallow velocity prior, the model recursively generates deeper patches in a top-down manner, and a depth-attenuated Gaussian blending scheme merges overlapping predictions into spatially coherent 3D volumes while accounting for the progressive accumulation of uncertainty at greater depths. Numerical experiments on both in-distribution benchmark models, including Overthrust, SEAM Arid, and SEG/EAGE, and the out-of-distribution Marmousi model demonstrate that the method effectively propagates shallow velocity priors downward and produces high-quality 3D velocity volumes across a range of geological settings and conditioning scenarios. Compared to a 3D benchmark of equivalent architecture, the proposed method significantly reduces parameter count, training cost, and inference cost, while maintaining superior VMB accuracy.
[AI-546] he Future of AI-assisted Physics: Discovering Verifying and Organizing Results via Formalization
链接: https://arxiv.org/abs/2610.14059
作者: Sven Krippendorf,Leopoldo Sarra,Joseph Tooby-Smith
类目: High Energy Physics - Phenomenology (hep-ph); Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO); High Energy Physics - Theory (hep-th)
备注: 19 pages, 2 figures
Abstract:Rapid advances in AI have opened a new era for physics, enabling unprecedented capabilities in research and discovery. However, these advances also bring challenges in verifying, trusting, and reusing the results they so rapidly produce. Formalization is a necessary part of the solution to these challenges. In this paper, as well as giving an overview of the role of AI and formalization in physics, we make the case that: (1) formalization is an essential layer for AI-driven discoveries in both computational and theoretical physics; (2) AI is lowering the barrier to, and increasing the efficiency of, formalization in physics, though it remains weaker at choosing definitions and designing libraries than at writing proofs; (3) the speed AI brings to formalization in physics currently comes at the cost of trust, credit, and community; (4) shared, community-reviewed formal libraries such as Physlib should be central in making progress in this field, both as a trusted, reusable foundation for physics results and as the environment in which AI systems can reason about physics; (5) formalizing physics raises challenges beyond those in mathematics, most importantly that a checked proof does not establish that its definitions faithfully capture the intended physics. We also give our perspectives on near-term and long-term directions in these areas.
机器学习
[LG-0] Exact Topological Compliance: Generating Persistence-Equivalent Graphs NEURIPS2026
链接: https://arxiv.org/abs/2610.18856
作者: Mattie Ji,Indradyumna Roy,Vikas Garg
类目: Machine Learning (cs.LG); Algebraic Topology (math.AT)
*备注: Accepted by NeurIPS 2026
Abstract:Persistent homology summarizes the topological evolution of a graph as persistence diagrams (PDs). Generating graphs whose topology matches a target descriptor is increasingly important, yet existing topology-aware generative models rely on soft regularization, producing graphs whose PDs may only be approximately similar. We initiate a principled investigation of the generation problem: produce graphs that realize a target PD exactly. We propose two approaches guided by the implicit local and global constraints a target PD imposes. The local approach builds degree-based PD-equivalent graphs iteratively, and we prove that every output realizes the target PD and that every degree-based PD-equivalent graph is reachable. However, as shown thereafter, local generation necessarily trades off between sample diversity and reaching infeasible states. We therefore recast generation as a global constraint satisfaction problem with a complete encoding solvable by modern constraint programming solvers. This is applicable to any vertex-based permutation equivariant filtration scheme and is minimal under the degree filtration. Experiments on four standard graph generation benchmarks confirm that both methods realize the target PD exactly. Together, our methods provide a principled treatment of strict topology-compliant graph generation. Code is available at this https URL.
[LG-1] Statistical Vignettes on the Grokking Phenomenon
链接: https://arxiv.org/abs/2610.18811
作者: Baye Malick Gning,Elvis Dohmatob
类目: Machine Learning (cs.LG)
*备注:
Abstract:In this work, we study grokking in a simple, analytically tractable setting whose rich phenomenology allows us to disentangle the effects of initialization, signal strength, weight decay, sample size, and input dimension. These factors can produce superficially similar learning curves through distinct mechanisms, complicating their interpretation. Our analysis distinguishes early fitting through nuisance features, core feature learning, and delayed generalization. We identify sharp conditions under which grokking provably occurs and derive analytic asymptotic expressions for the lengths of the generalization plateaus. Our results also characterize regimes of early generalization and failure to reach a prescribed test accuracy target. We show that the interpolation edge constitutes a distinct dynamical regime, where vanishing spectral gaps separate core learning from generalization, helping clarify the role of proximity to interpolation in existing accounts of grokking. The resulting picture is analogous to fast and slow phases in chemical kinetics and suggests that the lazy-to-rich transition discussed in the literature may reflect a similar separation of dynamical time scales. Our theoretical results are faithfully supported by extensive experiments.
[LG-2] Depth Memory
链接: https://arxiv.org/abs/2610.18809
作者: David González-Martínez,Philipp Nazari,T. Konstantin Rusch,Shiwei Liu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Residual connections propagate information through a network by additively accumulating layer outputs. Recently proposed alternatives, such as Manifold-Constrained Hyper-Connections (mHC) and Attention Residuals (AttnRes), instead use more expressive depth aggregation mechanisms that improve language modeling performance at scale. In this work, we consider the mixing of representations over depth as a problem of storing and retrieving information across layers. Formalizing this perspective, we introduce Depth Memory (DM), a unified framework that describes the standard pre-layer-norm (PreLN) transformer, mHC, and AttnRes in terms of streams that carry representations through the network. One such stream, for example, could be an explicit memory stream that is queried to enrich the input to every layer, and whose output is subsequently written back into the memory for use by later layers. For AttnRes, this memory stream is a growing cache of explicit layer representations. For mHC, it is an expanded residual stream. Building on this view, we propose Linear Depth Memory (LDM), which formulates information retrieval across depth as a regression problem solved using recursive least squares. The resulting architecture propagates three streams across depth: the standard residual stream, a linear associative memory stream, and an inverse covariance stream. For high-throughput training and inference, we propose a hardware-aware implementation of LDM that reduces peak activation memory and memory traffic. LDM consistently achieves the lowest perplexities among PreLN, mHC and AttnRes baselines at the 400M and 1.5B parameter scales, alongside competitive zero-shot accuracy.
[LG-3] Enabling Token-wise Importance Sampling in Off-Policy RLVR with Proximal KL Regularization
链接: https://arxiv.org/abs/2610.18738
作者: Hao Bai,Jiarui Yao,Tong Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Off-policy data in reinforcement learning with verifiable rewards (RLVR) can arise from asynchronous generation, rollout-training mismatch, and response reuse. Token-wise importance weighting avoids multiplying likelihood ratios across a response but generally biases policy gradients. We prove that, under coverage and local expressiveness conditions, full future KL credit makes the exact proximal optimizer the update’s unique stationary policy, even for mixtures of older policies, without full-response weights or a fitted critic. We further establish convergence of continuous-time tabular population updates under persistent coverage and time-varying behavior policies, and a finite-sample reward guarantee for linear logistic models under additional assumptions. These results motivate Proximal KL, a practical off-policy RLVR method that separates the proximal reference from the sampling policy and replaces PPO clipping with bounded future KL credit. Across all seven matched off-policy comparisons on Qwen3-1.7B and Qwen3-4B, spanning two learning rates, response reuse, and rollout staleness, Proximal KL outperforms GRPO in micro-averaged accuracy over eight math benchmarks by up to 2.58%. A separate on-policy comparison yields comparable aggregate accuracy. In periodic-refresh experiments, Proximal KL avoids the collapses observed in PG and CISPO, exceeds GRPO’s training accuracy, and reaches higher accuracy in fewer policy updates. Ablations show that token-wise correction outperforms sequence-level and unweighted controls. With token-wise correction retained, future KL feedback reduces sensitivity to the tested baseline perturbation relative to PPO clipping. A finite future horizon improves AIME24 mean@16 and pass@16 over the full tail, while fixed-policy measurements show that truncation reduces correction variance over the tested horizons.
[LG-4] SP2ec: Adaptive Self-Speculative Decoding for Vision-Language Models
链接: https://arxiv.org/abs/2610.18735
作者: Yuqi Huang,Xingyao Li,Yunlong Hou,Fengzhuo Zhang,Jiachun Pan,Vincent Y. F. Tan
类目: Machine Learning (cs.LG)
*备注:
Abstract:Speculative Decoding (SD) has demonstrated its efficacy in accelerating large language model inference, but for Vision-Language Models (VLMs), the diverse input modalities make it difficult for a small trained drafter to achieve consistently high throughput. Self-speculative decoding reuses parts of the target model as a drafter which retains the multi-modal capability of the VLMs, yet its efficiency depends on choosing a suitable draft configuration for each prompt. We propose Single-Peak SPeculative decoding (SP ^2 ec), a training-free bandit algorithm that adapts the draft configuration during generation. SP ^2 ec exploits the single-peak throughput pattern observed in existing self-speculative methods, including SWIFT, KnapSpec, SpecVLM, and our simple cosSim. Our algorithmic design explicitly minimizes a novel notion of wall-time regret for each given query, corresponding to the empirical wall-time latency. Theoretical guarantees on the wall-time regret show that SP ^2 ec can exploit the optimal draft configuration, with negligible exploration cost. Extensive experiments on Qwen2.5-VL-7B-Instruct and LLaVA-NeXT-Mistral-7B across visual and textual tasks show consistent improvements of up to 11% over the original self-speculative methods, demonstrating the wide applicability of the proposed SP ^2 ec algorithm.
[LG-5] Learning When to Transfer: Reinforcement Learning for Carbon-Footprint Modeling in New Manufacturing Domains
链接: https://arxiv.org/abs/2610.18730
作者: Muhammad Umar Farooq,Daniel R. Cooper
类目: Machine Learning (cs.LG)
*备注:
Abstract:Manufacturers often need to characterize a new material, process, or product before enough operating data are available to train reliable predictive models for this new, target domain. Existing manufacturing operations contain potentially useful knowledge to accelerate this characterization, but indiscriminate reuse of data from existing source domains can cause negative transfer in learning of the new, target domain when the manufacturing domains differ sufficiently. This work proposes a reinforcement-learning method for selective knowledge transfer, where its reward scores each opportunity to transfer against supervised alternatives, so negative transfer is avoided and controlled during forward prediction instead of being diagnosed afterwards. Given limited data from a new target domain, the learned policy chooses among predictors based on individual source operations, subsets of the sources, or the supervised alternatives: fully pooled sources, or target data alone. A validation gate deploys the transferred prediction only when it improves upon the supervised alternatives. The method is designed for learning variables supporting carbon accounting, as well as performance and quality outcomes. We demonstrate the approach by quantifying the embodied greenhouse-gas emissions of parts produced via material-extrusion additive manufacturing, using data from 729 experiments spanning five materials: ABS, PETG, TPU, PLA, and recycled PLA. The response combines measured fabrication electricity with polymer-specific cradle-to-gate production factors, while 22 material, process, and toolpath features characterize each experiment. In leave-one-material-out evaluation, this method achieves a mean test RMSE of 5.2 g CO2e per part, 20.00% lower than target-only training (32.5% lower when only 5-10% of the target training data are available) and 10.30% lower than indiscriminate source-target pooling.
[LG-6] Learning Source Structure: Bayesian Prediction with Transformers
链接: https://arxiv.org/abs/2610.18725
作者: Meir Feder,Ruediger Urbanke
类目: Machine Learning (cs.LG)
*备注: 45 pages, 12 figures. 9 pages of main text. Code at this https URL
Abstract:A transformer trained on sequences whose source parameters are drawn afresh each time can learn only what the sequences share: the source family and its prior. The Bayesian predictor that knows both, which we call informed Bayes, is the unique minimizer of the training loss and is computable from the family alone; we use it as a quantitative, falsifiable reference. Where earlier work compared transformers with Bayes on sources whose contexts can be learned independently, we build Markov sources with up to 64 symbols in which observations in one context inform prediction in another: product sources with known or hidden encodings, transition rows sharing an unknown base law and depth, and a mixture of three families. Our main result is that informed Bayes predicts, from the family and prior alone, how a trained transformer behaves in four tests that go beyond the learning curve: the loss that remains under a source outside the training family (predicted 2.237 bits, measured 2.309), the doubling of the extra loss when two independent components change rather than one (0.345 predicted, 0.350 measured), the benefit of other contexts when a token appears for the first time, and the depth at which copying an earlier continuation starts to pay. On the learning curves themselves, small transformers match informed Bayes when the shared structure is visible in the tokens, and hidden structure needs more layers and training, with gaps below 0.01 bits per token for the largest models. We explain why the imitation is possible and where it must fail: informed Bayes decomposes into sufficient statistics that attention selects and accumulates and an estimator, fixed by the prior, that the feed-forward network learns; training pays the complexity of informed Bayes under the architecture’s prior, and finite-memory universal prediction bounds how closely a bounded network can reach it.
[LG-7] On Collusion-Free Teaching in Planar Graphs
链接: https://arxiv.org/abs/2610.18714
作者: Khazaliya Liana,Kuske Clemens
类目: Discrete Mathematics (cs.DM); Computational Complexity (cs.CC); Machine Learning (cs.LG)
*备注:
Abstract:Non-clashing teaching minimizes teaching complexity among models satisfying the Goldman–Mathias collusion-avoidance criterion. While the Vapnik–Chervonenkis dimension (VCdim) measures the complexity of learning from samples, non-clashing teaching dimension (NCTD) measures the number of examples required when a teacher can select them. Whether NCTD is always bounded above by VCdim remains a fundamental open question in learning theory. In this paper, we investigate this question for the concept classes of closed neighborhoods in planar graphs. We prove a tight upper bound of two on their NCTD, thereby establishing the conjectured inequality for this family. For the positive variant NCTD+, in which teaching sets contain only positive examples, we obtain a tight upper bound of four. Finally, we show that deciding whether either NCTD+ or NCTD is at most one is NP-complete, even for concept classes of closed neighborhoods in planar graphs. In particular, despite the upper bound of two on NCTD for such classes, distinguishing between NCTD values of one and two remains NP-hard. Subjects: Discrete Mathematics (cs.DM); Computational Complexity (cs.CC); Machine Learning (cs.LG) Cite as: arXiv:2610.18714 [cs.DM] (or arXiv:2610.18714v1 [cs.DM] for this version) https://doi.org/10.48550/arXiv.2610.18714 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-8] A Refresh Interval Does Not Fully Specify a Refresh Schedule: Learning-Rate-Dependent Placement in Kronecker-Preconditioned Language Model Training
链接: https://arxiv.org/abs/2610.18695
作者: Xinyu Wang(1),Zhiwei Zhai(2),Yu Gu(1),Zijun Yu(1),Sicheng Lyu(1),Jingrui Tian(1),Lei Ding(3),Ziyang Song(4),Xiao-Wen Chang(1) ((1) McGill University, (2) The Chinese University of Hong Kong, (3) University of Manitoba, (4) Ohio University)
类目: Machine Learning (cs.LG)
*备注: 18 pages, including references and appendices
Abstract:Kronecker-preconditioned optimizers recompute a cached curvature eigendecomposition only periodically, and implementations and adaptive methods control this through the refresh frequency. We show that frequency leaves part of a schedule free: at a fixed number of refreshes, when they occur changes held-out loss at the same compute. Training GPT-style models with the Mousse optimizer across model sizes, we move the middle refreshes of a four-refresh schedule while fixing the first and last. On FineWeb the two placements exchanged rank as the learning rate rose (later placement was better at low rates, earlier placement at high ones), so a single-learning-rate comparison can report either ranking. The reversal held under controls including the published optimizer internals, and placement still mattered at the published refresh density. The learning rate at which the preference flips fell with model size, consistent with a held-out smaller model, and shifted with training length. A spectral-width staleness score predicted the same ranking at every learning rate and missed the reversal. Refresh schedules should be reported in full and compared across the learning rates at which they will be used.
[LG-9] BridgeTwist: Twisting Schrödinger Bridges for Training-Free Conditional Sampling NEURIPS2026
链接: https://arxiv.org/abs/2610.18658
作者: Mengyu Li,Qianqian Qu,Jun S. Liu
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: Accepted at NeurIPS 2026
Abstract:Generative models are often required to serve many downstream conditional tasks. In the training-free setting, a pretrained unconditional model is reused as a prior and combined with a task-specific likelihood to sample from the corresponding posterior. While such training-free samplers have been developed for diffusion and flow models, their counterpart for Schrödinger bridge (SB) models remains unexplored, despite the growing appeal of SBs as flexible stochastic transport models between general source and target distributions. We propose BridgeTwist, a training-free conditional sampler for pretrained SB generative models, formulated as a tempered twisted sequential Monte Carlo scheme. Its twist is a closed-form plug-in surrogate, obtained by combining the velocity and score fields exposed by SB pretraining into an endpoint predictor and evaluating the likelihood there. Since this surrogate is less reliable near the source, we temper the twist with a time-varying exponent, downweighting its influence at early steps. We further introduce a bridge-shaped inference-time diffusion schedule that concentrates the proposal noise in the interior of the path. Telescoping importance weights then cancel all intermediate twist factors, yielding an asymptotically exact sampler for the posterior. Empirically, BridgeTwist outperforms training-free baselines on class-conditional and text-prompted image generation, and is competitive with specialized inverse problem solvers on inpainting, deblurring, and super-resolution.
[LG-10] MultiAlloc: Budget-Aware Precision Allocation for KV Cache Quantization
链接: https://arxiv.org/abs/2610.18642
作者: Yangchengyu Zhou,Zhiping Wu,Runchang Jiang,Dongdong Ren,Youfang Han,Wei Zhao,Dongqi Cai,Wenbin Li
类目: Machine Learning (cs.LG)
*备注: Code: this https URL
Abstract:The Key-Value (KV) cache constitutes a dominant source of memory consumption during large language model (LLM) inference, particularly in long-context and complex reasoning scenarios. Existing KV cache quantization methods have explored different compression granularities, including layer-wise precision adjustment and token-level information preservation. In practical deployment, however, the key challenge is not merely compression, but how to allocate limited quantization resources among heterogeneous precision preservation strategies under a fixed KV cache memory budget. We propose MultiAlloc, a profiling-guided precision allocation framework for KV cache quantization that adaptively distributes limited quantization resources between layer-wise precision enhancement and token-wise preservation. MultiAlloc employs lightweight offline profiling to characterize layer sensitivity and token importance, enabling adaptive precision allocation under a fixed memory constraint. Guided by the profiled memory–accuracy trade-off, it automatically derives an effective precision allocation strategy without online allocation search, and applies the resulting policy during inference. Extensive experiments on reasoning and long-context benchmarks demonstrate that MultiAlloc consistently improves over layer-wise, token-wise, and recent multi-granularity KV cache quantization methods under matched memory budgets across diverse models, including Mistral, Llama, DeepSeek-Qwen, and QwQ.
[LG-11] Deterministic Prediction as Global Velocity: Flow Matching for Probabilistic Time-Series Forecast
链接: https://arxiv.org/abs/2610.18621
作者: Jiajing Hu,Xu Han,Li-Ping Liu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Accurate uncertainty quantification is critical for many time series forecasting applications, and recent studies have increasingly adopted generative models to capture predictive uncertainty. Existing approaches often combine deterministic forecasting with generative modeling to improve predictive performance. However, their joint training is typically unstable because the deterministic model is embedded within the generative process, resulting in deep computational graphs and changing the probability transport path. In this work, we propose DV-Flow, a unified flow-matching formulation that incorporates the deterministic predictor as a parallel velocity component rather than as the endpoint of the transport path. This design enables stable end-to-end optimization while subsuming several existing formulations as special cases, allowing direct comparisons among them. Extensive experiments on six real-world datasets demonstrate that DV-Flow consistently improves probabilistic forecasting performance over strong baselines. Our code is available at this https URL.
[LG-12] Autonomous Cyber Operations Gym for Training and Evaluating AI Models for Defence
链接: https://arxiv.org/abs/2610.18594
作者: Tobias Hürten,Johannes F. Loevenich,Florian Spelter,Erik Adler,Laurin Holz,Maxime Schwarzer,Thies Moehlenhof,Roberto Rigolin F. Lopes
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注: 6 pages, 5 figures, 1 table. Accepted and presented at the 2026 IEEE International Conference on Cyber Security and Resilience
Abstract:This paper proposes an Autonomous Cyber Operations (ACO) gym, a unified framework integrating high-fidelity network emulation, realistic user behaviour modelling for benign network traffic and adversarial profiles, and the generation of cybersecurity scenarios through open toolkits. By supporting live and deferred feature extraction at the packet and flow levels, the gym facilitates the creation of labelled synthetic datasets for training Deep Neural Network (DNN)-based Intrusion Detection Systems (IDSs), as well as developing and evaluating modular components for autonomous defence agents through emulated interactions. While evaluation of a DNN classifier within the gym shows high in-domain performance, achieving near-perfect recall on adversarial network flows, our experiments reveal a significant sensitivity to domain shift, showing that even minor changes in network composition, such as variations in traffic protocol distribution, can lead to a catastrophic collapse in model transferability and detection performance.
[LG-13] Fairness under Prediction Error in Predict-then-Optimize Pipelines
链接: https://arxiv.org/abs/2610.18593
作者: Hilda Adwubi Osei
类目: Machine Learning (cs.LG); Computers and Society (cs.CY)
*备注: 11 pages, 1 figure
Abstract:Fairness constraints in predict-then-optimize pipelines are necessarily enforced on predicted outcomes \haty , since true outcomes y are unobserved at decision time. We study whether a demographic parity constraint enforced this way continues to hold against y , and how the resulting violation scales with prediction error. Using a budget-constrained allocation problem solved as a linear program, we show empirically that the true-outcome fairness gap grows with prediction error \sigma and saturates, well fit by c(1 - e^-\sigma/\tau) with c = 0.1545 \pm 0.0103 and \tau = 3.9045 \pm 0.3745 ; its mean is invariant to sample size while its variance decays at the standard O(1/\sqrtn) rate. We derive a two-part theoretical bound: a closed-form local sensitivity result showing the gap’s first-order response to prediction error depends on only two individuals per group, validated against the solver to within three significant figures; and a universal, error-independent ceiling following from the boundedness of the decision variable, confirmed consistent with but looser than the empirical saturation level. On US Census income data, enforcing demographic parity on a trained predictor’s output reduces an unconstrained disparity in true income to roughly one-third of its original size, leaving 34.4% \pm 3.3% (California) and 35.7% \pm 1.8% (Texas) of the disparity intact and undetectable from the predictions alone – consistent not only in mechanism but in approximate magnitude across two states with different demographics and income distributions, and stable across ten independent train/test splits in each. A fairness constraint that appears fully satisfied at decision time can mask a substantial, systematic violation of the outcome it was meant to protect.
[LG-14] Linear Last-Iterate Convergence Rates for Optimistic Learning in Harmonic Games
链接: https://arxiv.org/abs/2610.18560
作者: John Lazarsfeld,Anas Barakat,Jacob Abernethy,Vidya Muthukumar,Georgios Piliouras,Andre Wibisono,Juba Ziani
类目: Computer Science and Game Theory (cs.GT); Machine Learning (cs.LG)
*备注:
Abstract:We prove that the last iterate of Optimistic Follow-The-Regularized-Leader (OFTRL) converges at a linear rate to a Nash equilibrium in multiplayer harmonic games. Our result holds under a mild game-regularity condition and for a large family of regularizers, including entropy, Tsallis, and log-barrier. Our analysis is based on a dual perspective of OFTRL: we prove that the interaction between optimism and the global structure of harmonic games leads to the dissipation of an energy function along the dual iterates, for which we establish tight quantitative bounds. Our result also holds for the Extragradient FTRL variant.
[LG-15] Last-Iterate Convergence of Clipped SGD with Linear Decay
链接: https://arxiv.org/abs/2610.18555
作者: Wei Jiang,Haomin Bai,Zechao Li
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:In this paper, we study the last-iterate convergence of clipped stochastic gradient descent under heavy-tailed noise. We assume that the stochastic gradients are unbiased and that the noise has a bounded central p -th moment, where 1p\le2 . For smooth convex objectives, a linearly decreasing learning rate gives a high-probability error bound of order O(LR^2\log(1/\delta)/T+\sigma R(\log(1/\delta)/T)^(p-1)/p) . We further establish last-iterate guarantees for nonsmooth composite objectives and smooth composite objectives with proximal updates. In addition, when a causal adversary replaces at most q gradients, the same clipped update achieves a bound with \log(1/\delta) replaced by \log(1/\delta)+q . The corresponding lower bounds for the clean and corrupted oracle models are also provided.
[LG-16] Euclidean Distance-Based Flow Matching
链接: https://arxiv.org/abs/2610.18534
作者: Paul-Henrik Heilmann,Peter Ochs
类目: Machine Learning (cs.LG)
*备注:
Abstract:Flow Matching has recently become popular as a technique for generative modeling with applications in many domains. In the standard Flow Matching formulation, we interpret the vector field as the solution of a least-squares regression problem, which is known to be sensitive to contamination. Therefore we leverage the viewpoint of robust statistics and replace the squared Euclidean distance with the Euclidean distance as a regression objective. This seemingly minor change strongly affects the properties of Flow Matching. We provide a probabilistic interpretation of the new regression target as a conditional geometric median and use this characterization to study distributional robustness of the Euclidean Distance-based Flow Matching. We analyze the mathematical properties of this changed objective and the effect of a contaminated target distribution on the involved conditional velocity distributions. We discuss the challenge of directly characterizing the distribution that our method induces. In specific contamination settings we show theoretically that our method yields an improvement for severe contamination. Finally, we present numerical experiments conducted on the CIFAR-10 image dataset where a proportion of the samples has been contaminated. Our Euclidean distance-based Flow Matching shows a clear advantage in scenarios of severe contamination of the target distribution.
[LG-17] wo For The Price Of One: Curvature-Aware Pruning Preserves Influence Functions Too
链接: https://arxiv.org/abs/2610.18516
作者: Mehdi Makni,Natalia Ponomareva,Rahul Mazumder
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Post-training pruning enables efficient deployment, but conventional methods do not explicitly preserve internal-gradient alignment between sparse and dense models. We study novel connections between downstream performance of compressed models and the ability to retain gradient information via influence-function preservation. Central to this connection is a curvature-aware objective that accounts for both layer activations and backpropagated gradients. Despite its promise, this objective remains underexplored in pruning due to a lack of efficient solvers. We introduce \textttCACTUS , a novel ADMM-based optimizer supporting both pruning and sparse-plus-low-rank decompositions of pre-trained weights. Compared to LLM-Surgeon which targets the same objective, \textttCACTUS can achieve over an order-of-magnitude speedup. At 50% pruning of Qwen3-4B, \textttCACTUS reduces the C4 perplexity gap to the dense model by 45% , improves compressed-to-dense per-sample gradient cosine similarity on WikiText-2 from 0.670 to 0.755 , and runs over 28\times faster than LLM-Surgeon on substantially weaker hardware. By preserving influence functions, \textttCACTUS supports effective data selection. On Llama-3.2-3B, fine-tuning a fresh dense model on the 5% of data selected by \textttCACTUS achieves 49.44% accuracy on BIG-Bench Hard, compared with 47.22% using ALPS and 49.54% using dense-model selection. Finally, integrating \textttCACTUS with LoGRA, a recent gradient-projection strategy, on Qwen3-4B with MLP layers pruned to 2:4 sparsity and a rank-64 correction retains a 0.711 cosine alignment with dense LoGRA-projected C4 gradients.
[LG-18] Approximating Optimal Policies in Heterogeneous Asynchronous Multi-player Bandits
链接: https://arxiv.org/abs/2610.18514
作者: Sacha Arroues-Paykin,Etienne Boursier,Vianney Perchet
类目: Machine Learning (cs.LG)
*备注:
Abstract:Motivated by applications to cognitive radio networks, multi-player multi-armed bandits (MMAB) have attracted considerable attention in recent years. The asynchronous variant, in which players are active only at a fraction of timesteps, better reflects real-world decentralized networks, where devices are not permanently active. Despite its practical relevance, this setting remains relatively understudied. Existing work on asynchronous MMAB has focused on homogeneous players, who share the same activation probability, a setting in which the optimal centralized allocation can be easily characterized. In contrast, efficiently computing the optimal allocation in the heterogeneous setting, where players have different activation probabilities, has remained an open problem. In this work, we investigate this problem and first establish its NP-hardness. We then turn to the greedy algorithm, which runs in polynomial time, and prove that it achieves a \frac1e -approximation of the optimal allocation. Furthermore, when players are processed in decreasing order of their activation probabilities, the approximation guarantee improves to \frac1\sqrte . Overall, our results establish the greedy algorithm as an efficient approximation to the optimal allocation, providing a practical oracle baseline for the design of online algorithms for heterogeneous asynchronous multi-player bandits.
[LG-19] owards Safe Autonomy for Computer Use
链接: https://arxiv.org/abs/2610.18508
作者: Dominik Seip,Bartol Markovinović,Matthias Hein
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注:
Abstract:The safety of LLM-based agents currently rests on alignment training and safety classifiers, neither of which can rule out that a sophisticated indirect prompt injection (IPI) goes through and causes significant harm. Recent work therefore builds agents with information-flow guarantees that rely on structured tool interfaces to interact with the environment. Computer Use Agents (CUAs) act visually on the computer screen, which does not provide a boundary between trusted first-party interfaces and untrusted, potentially attacker-authored content. Actions are equally ambiguous: a click may land on an application toolbar or a button controlled by an attacker. Treating the user as the source of trust and confirming each sensitive decision produces confirmation fatigue, with the risk of missing harmful actions. We introduce SA-CUA, a dual-LLM framework that generates a reliable trust boundary for CUAs by deriving conservative deterministic rules from the accessibility tree for unmasking trusted regions from the privileged LLM’s screenshot, keeping trusted and untrusted content separated. Human interaction is only required for proper task specification in the beginning. During the task a robust endorsement channel allows values to pass from the untrusted to the trusted domain by verifying information via trusted channels. This allows the agent to autonomously handle tasks involving untrusted information. SA-CUA maintains much higher utility than CaMeL-CUA while keeping the attack success rate at zero.
[LG-20] Recalibrating Many Questions with Few Labels: Minimax Rates and Identification Limits
链接: https://arxiv.org/abs/2610.18488
作者: Xiaoyu Li,Chiwun Yang
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 69 pages, 7 figures, 1 table
Abstract:Many related prediction problems, which we call questions, come with few labels each. What information can be shared across them, and what is actually necessary for good decisions? We answer both for recalibration, motivated by decision models such as Jev (TypeSafe AI, 2026) that return a probability for any yes/no question a developer writes. A predictor reports a probability P with output law \mu on m questions, each with n labels and a nondecreasing map g_j§=\mathbbP(Y=1\mid P=p) ; the compound risk is the excess Brier score over the true maps, averaged over questions. Over maps within sup-distance \tau\le 1 and total variation V\le 2 of one nondecreasing shape (fixed effects), the minimax compound risk is (mn)^-2/3+\min\tau^2,1/n+\min\tau^2,V^2,(V/n)^2/3\ up to universal constants, for every non-atomic \mu . A pooled isotonic fit with a per-question correction sized once, on held-out labels of all questions, attains it without knowing \tau , V or \mu ( n\ge 2 , m\ge 3 ); tuning each question on its own labels loses a factor of order \min\n,\log(m^2/n)\ when the maps differ by constants and m^2/n is large. When the maps are drawn from an unknown prior (random effects), other questions need not identify the Bayes rule. The loss, the identification regret, is at most 2^-n once the law of a question’s data is known, extending a bound of Skibinsky (1968) for coins, and in the worst case at least \beta_\star^-n/(72\sqrtn) at every finite m , where \beta_\star\approx 3.583 . Beyond a few labels per question, what cannot be learned is nearly unnecessary. For constant maps (an uninformative predictor) and constant-map priors, the exponential base is exactly \beta_\star , not the 4 suggested by the moment range; for monotone maps it lies in [2,\beta_\star] .
[LG-21] SoK: Rethinking Membership Inference in Tabular Machine Learning
链接: https://arxiv.org/abs/2610.18486
作者: Cristina Pêra,Tânia Carvalho,Maxime Cordy,Luís Antunes
类目: Machine Learning (cs.LG)
*备注:
Abstract:Membership Inference Attacks (MIAs) are currently the dominant approach to evaluating privacy in machine learning applications. Despite their significance, several concerns remain unexplored, particularly with regard to tabular data. Existing evaluations focus predominantly on high-dimensional datasets, leaving the privacy behavior of small-to-medium datasets unexplored. In this paper, we present a new taxonomy for MIAs and conduct a comprehensive study of the factors influencing vulnerability. We evaluate four distinct state-of-the-art attacks across diverse tabular datasets and model architectures. Our results show that datasets with higher class granularity and majority class percentage are more susceptible to MIAs. We further demonstrate that advanced MIAs are highly dependent on adversarial knowledge of the target model, while reference model-free attacks often achieve statistically comparable results. Most importantly, we show that records with unique signatures (single-outs) remain highly vulnerable, even when common defenses are applied. These findings indicate that current defenses reduce average leakage, but fail to adequately protect the most sensitive individuals.
[LG-22] ProTok: Unified Protein Tokenization via Modular Multi-Codebook Quantization
链接: https://arxiv.org/abs/2610.18481
作者: Romain Hardy,Xiaoman Zhang,Pranav Rajpurkar
类目: Machine Learning (cs.LG)
*备注: 22 pages, 2 figures, 10 tables
Abstract:Protein sequence, structure, and functional annotations each capture complementary aspects of protein biology, yet existing tokenizers encode each modality independently, limiting the expressiveness of discrete protein representations. We present ProTok, a unified protein tokenizer that maps all three modalities into a modular discrete vocabulary via product quantization with per-modality codebook groups and modality dropout. A single ProTok checkpoint supports flexible inference from any modality subset without retraining. On seven StructTokenBench benchmarks spanning 14 tasks and 28 test splits, ProTok achieves an average rank of 1.4 across all structure-only tokenizers, placing first on 21 splits and among the top two on 27 of the 28. These gains come without sacrificing reconstruction: ProTok’s tokens reach 0.939 lDDT on CATH 4.2, showing that multi-codebook quantization and multi-modal training yield tokens that are both reconstructable and informative. ProTok’s structure tokens further lead all baseline structure tokenizers on zero-shot functional retrieval across CATH, Enzyme Commission, and Gene Ontology label hierarchies. The vocabulary also supports generation: a compact language model trained from scratch over ProTok tokens jointly generates backbone and sequence, with 52% of sampled backbones designable versus 21% for a parameter-matched GeoBPE baseline. ProTok shows that structure, sequence, and function can share one discrete vocabulary, motivating protein language models that reason over a unified token stream.
[LG-23] Graph Metanetworks for Transformer Based Language Models
链接: https://arxiv.org/abs/2610.18471
作者: Yonathan Wolanowsky,Guy Bar-Shalom,Fabrizio Frasca,Haggai Maron
类目: Machine Learning (cs.LG)
*备注:
Abstract:Weight-space learning treats the weights of neural networks as data, processed by other models called metanetworks. A key challenge is that different weights can realize the same function, related by a group of symmetries the metanetwork should respect. While symmetry-aware metanetworks for simple architectures such as MLPs are well developed, a comprehensive treatment of transformers, and in particular transformer-based language models, is still missing. Importantly, their weight space has a richer symmetry group: beyond neuron permutations, attention introduces continuous general-linear symmetries of the attention matrices and permutations of entire heads. Existing methods for transformer weights either capture only part of these symmetries or are restricted to a single, fixed architecture, and none accounts for language-model-specific components such as the token embeddings. Building on graph metanetworks, we propose a graph-based approach that is both symmetry-aware and applicable across depths and widths. We formalize a weight-space symmetry group of complete decoder-only language models and provide a graph construction that is invariant to it. We further introduce two new transformer-population datasets with novel tasks, such as training-corpus prediction and fine-tuning detection, on which our approach notably outperforms existing baselines.
[LG-24] Kairos: Transactional Security Enforcement at the System Boundary for LLM Agents
链接: https://arxiv.org/abs/2610.18457
作者: Kristoffer Hempel,Nils Loose,Lorenzo Cavallaro,Thomas Eisenbarth
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注:
Abstract:Large language model (LLM) agents are becoming an integral part of the software development lifecycle, and the capabilities granted to them allow an agent to act with the user’s authority on any resource it can access. The reach that gives an agent its utility is equally available to an attacker who can steer it. Anything an agent ingests can carry an injected instruction, and since the model does not reliably distinguish instructions from data, indirect prompt injection should be mitigated by mediating the effects the agent produces. Such mediation faces a structural difficulty. Legitimate work and the first step of an attack often look alike when an individual action is observed, and what separates them lies later in the trace, so a monitor limited to allow and deny is forced to decide before the evidence accumulates. We present Kairos, a reference monitor at the isolation boundary of an unmodified, opaque agent, where a tool call, a shell command, and a script the agent writes all cross the boundary as the same file and network operations, so that confinement holds regardless of the agent’s behavior. A declarative policy that states which sources are trusted decides most effects deterministically and defers the rest, which are held while the agent continues speculatively. A decision oracle with bounded authority, such as a language model or a human, releases or refuses each deferred effect on the trace. Across three benchmarks and three agent models, Kairos outperforms existing defenses in the trade-off between attack success and utility. In particular, on AgentDojo, it reduces attack success from 47.0% to 2.7% while its benign utility stays at the level of the undefended agent.
[LG-25] FastEEP: Fast Intra-Instance Elastic Scaling for Expert-Parallel MoE Serving
链接: https://arxiv.org/abs/2610.18451
作者: Yunfei Feng,Jingkai He,Shiqi Liu,Cheng Zhang,Dong Du,Erhu Feng
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注:
Abstract:Large-scale expert parallelism (EP) has become the standard deployment configuration for Mixture-of-Experts (MoE) models in production. Meanwhile, the highly dynamic nature of LLM workloads requires providers to continuously adjust serving capacity to match fluctuating demand. Scaling MoE serving cannot be efficiently achieved by simply adding or removing serving instances; instead, it often requires resizing the EP group within an existing instance, which we refer to as intra-instance this http URL inference engines primarily rely on instance-level scaling, which destroys and reconstructs entire serving instances, incurring substantial scaling overhead. We present FastEEP (Fast Elastic Expert Parallel), a fast intra-instance scaling system for MoE serving based on the principle of Partial Reuse. FastEEP reuses existing runtime state across worker execution, GPU memory, and communication by separating reusable execution structures from the resources activated by the current EP configuration. This separation enables coordinated resource reclamation and reconfiguration while preserving costly-to-reconstruct state, allowing fast EP scaling without full instance teardown and reconstruction. Our evaluation shows that FastEEP achieves 4.5x–11.7x faster scaling than the existing vLLM-based approach across different models, enabling efficient, fine-grained scaling of MoE serving. Subjects: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG) Cite as: arXiv:2610.18451 [cs.DC] (or arXiv:2610.18451v1 [cs.DC] for this version) https://doi.org/10.48550/arXiv.2610.18451 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-26] LunaTP: Enabling Efficient and Scalable Tensor Product for Equivariant MLIPs
链接: https://arxiv.org/abs/2610.18393
作者: Yuanchang Zhou,Mingzhen Li,Chen Wang,Xiangyu Zhang,Guangming Tan,Weile Jia
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注: 26 pages, 11 figures
Abstract:Equivariant machine learning interatomic potentials (MLIPs) promise atomistic simulations with accuracy approaching that of quantum-mechanical calculations. However, the high computational cost of their core Clebsch–Gordan tensor product (CGTP) convolutions severely limits model scalability and efficiency. In this paper, we rethink the parallel execution of CGTP convolutions and introduce LunaTP. Built on node-centric execution, LunaTP exploits path-group parallelism and multi-level reuse to accelerate CGTP convolutions and higher-order derivative computation while preserving the underlying mathematical formulation. On an NVIDIA H100 GPU, LunaTP achieves kernel speedups of up to 250.0 \times over e3nn and 17.1 \times over OpenEquivariance. It also delivers end-to-end training throughput gains of up to 28.7 \times over e3nn and 3.11 \times over OpenEquivariance. Finally, LunaTP demonstrates scalability and robust performance across models, workload scales, and GPUs, delivering substantial acceleration for training and atomistic simulation with high-accuracy equivariant MLIPs.
[LG-27] Beyond Task Success: A Theory of Representation Reuse
链接: https://arxiv.org/abs/2610.18375
作者: Zibo Meng,Andrew C. M. Austin,Alin Achim
类目: Machine Learning (cs.LG)
*备注: 42 pages, 10 figures
Abstract:A learned representation is useful beyond its source task only if later tasks can make use of what the encoder has already learned. Source-task performance, however, gives no guarantee that the representation itself will remain useful once the original head is removed. We study representation reuse with respect to a specified family of future tasks. For such a family, we define the task-relevant state as the coarsest deterministic state sufficient for all of its tasks. A larger task family can require a finer state, and losing a required distinction makes at least one future task unrecoverable on a target support containing that conflict. Exact reuse also depends on how the retained information is presented to the downstream rule. We separate this into structural accessibility, which concerns whether the required distinctions remain recoverable, and coordinate consistency, which concerns whether those distinctions can be interpreted by the same downstream rule across samples. Both failure modes appear separately in the experiments. Reuse improves as task-relevant distinctions become more accessible. Reversible sample-dependent coordinate changes can sharply reduce reuse even when the representation remains recoverable, and inverting the applicable coordinate change restores performance. The same accessibility-reuse ordering appears in prospectively specified CIFAR-100 experiments with both ResNet-18 and DenseNet-121. Finally, preserving a candidate state that contains the distinctions needed by the anticipated task family improves reuse on compatible held-out tasks, with the clearest benefit when target supervision is limited. Representation reuse is therefore relative to the tasks that follow source learning: a useful representation must retain the distinctions those tasks need and make them available in a form that a shared downstream rule can use.
[LG-28] Over-parameterization Is Neither Robust nor Frag ile in Matrix Sensing: A Geometric Theory of Solution Robustness
链接: https://arxiv.org/abs/2610.18318
作者: Runze Shi,Ziye Ma
类目: Machine Learning (cs.LG)
*备注:
Abstract:Over-parameterization is ubiquitous in modern machine learning, yet its effect on robustness remains poorly understood. We study this question in matrix sensing, a canonical non-convex problem that can also be viewed, under an appropriate parameterization, as training a two-layer quadratic neural network. We develop a rank-aware robustness theory for the Burer-Monteiro factorization when the search rank k may exceed the target rank r_star, thereby capturing over-parameterization. Our analysis shows that robustness depends jointly on the degree of over-parameterization and the local geometry of the attained second-order stationary point: increasing k generally worsens sensitivity, but stationary points with the same search rank can exhibit substantially different robustness. Motivated by this observation, we propose a trajectory-aware algorithm that uses spectral balancing and randomized sketching to steer optimization toward less sensitive solutions. In matrix-sensing experiments, the method reduces paired sensitivity by 18-29 percent for isotropic operators and 87-89 percent for weak-tail operators, without degrading reconstruction accuracy. These results demonstrate that parameter count alone does not determine robustness and identify solution geometry and the optimization trajectory as important levers for improving it.
[LG-29] Bayesian Optimization of CVaR with Unknown Hyperparameters
链接: https://arxiv.org/abs/2610.18306
作者: Valentin Margraf,Jonas Hanselle,Viktor Bengs,Serafima Lebedeva,Eyke Hüllermeier,Marcel Wever
类目: Machine Learning (cs.LG)
*备注: 17 pages, 25 figures
Abstract:We study Bayesian optimization of the Conditional Value-at-Risk (CVaR) of an expensive black-box function that depends on a decision and an environmental variable. Existing no-regret algorithms assume that the correct Gaussian process (GP) kernel and its hyperparameters are known. In practice, however, they must be estimated from limited data, often yielding a misspecified surrogate. This is especially damaging for CVaR because misspecification can distort the estimated tail-risk and thus lead to catastrophic decisions. We propose HE-CV-UCB, the first no-regret algorithm for CVaR Bayesian optimization with unknown GP hyperparameters. Given a finite candidate set containing a well-specified model, HE-CV-UCB adaptively eliminates those inconsistent with the observed data, actively queries where the survivors disagree, and returns a design that is robust across the surviving candidates. We prove that it attains cumulative regret matching the oracle rate of CV-UCB up to a factor depending only on the candidate set’s size. Across synthetic benchmarks and three real-world applications, HE-CV-UCB achieves competitive performance and frequently outperforms existing CVaR Bayesian optimization methods.
[LG-30] Reliable Risk Estimation with Kernel Mean Matching under Dependent Sampling
链接: https://arxiv.org/abs/2610.18292
作者: Alexey Zaytsev,Diana Koldasbayeva,Egor Serov
类目: Machine Learning (cs.LG)
*备注: 35 pages, 6 figures
Abstract:Kernel Mean Matching (KMM) estimates risk under covariate shift by reweighting labelled source samples so that their kernel mean embedding matches that of unlabeled target samples. Existing finite-sample guarantees are usually stated for independent samples, although many applications involve temporal, spatial, or clustered dependence. We give a finite-sample KMM analysis under dependent sampling by replacing nominal sample sizes with effective sample sizes in the concentration steps used by the classical proof. The approach uses Bernstein inequalities for geometrically strongly mixing processes and derives a model-based effective sample size. The resulting bound separates overlap, dependence, and smoothness effects with specific constants derived under a Gaussian process assumption. This analysis yields an additional tradeoff: smoother losses preserve more input dependence and therefore reduce effective sample size, while also reducing function complexity and empirical-to-population risk gaps. Applying the result to prediction yields confidence intervals that achieve near-nominal coverage in synthetic experiments and substantially reduce undercoverage on real monitoring data.
[LG-31] PING: physics informed graph neural network for advection-diffusion processes with applications to solar power forecasting
链接: https://arxiv.org/abs/2610.18280
作者: Jelena Simeunović,Baptiste Schubnel,Pierre-Jean Alet,Pascal Frossard,Rafael E. Carrillo
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注: 23 pages, 11 figures, submitted to the International Journal of Forecasting
Abstract:Advection-diffusion plays an essential role in transport phenomena such as atmospheric and cloud dynamics. Improved forecasting of cloud motion, in turn, enhances solar power production forecasts and facilitates the integration of additional photovoltaic (PV) sources into the power grid. However, both classical numerical models and emerging deep learning approaches face challenges when applied to spatially irregular sensor networks such as those formed by weather stations. In this work, we adopt a graph signal processing perspective and model advection-diffusion particles as signals on a graph to capture the dynamics of these processes. We introduce a Physics-Informed Graph Neural Network (PING) for forecasting future advection-diffusion particle concentrations on irregular grids. The model incorporates a discretized advection-diffusion equation as a soft constraint in the loss function to guide learning toward physically consistent solutions. This model is on par with state of the art models on synthetic and real transport phenomena datasets, including sea surface temperature and cloud index datasets. Furthermore, the PING model is successfully coupled with models for forecasting PV power production. This hybrid approach leverages data fusion from ground-based sensors at PV stations and satellite images with cloud concentration to improve the accuracy of intra-day PV power production.
[LG-32] Optimal Online Regret for Context-Lumpable Bandits
链接: https://arxiv.org/abs/2610.18258
作者: Chenyu Lu,Zihan Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:We study stochastic contextual bandits with an unknown context-lumpable structure. The S contexts are partitioned into r blocks, and every arm has the same mean reward across contexts in the same block. Under arbitrary context distributions and without separation assumptions, we design an algorithm with expected regret \widetildeO(\sqrtr(S+K)T) , matching the lower bound of Lee et al. (2023) up to logarithmic factors and improving their previous upper bound by a factor of r . The algorithm operates over dyadic accuracy levels, using a baseline policy and confidence-based early stopping to control exploration. It then performs cross-context elimination: since rewards are identical within a block, one selected arm eliminates all surviving records from that block, requiring at most r selections at each level. Multi-scale exploration assigns finer accuracies to frequent blocks and coarser accuracies to rare blocks, yielding a near-optimal arm for every block while keeping the weighted policy error small. Subjects: Machine Learning (cs.LG) Cite as: arXiv:2610.18258 [cs.LG] (or arXiv:2610.18258v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.18258 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-33] ach Me After You Learn It: Lookahead Self-Distillation for RLVR
链接: https://arxiv.org/abs/2610.18256
作者: Zepeng Zhai,Meilin Chen
类目: Machine Learning (cs.LG)
*备注: 18 pages, 3 figures
Abstract:On-policy self-distillation (OPSD) has emerged as a popular training paradigm for LLMs, providing dense, token-level supervision by aligning a model’s own distribution with the distribution induced by privileged information. However, existing OPSD methods derive supervision solely from privileged information of the same instance, leaving inter-instance supervision largely unexplored. We propose Lookahead Self-Distillation (LEAD), which introduces inter-instance distillation by allowing the teacher to first learn from other on-policy rollouts before teaching the student. LEAD constructs lookahead teachers through virtual RL updates on disjoint subsets of on-policy rollouts, and uses each teacher to provide cross-group token-level supervision for the other subset. To reconcile the complementary yet potentially conflicting signals from OPSD and RLVR, we introduce harmonious modulation, which adaptively modulates the GRPO advantage based on the agreement between the two signals. Extensive experiments on scientific reasoning and interactive-agent benchmarks show that LEAD achieves the best in-domain average performance across model scales while maintaining out-of-domain generalization comparable to GRPO.
[LG-34] Operator Surrogates
链接: https://arxiv.org/abs/2610.18210
作者: Christoph Schwab,Josephine Westermann,Jakob Zech
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:This chapter provides a mathematical introduction to operator surrogates, which are finite-parametric maps approximating complex (typically nonlinear) operators between infinite-dimensional (function) spaces. A typical approach to constructing operator surrogates is the encoder-approximator-decoder paradigm, whereby infinite-dimensional inputs are mapped to a finite-dimensional latent space, processed, and then mapped back to the infinite-dimensional output space. We discuss various methods for creating efficient representation systems, including stable linear frameworks like Banach frames and data-driven techniques such as Principal Component Analysis (PCA), Active Subspaces, and nonlinear autoencoders. Subsequently, we survey key surrogate architectures, contrasting spectral methods, based on polynomial chaos expansions and related technologies, with more recent neural network approaches. We further discuss architectures that do not rely on representation systems, such as Fourier Neural Operators and transformers. We elaborate on their relationship and how they relate to other neural network operator surrogates, particularly those obtained by so-called “algorithm-unrolling”. Additionally, we provide an introduction to the theoretical analysis of approximation and generalization errors. We explain how underlying low-dimensional structures, provided for example by holomorphy, are essential to overcome the so-called “curse of dimensionality”: this refers to the fact that the complexity of numerical algorithms tends to increase exponentially in the problem dimensionality. We present convergence rate bounds for holomorphic operators that depend on the number of trainable parameters and the regularity of the input and output data. Throughout, we prioritize intuition over detailed technical proofs and provide precise pointers to the literature for detailed arguments. Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML) Cite as: arXiv:2610.18210 [cs.LG] (or arXiv:2610.18210v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.18210 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-35] Reinforcement Learning for Large-Scale Lifelong Multi-Agent Path Finding with Shared Spatial Fields
链接: https://arxiv.org/abs/2610.18186
作者: Jean Seong Bjorn Choe
类目: Machine Learning (cs.LG)
*备注:
Abstract:Large-scale applications of lifelong multi-agent path finding (LMAPF) call for computationally inexpensive movement decisions under changing traffic conditions and across varied layouts. We propose a navigation policy in which a convolutional network computes a shared spatial field over the map once per timestep and a compact goal-conditioned operator derives agents’ movement preferences. The complete policy is trained end-to-end using average-reward reinforcement learning with throughput as the reward, without planner demonstrations or a separate online path planner. For structured and unstructured layouts, we train one policy per class on mixtures of small maps and deploy each without retraining on larger, unseen layouts and agent populations. Across six layouts with 10000 agents, our policies achieve throughput comparable to or higher than the strongest evaluated baselines. The proposed method has an average execution time below 10 ms per step, more than an order of magnitude faster than several established planners.
[LG-36] Random Probing for Structural Self-Interactions in Graph Neural Networks
链接: https://arxiv.org/abs/2610.18164
作者: Saku Peltonen,H. Çağrı Bilgi,Kubilay Atasu
类目: Machine Learning (cs.LG)
*备注: Accepted at LoG 2026 (oral)
Abstract:Graph Neural Networks (GNNs) fundamentally lack the ability to capture structural properties defined by node self-interactions, such as cycle counts, subgraph centrality, and heat kernel signatures. Existing solutions often rely on computationally expensive higher-order architectures or hand-crafted features. In this work, we introduce DIAL, a message-passing layer that gives nodes access to graph structure through the diagonal of graph-derived operators. Our approach uses randomized probing to provide nodes with learnable, permutation equivariant access to diagonal entries. We prove that our approach can express important structural features, including cycle counts, subgraph centrality, and heat kernel signatures. The proposed method can be integrated directly into message passing layers. Experiments on molecular property prediction, cycle counting, and pattern detection demonstrate the effectiveness of DIAL across a range of graph learning tasks.
[LG-37] Learning a Group Element from Its Action: Galois Correspondences Optimal ERM and an Abel-Ruffini Threshold for the Natarajan Dimension
链接: https://arxiv.org/abs/2610.18154
作者: Xiaoyu Li
类目: Machine Learning (cs.LG); Group Theory (math.GR); Machine Learning (stat.ML)
*备注: 55 pages, 5 figures. The abstract has been shortened to meet arXiv’s submission requirements
Abstract:We study learning an unknown element g of a group acting with finite orbits on a countable set, from examples (x,gx) . After a sample on A , the consistent hypotheses are the restrictions of a coset of the pointwise stabilizer of A , and the sample forces exactly the labels of points this stabilizer fixes; for a Galois group, they form the field generated by the data. Through this correspondence, multiclass dimensions become group invariants: the DS and graph dimensions equal the height, the Littlestone dimension the longest irredundant sequence length, and the teaching dimension the minimal base size. Among deterministic learners, symmetry makes empirical risk minimization optimal: on nondegenerate classes every consistent learner, whatever its tie-breaking, attains the optimal realizable sample complexity up to constant factors, and every consistent online learner the optimal mistake bound. Agnostically, every ERM’s sample complexity exceeds the known lower bound by at most the DS-to-Natarajan ratio, up to constants. For finite Galois extensions of countable fields, the Natarajan dimension equals a group invariant, the coset-cube number, which is at least the number of complemented factors of any chief series of any subgroup. Hence the dimensions coincide for every solvable Galois group, while for the groups A_5 and S_5 of the quintic they are 23 and 34 , and A_5 is the smallest group with a gap. For the Galois group S_n of the general equation of degree n , they coincide exactly when n\le 4 , i.e., in characteristic zero, exactly when the equation is solvable by radicals; the implication is one-way, as the simple group \mathrmPSL(2,8) has no gap. For linear, orthogonal and unitary maps of \mathbb R^d or \mathbb C^d both dimensions equal d , and a two-fibered pseudo-cube is a disjoint union of Natarajan cubes exactly when its monodromy group is abelian.
[LG-38] Revisiting Error Feedback for Model Quantization
链接: https://arxiv.org/abs/2610.18139
作者: Wai-Yiu Keung,Chi Cui,Wing-Kin Ma,Jesús Cid-Sueiro,Hei Victor Cheng
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注:
Abstract:We study causal error-feedback coefficients as design variables for model quantization. Modeling instantaneous quantization errors as independent, zero-mean variables with a common fixed variance, we formulated a metric-dependent trace problem for designing feedback coefficients. The formulation resembles fixed-order GPTQ/LDLQ feedback and supports Jacobian-aware designs that account for downstream error propagation. For a positive-definite metric \boldsymbolH , we prove that the joint rotation-feedback trace minimum is n\det(\boldsymbolH)^1/n and present an attaining pair using the geometric mean decomposition. Experiments on synthetic networks, trained MLPs, and language models show benefits from downstream-sensitive feedback, but also reveal that trace optimality need not improve finite-level performance. Small QAT experiments further illustrate direct loss-driven feedback tuning with fixed or trainable network weights.
[LG-39] he Effect of Layer Ordering in Hybrid Architectures
链接: https://arxiv.org/abs/2610.18104
作者: John Cooper,Mingchen Ma,Jiaqi Mo,Shengqi Qiu,Nathaniel Bastian,Ilias Diakonikolas,Frederic Sala
类目: Machine Learning (cs.LG)
*备注:
Abstract:Hybrid attention is rapidly becoming the dominant architectural choice for frontier models. Despite their successes, such architectures complicate model design by introducing additional choices, including the types and locations of layers. This combinatorially large design space has been studied empirically, but given the vast expense of training and evaluating models with different configurations, a principled understanding of the effects of layer selection and ordering is needed. We provide the first such analysis, showing that the dominant hybrid architecture is highly task-specific. Concretely, for any layer configuration, we prove that there exists a task where we can construct a model with this configuration that provably outperforms any other model of any different configuration under fixed resource constraints. Empirically, we observe that for standard training, learned hybrid architectures exhibit separations between layer orderings that match our theoretical findings. For instance, under the same memory budget, the optimal architecture predicted by our theory for a segmented retrieval task consistently achieves perfect accuracy, whereas the next strongest plateaus below 0.5. These results extend to deeper models, where ``containing’’ the dominant architecture for a task indicates whether the model attains high performance. In addition, we investigate the mechanisms learned by hybrids, observing that Transformer layers contribute the largest share of performance on associative recall across architectures. This work establishes insights into task-dependent hybrid architecture design relevant to the training and scaling of frontier-scale hybrids.
[LG-40] From Spectral Saliency to Finite Deletion: Evaluating Selective Language Model Edits
链接: https://arxiv.org/abs/2610.18098
作者: Ibne Farabi Shihab,Joyanta Jyoti Mondal
类目: Machine Learning (cs.LG)
*备注: 25 pages, 9 figures
Abstract:Deleting singular components offers a precise way to remove a capability from a language model, yet the first-order spectral saliency used to rank components predicts only a local loss change, whereas deletion is a finite intervention. In this paper, we study how prediction error affects selective editing, where selective editing requires degradation of a target task while preserving others. We decompose the error into derivative transfer across item splits and nonlinear response along the deletion path. We establish that outer interval screening retains every feasible candidate when every candidate’s intervals contain its finite validation loss changes on every task. We evaluate these effects using one answer-only negative log-likelihood objective for search, validation, and test, with disjoint calibration and evaluation masks. The primary configuration targets arithmetic in Mistral-7B-Instruct-v0.3 while preserving code and question answering; extensions cover Llama-3.1-8B-Instruct and reversed target roles. In the primary study, which deletes 1024 coordinates from a support of 7168, calibrated intervals cover 968 of 1024 evaluation masks. Outer screening reaches the direct-selected edit after 25 verification checks, whereas direct verification abstains at a 32-check cap, but it rejects 4 of 74 feasible masks. All three selected edits pass validation, yet only one meets every test constraint. Interval coverage, feasible-candidate retention, and independent test preservation are therefore distinct properties and require separate assessment.
[LG-41] Same Function Different Trajectory: No Standard Optimizer Is Equivariant to Transformer Reparameterization
链接: https://arxiv.org/abs/2610.18085
作者: Bum Jun Kim
类目: Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
*备注: 55 pages, 8 figures
Abstract:We show that two Transformers representing the same function can follow different training trajectories. For example, doubling a query matrix and halving its key matrix preserves the model’s function but changes the gradient scales seen by the optimizer. Our theoretical analysis establishes that standard block-local first-order optimizers fail to scale updates with parameters under a shared learning rate. For AdamW, transforming the first- and second-moment estimates alone does not correct the mismatch between parameter and update scales. To make runs restarted from equivalent checkpoints follow the same trajectory, we balance matrix pairs and transform their moment estimates together, bringing distinct but equivalent training states to the same starting point. To correct scale imbalance between paired Transformer matrices without changing predictions, we propose Symmetry-Aware Gradient Geometry Stabilization (SAGGS), which periodically equalizes the Frobenius norms of exact symmetry pairs, and its damped variant, Soft SAGGS. SAGGS reduces perplexity (PPL) by 0.374\pm0.026 at 101M parameters on WikiText-103, and Soft SAGGS reduces PPL by 0.780\pm0.040 in a 31M modern-architecture replication, both under scale imbalance. In a 123.8M checkpoint-resume experiment, reciprocally rescaling matrix pairs barely changes validation loss at restart but increases final PPL by 3.78\pm0.42 ; hard and soft rebalancing recover to within 0.02 PPL of the control.
[LG-42] Failure Recovery via Belief Updates and World Modeling for Long-Horizon Agents
链接: https://arxiv.org/abs/2610.18082
作者: Xiyang Wu,Zhichao Liu,Shengxin zhang,Zongxia Li,Ming Li,Tianyi Zhou,Dinesh Manocha
类目: Machine Learning (cs.LG)
*备注:
Abstract:Long-horizon agents often struggle to recover from failures whose causes lie several steps earlier, beyond what local reflection or backtracking can resolve. We introduce BELIEFWEAVER, a framework that combines persistent evidence across attempts with prospective reasoning over repair directions. It maintains a relational belief state over frontiers, attempts, observations, and checkpoints. Execution evidence updates related frontier priorities across attempts and rollback, while diagnostic evidence gates execution. An implicit world model (IWM) evaluates repair hypotheses, predicts progress, and identifies relations among frontiers to guide exploration. We further distill these recurring IWM reasoning components into a 9B student through supervised fine-tuning and on-policy distillation (OPD). Across three terminal-based benchmarks, BELIEFWEAVER improves aggregate final score by 5.1 percentage points over native Terminus-2 continuation, with positive trends in exploration efficiency. The distilled 9B student improves final score by 9.8 percentage points over the reported native baseline while reducing estimated teacher API cost by 54.8% relative to the matched teacher. These results show that persistent evidence and prospective reasoning can improve recovery in long-horizon agent tasks while substantially reducing reliance on teacher-model inference.
[LG-43] Causal Discovery Under Hard Selection Bias: A New Robust Score-Matching Approach NEURIPS2026
链接: https://arxiv.org/abs/2610.18075
作者: Yiwen Qiu,Francesco Montagna,Shimeng Huang,Francesco Locatello
类目: Machine Learning (cs.LG)
*备注: 30 pages, NeurIPS 2026
Abstract:Selection bias is a difficult yet widespread problem in causal discovery, occurring whenever non-random selection processes lead to data that is not representative of the underlying populations. Due to its practical importance, related works have been proposed to address the problem under local [Versteeg et al., 2022] or interventional settings [Dai et al., 2025a], but general algorithms for observational data remain elusive. In this paper, we focus on truncated data (hard selection), and provide a general result for Additive Noise Models (ANM). We show that score-based methods are a surprisingly suitable candidate due to a special property of the score function ( \nabla \log p(x) ): deterministic truncation preserves both the score of the joint distribution as well as the Jacobian of the score. While established score-matching algorithms still fail in most settings, we actively leverage this observation to propose a new method for causal discovery that is robust under truncated data with ANMs. Overall, our approach offers a general, robust solution that is agnostic to external information about the selection process and still achieves comparable performance to the state-of-the-art approaches when selection is not present.
[LG-44] Privacy-Preserving GNN Inference on Financial Transaction Graphs via Homomorphic Encryption
链接: https://arxiv.org/abs/2610.18054
作者: H. Çağrı Bilgi,Kemal Mutluergil,Ceren Yıldırım,Kubilay Atasu,Erkay Savas
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注:
Abstract:Anti-money laundering (AML) in financial networks requires modeling transaction patterns that span multiple institutions. In practice, however, banks cannot share account and transaction features due to strict privacy regulations. We propose a privacy-preserving graph neural network (GNN) inference framework tailored to this multi-bank setting. A central server holds the graph connectivity and pretrained model weights in plaintext, while each bank keeps its transaction features encrypted. Inference is then carried out entirely on encrypted features under the CKKS homomorphic encryption scheme, so that no plaintext transaction features or predictions are exposed to the server. The key to making this practical is an edge-centric GNN architecture, wherein message passing is expressed through operations on node–edge incidence matrices. This formulation not only preserves the edge-attributed multigraph structure of transaction networks but also reduces every aggregation step to plaintext–ciphertext matrix products. Since CKKS can evaluate only polynomial circuits, we also approximate each nonlinearity with a composite polynomial of the sign function, which we calibrate per activation site. On the IBM AML benchmark, our model achieves performance competitive with plaintext baselines, and with a consumer-grade GPU, classifies each queried transaction end-to-end in as little as \textbf5.31 seconds, while per-transaction latency can be amortized to \textbf1.65 seconds when optimized for throughput.
[LG-45] LADDER: A Long-Horizon Robot Manipulation Benchmark
链接: https://arxiv.org/abs/2610.18053
作者: Petko Petkov,Giuliano Albanese,Nikolay Nikolov,Luc Van Gool,Jonathan Tremblay,Danda Pani Paudel
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: 30 pages. Project website: this https URL
Abstract:General-purpose robot manipulation policies learn short-horizon skills effectively, but degrade rapidly on long-horizon tasks. Existing benchmarks cannot isolate failures because task horizon is either too short or entangled with scene and object variations. We propose LADDER, a long-horizon robot manipulation benchmark scaling from 1 to 10 subgoals within the same environment over three suites (SimplerEnv-Long, RobotArena- \infty -Long, and RoboLab-Long), 145 tasks, eight categories, two embodiments, and 2,024 simulated episodes per system. We evaluate two orchestration paradigms: text subgoals for learned low-level policies (VoLo) and tool selection orchestration (CaP-X, GPT-Policy, PSP-Agent, TiPToP). Given the full instruction directly, low-level policies achieve 85.4% on single-subgoal tasks, but drop to near zero on extended tasks. While orchestration remedies this, it remains bounded by physical execution errors of the low-level policy rather than high-level planning and achieves only ~25%. Tool selection orchestration with GPT-Policy on GPT-6 Astra significantly outperforms learned-policy methods, but incurs significant time and cost budget and succeeds only ~62%. Real-world experiments across two embodiments confirm the collapse of the low-level policy and the gain from orchestration. Project website: this https URL
[LG-46] FRACTION: Fraud Explanation Generation via Agent ic Graph Context Navigation
链接: https://arxiv.org/abs/2610.18023
作者: Cecelia Soh,Jun Hu,Yuan Li,Cheng Chen,Hui Liu,Yanjia Zhu,Hao He,Bryan Hooi,Bingsheng He
类目: Machine Learning (cs.LG)
*备注:
Abstract:Graph fraud explanation generation emerges as an important task, which requires jointly leveraging graph structure and content information to produce human-readable explanations for a given risk label. An intuitive approach is to adapt existing graph-enhanced large language models (LLMs), which inject graph structure into LLMs via graph-conditioned instruction tuning or structure-aware embeddings. However, existing methods face two key challenges in graph fraud explanation generation: a lack of evaluation protocols tailored to explanation quality, and a mismatch between the granularity of graph information they provide and what explanation generation requires. In this paper, we propose FRaud Explanation Generation via Agentic Graph Context NavigaTION (FRACTION), a framework for graph fraud explanation generation. To evaluate explanation quality, we introduce a task-specific rubric that assesses faithfulness, graph grounding, helpfulness, consistency, and conciseness, and validate it against human judgments. To focus on fine-grained details of relevant neighbors without reading the entire neighborhood, we equip a language model with graph tools that progressively disclose neighbor records. To train our graph agent, we propose a rubric-guided trajectory distillation strategy through which the agent learns both tool-use behavior and structured explanation generation. Experimental results on three fraud datasets demonstrate consistent improvements over state-of-the-art baselines under a human-aligned rubric.
[LG-47] Improving Neural Symbolic Regression with a Diverse Prior and Post-training Alignment
链接: https://arxiv.org/abs/2610.17994
作者: David Otte,Jörg K.H. Franke,Frank Hutter
类目: Machine Learning (cs.LG)
*备注:
Abstract:Symbolic regression is the problem of finding an expression that fits a dataset. Solving it requires searching a combinatorial space of candidate expressions. Neural symbolic regression can amortize that search by training a model once on synthetic data and then predicting expressions for a new dataset. But existing models still underperform genetic programming on real, noisy data. We argue that the deficit lies in the training rather than in the paradigm, and present RASR (Reinforced Amortized Symbolic Regression), a method with two key changes to the training recipe. First, we design a synthetic prior that narrows the gap with real-world data: inputs come from the prior of a tabular foundation model; constants are sampled bottom-up, conditioned on the data; targets carry heteroscedastic and structured noise; and everything is standardized. Second, we split training into two stages. Supervised pre-training maximizes the token-level likelihood of a reference expression. This is followed by reinforcement-learning post-training that aligns the model with its inference-time use by rewarding held-out fit and penalizing expression size. On the black-box track of SRBench 2.0, RASR achieves the best mean test R^2 rank of all evaluated methods. Three RASR variants, trained with different size penalties, make up most of the accuracy-complexity Pareto front. On Strogatz, our method recovers more ground-truth expressions than any baseline, and on Feynman it is competitive with the best-performing methods.
[LG-48] Counterfactual Certification for Sequential Unlearning with Long-Run Performance Guarantees
链接: https://arxiv.org/abs/2610.17993
作者: Hanna Benarroch(DI-ENS),Jamal Atif(CMAP),Olivier Cappé(DI-ENS)
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Certified machine unlearning ensures the proximity between the unlearned model and a model retrained without the data to be forgotten. Extending this framework to sequential deletion requests, typically arriving as groups of variable size (e.g., all records of one user or of coordinated users), is challenging. For unlearning methods based on fine-tuning on the retain sets, this leads to either degrading performance or increasing computation as the number of unlearning requests grows. For independent successive requests, we instead certify each model release against a counterfactual unlearning trajectory in which the group being deleted never took part. With a fixed fine-tuning budget per deletion, we show that, for smooth strongly convex objectives, this approach keeps the performance loss bounded, uniformly in the number of requests. Experiments with multiclass logistic regression on FEMNIST show that the resulting algorithm needs fewer corrective steps than existing sequential methods and outperforms training-time differential privacy, particularly when group sizes are heterogeneous.
[LG-49] Exact power indices for plurality-voting ensembles NEURIPS2026
链接: https://arxiv.org/abs/2610.17980
作者: Ilie Sarpe,Theofanis Georgakopoulos,Aristides Gionis
类目: Machine Learning (cs.LG); Data Structures and Algorithms (cs.DS); Computer Science and Game Theory (cs.GT)
*备注: Extended version of a work accepted at NeurIPS 2026. 59 pages, 9 figures, 11 tables
Abstract:Plurality-voting ensembles aggregate the predictions of multiple base classifiers and output the most voted class, with ties typically broken at random. Despite the simple aggregation rule, attributing the impact of each base classifier to the ensemble’s predictions is far from straightforward. That is, a single base classifier’s impact depends on how it interacts with the votes of all other classifiers in the ensemble, for example, in resolving ties, and not only on its own predictions. We address the attribution problem using established tools from cooperative game theory, studying the Banzhaf and Shapley power indices for new games modeling plurality voting and random tie-breaking, for binary and multiclass classification settings. Computing power indices is, in general, intractable. Surprisingly, our methods, collectively named ENPOWER, enable the exact computation of Banzhaf and Shapley values in linear time for the binary classification setting and in low-degree polynomial time for the multiclass setting. The algorithms in ENPOWER rely on new combinatorial formulations of the Banzhaf and Shapley indices, yielding closed-form solutions for the binary case and ordinary generating function representations for the multiclass case. We validate empirically the methods in ENPOWER demonstrating their high applicability, providing attribution values not captured by existing methods. We apply the attribution of ENPOWER to: ensemble pruning and prompt-based ensembling of vision-language models.
[LG-50] Rarity Determines Minimax Rates of Semantic Entropy From Weak Pairwise Judgments With a Known Channel
链接: https://arxiv.org/abs/2610.17971
作者: Xiaoli Yu,Jiamiao Liu
类目: Information Theory (cs.IT); Machine Learning (cs.LG)
*备注: 39 pages, 7 figures, 7 tables. Code: this https URL
Abstract:The semantic entropy of a language model is the entropy of the meanings of its sampled answers. The meanings are never observed, and the entropy is estimated from the verdicts of a pairwise equivalence judge. Given the meanings of n answers, a known channel declares each pair equivalent independently, with probability q+\Delta within a meaning and q across meanings. At most K meanings occur, and q is bounded away from 0 and 1 . For a weak judge, with n\Delta^2\to0 , the global minimax rate is set by distributions near a vertex of the simplex and overstates the difficulty elsewhere. We show that the difficulty is set by how many meanings can be rare. Over distributions with at most s meanings below a fixed probability \tau1/(K-s) , the minimax squared risk is of order N^-1/s for s\ge2 , N^-1\log^2(e+N) for s=1 and N^-1 for s=0 , capped at one, where N=n^K\Delta^2K-2 . The matching lower bounds hold on every fixed polynomial window n\Delta^2\le n^-\kappa . One estimator attains these rates without knowing s or \tau : it averages products of centered verdicts along all spanning trees of every k -set of answers, projects the resulting power sums onto the valid moment set and inverts them. The case s=K-1 gives the global rate N^-1/(K-1) and the consistency threshold: with \Delta=n^-\gamma the entropy is uniformly consistently estimable exactly when \gammaK/(2K-2) . On polynomial windows, the same estimator attains the sharp local minimax constant on shrinking neighborhoods of interior points with distinct probabilities. Restricting the average to stars multiplies the constant by K^K-3 .
[LG-51] AlignPred: Error-Controlled KV Cache Compression via Alignment and Predictive Coding
链接: https://arxiv.org/abs/2610.17970
作者: Xu Yang,Jiapeng Zhang,Zhiwei Zuo,Yuxin Chen,Feiqiang Sun,Chengguang Xu,Feng Jin,Zhuo Tang
类目: Performance (cs.PF); Machine Learning (cs.LG)
*备注:
Abstract:Long-running agents retain KV caches across turns and sessions for hours or days. Compressing these persistent caches requires reducing storage while controlling reconstruction error. We present AlignPred, a codec that combines local alignment with predictive coding under a configurable KV reconstruction-error budget. Local orthogonal transforms align related heads or layers into a common coordinate frame, allowing AlignPred to combine token-wise prediction with same-token prediction across heads or layers. Residuals are quantized on an error-bounded lattice and entropy-coded, without learned predictors, encoder-side mode search, or large joint transforms. Across three attention mechanisms, AlignPred requires less storage than the evaluated baselines at comparable measured reconstruction error. Against the evaluated CPU codec implementations, it decodes at least 1.6 times faster while matching or improving encode time. AlignPred achieves near-native task scores at substantially compressed operating points on LongBench-Pro at 128K context. AlignPred’s SGLang HiCache integration reduces persistent target-KV storage by 42-72% relative to native HiCache, with comparable latency.
[LG-52] Neural Dimensional Analysis: Foundations of Dimensionless and Scale-Consistent Computation
链接: https://arxiv.org/abs/2610.17969
作者: Bum Jun Kim,Makoto Kawano,Yusuke Iwasawa,Yutaka Matsuo
类目: Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
*备注: 56 pages, 9 figures
Abstract:Modern Transformers rely on many scale-control mechanisms, including normalization, residual scaling, attention temperatures, loss weighting, and adaptive updates. To explain how these mechanisms maintain scale consistency, we introduce Neural Dimensional Analysis (NDA), a unified framework based on dimensional consistency. We define neural units that specify how tensors rescale and which quantities can be added, just as physical units do in physics. Our rules match residual units, nondimensionalize attention and gate inputs, combine losses across units, and convert gradient units into parameter units. We implement NDA through dimensionless Unit Violation Indices (UVI) for diagnosing scale mismatches within Transformers, a unit-consistent multi-loss objective, and the Gauge-Canonical Transformer (GCT) with normalized coordinates and orthogonalized residual additions. In language-model experiments spanning 51M to 1B parameters, GCT remains stable during training across a broad sixteenfold learning-rate range, outperforming Pre-LN, LLaMA-style, and QK-Norm in 124M validation loss at tuned learning rates. GCT’s normalized residual stream also shows improved robustness to quantization. At 1B parameters, calibrated per-tensor int8 quantization increases GCT’s validation loss by only 0.009 nats, compared with up to 5.51 nats for the baselines. Vision and physics-informed neural network experiments further confirm NDA’s benefits beyond language modeling. These results establish dimensional consistency as a neural network design principle.
[LG-53] CALM-5G: Event-Level Conformal Risk Control for Predictive Service-Assurance Alarms
链接: https://arxiv.org/abs/2610.17962
作者: Ömer Faruk İşcanlı,Yağmur Sabucu Sandal
类目: Networking and Internet Architecture (cs.NI); Machine Learning (cs.LG)
*备注: 15 pages, 4 figures, 5 tables. Submitted to IEEE Transactions on Network and Service Management
Abstract:Predictive service assurance converts forecasts of key performance indicators into alarms, and that conversion decides whether the assurance function is deployable. We present CALM-5G, an assurance layer that operates above any forecasting backbone and issues alarms with distribution-free, finite-sample guarantees. It calibrates on service-level violation events, scores violation onset directly, selects its calibration scheme out of sample, and reports the alarm duty cycle with every certified miss rate. Four corpora are evaluated separately, namely a production cellular network, commercial 5G traces, an open radio access network testbed and our 5G standalone testbed. Risk control on the regression-derived score yields duty cycles of 0.24 to 0.91. Onset targeting with cross-validated calibration reduces time-in-alarm by 56.3% to 89.7% on three corpora, with no significant change on the fourth. The realised miss rate stays within a two-standard-error tolerance on three corpora and within one standard error of it on the fourth. Under label shift, pooled batch calibration exceeds that tolerance structurally, by 10.9 standard errors on the production corpus. Online calibration restores compliance. Calibration schemes cannot be ranked on the split that fits their thresholds. Miss rate and duty cycle are certified jointly where within-entity dependence leaves sufficient calibration units.
[LG-54] Stein Variational Distributional Reinforcement Learning: Score-Guided Particles for Multivariate Return Distributions
链接: https://arxiv.org/abs/2610.17925
作者: Manuel Haußmann,Melih Kandemir
类目: Machine Learning (cs.LG)
*备注: preprint
Abstract:Decisions that weigh several reward signals against each other require the joint distribution of the return vector. Categorical critics scale exponentially in the reward dimension, and quantiles have no canonical multivariate analogue. Multivariate distributional critics therefore represent each return distribution by M equal-weight particles trained with a discrepancy loss. Beyond their atoms, the particles provide no sampler and cannot resolve events rarer than 1/M. We propose Stein variational distributional RL (SVDRL). Its critic outputs Stein particles, which Stein variational gradient descent transports along a learned score field of the distributional Bellman target. We learn this field by bootstrapped denoising score matching across noise levels and tie it across the Bellman recursion by a consistency loss, which we prove is a non-expansive temporal-difference scheme in score space. The same field certifies the particles. Its reverse-diffusion sampler satisfies a supremal Wasserstein policy-evaluation bound that holds for bounded, possibly multimodal returns. The Stein particles inherit this bound up to a gap that we measure without ground truth, and the sampler answers tail-risk queries below the particle resolution at any sample size. Empirically, SVDRL is more accurate on rare events than particle critics with the same number of particles, and on par on MuJoCo continuous control.
[LG-55] NBCLT: Parameter-Agnostic Norm-Ball Clipped Local Training
链接: https://arxiv.org/abs/2610.17913
作者: Linxuan Pan,Shucheng Yang,Junchi Yang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Distributed learning with multiple local steps reduces communication but aggravates drift, and its practical performance often depends on carefully tuned hyperparameters under heterogeneous data and communication constraints. At the same time, recent Muon-type optimizers suggest that constraining updates within suitable norm balls provides a simple way to encode geometric priors into optimization. In this work, we propose NBCLT (Norm-Ball Clipped Local Training), a unified local-training framework for both federated learning (FL) and decentralized learning (DL). NBCLT combines a predict-and-correct local update rule with Euclidean projections onto flexible norm balls, thereby performing norm-ball clipping under flexible update geometries while retaining a simple implementation. In FL, its problem-parameter-free regime achieves \mathcalO(\epsilon^-2) communication complexity and \mathcalO(\epsilon^-4) total stochastic-gradient complexity without assuming bounded heterogeneity, when the per-round local-computation budget is scaled appropriately with the horizon. In DL, NBCLT is compatible with both static and time-varying mixing and preserves these accuracy exponents under the analogous computation scaling, up to fixed network constants and an additive initialization cost.
[LG-56] Stable-Plastic Subspace Construction for Differentially Private Continual LoRA
链接: https://arxiv.org/abs/2610.17904
作者: Jin Liu,Ning Xi,Yinbin Miao,Junkang Liu,Ke Cheng,Jianfeng Ma
类目: Machine Learning (cs.LG)
*备注:
Abstract:Continual learning (CL) adapts pretrained models to a sequence of tasks while retaining knowledge acquired from earlier tasks. When task data are sensitive, differential privacy (DP) provides record-level protection, but maintaining both adaptation and retention under privacy constraints remains challenging. We study this problem in continual low-rank adaptation (LoRA) and observe that: 1) private training increases forgetting primarily during subsequent adaptation rather than initial task acquisition; 2) jointly perturbing both LoRA factors introduces a non-negligible bilinear noise interaction; and 3) DP aggravates the stability-plasticity dilemma, hindering both knowledge retention and adaptation to new tasks. Based on these observations, we propose DP-SPA, which constructs a stable subspace from previously privatized updates and a plastic subspace from pretrained weights within its orthogonal complement. The method fixes these bases and privately optimizes only the LoRA factor, eliminating the bilinear noise interaction within each step. Before merging each task update, it rescales stable-direction coefficients using historical and current update energies. We establish a privacy guarantee and show that recalibration does not increase historical overlap and tightens a conditional upper bound on old-task loss increase. Experiments across five datasets, eight settings, and multiple privacy budgets show that DP-SPA improves average accuracy and reduces forgetting compared with the evaluated baselines.
[LG-57] FlashMask-3: Efficient and Expressive Mask-Aware Distributed Attention NEURIPS2026
链接: https://arxiv.org/abs/2610.17901
作者: Guoxia Wang,Qianyue He,Siming Wu,Haoyang Xie,Qiwen Bao,Jinle Zeng,Jiabin Yang,Tian Wu,Dianhai Yu
类目: Machine Learning (cs.LG)
*备注: Published as a main conference paper at NeurIPS 2026
Abstract:Long-context training of large language and multimodal models couples structured attention masks (document, prefix-LM, share-question, sliding-window, block-sparse) with multi-node context parallelism, making the mask a dominant bottleneck. FlashMask’s column-wise sparse mask compresses them from O(n^2) to O(n) via four query-axis boundary vectors, covering broad structured patterns. Implemented for Ampere GPUs, it neither exploits FlashAttention-3’s TMA/WGMMA warp-specialized pipeline on Hopper nor supports distribution. The column-wise sparse mask is indexed by key positions but stores query-axis interval boundaries. Under any multi-chunk context-parallel assignment, it reduces per-rank mask localization to a provably zero-communication \emphclip–shift–max on the boundary vectors, compatible with the single-GPU kernel. FlashMask-3 introduces a mask-aware Hopper kernel with three-role warp specialization atop FlashAttention-3’s TMA/WGMMA pipeline, plus a mask-aware distributed extension: zero-communication mask localization, a sparsity-aware load balancer (HCC-IPO), and a mask-aware compute-communication overlap with mask-driven communication pruning, whose forward all-gather uses topology-aware hierarchical routing. On a single H100 across 12 masks at 4K–128K context, it improves achieved TFLOPs over FlashMask, FlexAttention, and MagiAttention by 45–141%, 34–71%, and 2–44%. At context-parallel degrees 4–32 with 8K sequence per device, it attains 31–75% compute-communication overlap efficiency and improves throughput over Megatron-Core CP ring, Megatron-Core CP all-gather, and MagiAttention by 1.80\times – 3.71\times , 1.71\times – 2.92\times , and 1.09\times – 1.46\times . In end-to-end training of a 118B-A14B MoE at 128K on 128 H100 GPUs, it raises throughput by 7.3–11.0% over a Dual Chunk all-gather baseline, with attention forward+backward 24.5–44.9% faster.
[LG-58] Deep Mesoscopic Dynamics
链接: https://arxiv.org/abs/2610.17885
作者: Yue Zhao,Shiqi Wu,Yihang Ouyang,Sergey Grebenchuk,Magdalena Grzeszczyk,Xiaoxu Zhao,Kostya S. Novoselov,Maciej Koperski,Qianxiao Li
类目: Machine Learning (cs.LG); Statistical Mechanics (cond-mat.stat-mech); Computational Physics (physics.comp-ph)
*备注:
Abstract:Microscopic simulations resolve spatial patterns but are feasible only over small domains and short timescales, whereas bulk descriptions erase those patterns through spatial averaging. Mesoscopic fields preserve this spatial organisation, but their effective equations are known only for selected microscopic systems and scaling limits. We introduce Deep Mesoscopic Dynamics (DeepMesoD), which uses neural networks to learn an effective mesoscopic field equation from microscopic trajectories at a prescribed coarse-graining scale. For the nonconserved dissipative dynamics studied here, DeepMesoD learns the free-energy functional with a spatial interaction kernel and a state-dependent dissipation operator constrained by the fluctuation-dissipation relation. We first validate the construction in the Ising-Kac model, where the learned mesoscopic equation recovers the known hydrodynamic limit and rollouts remain stable far beyond the training horizon. We then treat two systems without closed-form mesoscopic descriptions. For a two-dimensional Heisenberg ferromagnet governed by atomistic spin dynamics, DeepMesoD learns a mesoscopic model whose predictions match atomistic references up to 1 ns, and a rollout with an unseen initial condition remains bounded up to 25 ns. For layered CrBr _3 , the learned kernel captures the suppression and sign reversal of interlayer coupling at a stacking fault whose location and strength are not supplied as inputs. Controlled changes to the learned interaction show how fault number and placement shape multistep magnetic responses, complementing our magneto-photoluminescence measurements. Across the two atomistic systems, DeepMesoD achieves a three-to-four-order-of-magnitude reduction in wall-clock time compared with atomistic spin dynamics.
[LG-59] Searching for What Matters: Open-Ended Discovery of Multiple Hypotheses with Agent ic Data Exploration
链接: https://arxiv.org/abs/2610.17874
作者: Sanchit Kabra,Nurendra Choudhary,Jiangshu Du,Eddie Huang,Danai Koutra,Chandan K Reddy
类目: Machine Learning (cs.LG)
*备注:
Abstract:Discovering what drives an outcome from data is central to science and often requires deciding what is worth investigating before a specific hypothesis is known. Yet most existing hypothesis generation settings substantially narrow this search by providing a research question, target relationship, or curated context. We introduce open-ended multiple hypothesis discovery, where a system receives only a large tabular dataset and an outcome and must discover multiple data-supported hypotheses without knowing in advance the relevant variables, functional forms, conditions, or number of underlying mechanisms. To study this problem, we introduce a benchmark built from structural causal models grounded in real-world data schemas, containing 351 planted hypotheses across 15 tasks and five domains, with 53–65 variables and 1M–10M rows per task. The causal model is hidden from the discovery method and retained only by the evaluator, allowing proposed hypotheses to be tested through interventions and separating observational discovery from causal validation. We further propose HypAgent, which treats the dataset as an interactive environment and performs iterative search by proposing candidate directions, triaging and refining promising hypotheses, and removing explained signal so that each accepted discovery redirects subsequent search. HypAgent achieves 0.47 precision, 0.51 recall, and 0.49 F1, compared with 0.35 F1 for the strongest baseline. Our analysis shows that many failures occur after a relevant variable has already been surfaced, suggesting that correctly characterizing a mechanism’s functional form and conditions, rather than merely identifying relevant variables, remains a key challenge in open-ended discovery.
[LG-60] Persistence Positional Encoding: A Minimal Transformer Recipe for Persistence Diagrams
链接: https://arxiv.org/abs/2610.17848
作者: Eunwoo Heo,Ganghyun Kim,Jaejun Yoo
类目: Machine Learning (cs.LG)
*备注: 25 pages, 6 figures
Abstract:Persistence diagrams (PDs) represent multiscale topological information as multisets of birth-death pairs (b,d) . How best to use them with Transformers, however, remains unclear. Connecting PDs to Transformers requires choosing how topological information is tokenized and exposed to self-attention. Prior methods use specialized designs for tokenization, aggregation, Transformer blocks, or attention, but are hard to compare since reported results rarely share PD inputs, splits, or model-selection rules. We instead confine the diagram-specific design to point representation, mapping each birth-death pair through fixed random Fourier features while leaving the Transformer backbone unchanged, and call this recipe Persistence Positional Encoding (PPE). In matched controls on eight TU graph benchmarks (TU8), changing only point representation shifts macro accuracy by 5.49 to 6.47 percentage points (pp), against 0.47 to 1.45 pp from changing the downstream module, identifying point representation as the dominant design variable tested. We further reevaluate vectorizations, kernels, learned representations, and Transformer-based models under one protocol with shared PD inputs, outer folds, and validation-only selection. Among all evaluated PD methods and downstream configurations, PPE achieves the highest mean on most benchmarks, including TU8, ORBIT5K, and MCB-C. Finally, we prove that grid averaging leaves a reconstruction error at any resolution, whereas a Transformer with fixed random Fourier frequencies is almost surely universal. PPE thus offers a minimal, Transformer-native baseline for persistence diagrams, and a common protocol for future comparison.
[LG-61] GeoNIRS: Geometry-Aware Pretraining for Heterogeneous fNIRS Representation Learning
链接: https://arxiv.org/abs/2610.17845
作者: Zongsheng Li,Shengyuan Zhu,Zimo Zhang,Hanyao Li,Shengming Wang,Quanying Liu
类目: Machine Learning (cs.LG)
*备注: 30 pages, 5 figures, and 21 tables. Code and pretrained checkpoints are available at this https URL
Abstract:Functional near-infrared spectroscopy (fNIRS) provides portable and spatially localized measurements of cortical hemodynamics, but differences in channel count, probe layout, signal distribution, and task type hinder joint pretraining across datasets. We introduce GeoNIRS, a geometry-aware framework that uses source-detector locations as a shared spatial reference across heterogeneous probe layouts. GeoNIRS compresses variable-size channel sets into a fixed-size latent representation and employs bidirectional HbO/HbR cross-reconstruction with contiguous temporal masking. GeoNIRS-base is pretrained on eight heterogeneous datasets, while GeoNIRS-large scales to 22 datasets. Neither downstream data nor labels are used during pretraining, enabling evaluation under a strict cross-task setting. To our knowledge, GeoNIRS is the first fNIRS framework to explicitly model physical channel geometry during joint pretraining across heterogeneous channel configurations, and the first to provide complete source code and downloadable pretrained checkpoints. We evaluate GeoNIRS on eight held-out tasks spanning clinical and auditory assessment, pain and affective processing, semantic decoding, and motor function. GeoNIRS achieves the highest task-averaged balanced accuracy among the evaluated conventional machine-learning, fNIRS-specific, general time-series, and fNIRS pretraining baselines. Compared with random initialization, pretraining improves average balanced accuracy by 4.12 percentage points under linear probing and 3.30 points under full fine-tuning, with larger benefits when labels are scarce. GeoNIRS also remains robust to missing input channels. Geometry perturbations and pretraining ablations further show that both channel geometry and cross-chromophore reconstruction contribute to the learned representations.
[LG-62] Follow the Perturbed Leader for Linear Bandits: sqrtT Regret over General Convex Sets
链接: https://arxiv.org/abs/2610.17818
作者: Chen-Yu Wei
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:We give the first adversarial linear bandit algorithm based on Follow-the-Perturbed-Leader (FTPL) that achieves \sqrtT regret over general convex action sets. Our algorithm attains \widetilde\mathcalO(d^5/4\sqrtT) regret using only linear optimization oracles, and the near-optimal \widetilde\mathcalO(d\sqrtT) regret when an optimal-design exploration distribution is further available. This improves on the T^2/3 regret of the FTPL algorithm of Dani and Hayes (2006) for general convex sets, and generalizes the \sqrtT regret of Lévy et al. (2026) over two special action sets (hypercube and Euclidean ball). Compared to previous oracle-efficient linear bandit algorithms based on continuous exponential weights, the FTPL approach generally has better computational efficiency, and is much simpler to implement. The key algorithmic element is a perturbation distribution that adapts to the estimated rewards. Its simplest form, an adaptive Gaussian perturbation, is straightforward to implement and already attains the regret bound above. We further extend it to two general constructions, random-weighting and elliptical perturbations, each covering distributions from light tails (Gaussian) to heavy tails (Student- t_2 ). On the technical side, we develop a systematic way to analyze the stability term in the FTPL regret through the Fisher information of the perturbation distribution. This builds a general framework for FTPL beyond existing ones such as the hazard rate and the FTRL-FTPL duality. Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML) Cite as: arXiv:2610.17818 [cs.LG] (or arXiv:2610.17818v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.17818 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-63] Hybrid Parallel Bessel KAN Model: a Kolmogorov Arnold Network: integrated framework for developing a novel architecture to reconstruct the fused-basis approach via a parameter free quadratic interaction
链接: https://arxiv.org/abs/2610.17779
作者: Mintu Pathak,Meshv Patel,Mohendra Roy,Jwngsar Brahma
类目: Machine Learning (cs.LG)
*备注:
Abstract:The Kolmogorov-Arnold Network (KAN) challenges traditional MLPs by placing learnable activation functions on network edges. However, scaling KANs to spatial data exposes limitations in default B-Splines. While hybrid architectures mix multiple basis families on a single edge to resolve this, monolithic superposition causes geometric overlap, parameter explosion, and severe numerical ill-conditioning ( \kappa(G) \approx 10^16 ).To overcome this, we propose Parallel BesselKAN, which isolates B-Spline and low-order Bessel branches into parallel processing pipelines integrated via a parameter-free quadratic interaction. This decoupled architecture acts as an efficient internal ensemble, bypassing monolithic stability bottlenecks. It compresses the parameter footprint by 35% while achieving 97.99% accuracy on MNIST. Although standard MLPs retain a hardware speed advantage for pure vision tasks, Parallel BesselKAN provides a robust, parameter-efficient framework stabilizing mixed-basis networks and paving the way for scalable spatial adaptations like Convolutional KANs.
[LG-64] MissPath-FM: Flow Matching with Structured Priors for Partially Observed Time Series NEURIPS2026
链接: https://arxiv.org/abs/2610.17715
作者: Genpei Zhang
类目: Machine Learning (cs.LG)
*备注: Accepted at NeurIPS 2026. 24 pages. Project page: this https URL Code: this https URL
Abstract:Flow matching has emerged as a powerful alternative to diffusion models for generative modeling, but standard conditional flow matching is fundamentally misaligned with partially observed time series: starting from an uninformed Gaussian source forces the model to transport even already-observed entries from noise to the target, wasting capacity on trivially known regions. We propose MissPath-FM, a partial-observation-aware flow matching framework that redesigns the source prior and probability path for incomplete time series. MissPath-FM employs structured source priors (interpolation priors and Gaussian process posterior priors) to initialize the flow close to the observed signal, paired with a missingness-aware smooth path that allocates transport asymmetrically between observed and missing regions. We evaluate systematically across four datasets, four missingness mechanisms (MCAR, block, MNAR, irregular), two missing ratios, and three sequence lengths (T in 24, 48, 96), totaling 48 settings against eight baselines. MissPath-FM wins 29 of 32 settings at T=48 and 45 of 48 overall (93.8%). Our ablations reveal a regime-dependent design hierarchy in which no single component suffices: under random missingness the source prior yields a large initial gain, but under structured block missingness it can fail entirely and the learned flow becomes indispensable; the missingness-aware path contributes a further 12%-26% across all regimes. Mechanistically, structured priors shorten observed-region transport by up to 15x, concentrating the velocity field on genuinely uncertain entries. Project page: this https URL
[LG-65] Confine to the Task Anchor the Rest: Concept-Guided Retention Regularizer for Parameter-Efficient Fine-Tuning
链接: https://arxiv.org/abs/2610.17713
作者: Ibna Kowsar,Ivan Medri,Fanyu Meng,Xin Liu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Parameter-efficient fine-tuning (PEFT) has been widely used, yet the adapted model suffers catastrophic forgetting of pre-trained knowledge. While most existing work hardens the adapter by constraining where the weight update may lie, we note that constraining this weight channel does not constrain the activation channel. In this paper, we propose Concept-Guided Retention (CGR), a plug-and-play regularizer that constrains both for any PEFT base. CGR reads a frozen sparse autoencoder (SAE) and treats two disjoint concept sets differently. It softly confines the update to the task-concept subspace (via weight channel), and anchors a selected set of general-purpose concepts to the frozen model (via activation channel). We also provide anchor value bounds based on weight updates to understand how they interact, and show that both terms are necessary. On a CLIP vision panel, CGR significantly improves retention at comparable accuracy for every base, and the same pattern holds for world-knowledge retention under Llama-3.1-8B instruction tuning. Because the protected directions are named SAE concepts, CGR remains inspectable and controllable.
[LG-66] Focus on What You Need: Global-to-Local Dynamic Attention for Multimodal Sarcasm Detection
链接: https://arxiv.org/abs/2610.17707
作者: Dehao Wu,Xingjie Zhuang
类目: Machine Learning (cs.LG)
*备注: 10 pages, 8 figures
Abstract:Multimodal sarcasm detection has attracted increasing attention due to the widespread use of images and text on social media platforms. Existing studies generally treat the semantic or emotional inconsistency between images and text as a key cue for identifying sarcastic expressions. However, images in real-world scenarios often contain multiple visual objects, many of which are irrelevant to sarcasm-related semantics. Such redundant visual information may introduce noise during cross-modal interaction, distracting models from focusing on critical sarcasm cues and potentially leading to inaccurate predictions. To address this issue, we propose GLDA (Global-to-Local Dynamic Attention), a multimodal sarcasm detection framework that models cross-modal interactions from a global-to-local perspective. Specifically, GLDA first employs a global multimodal fusion module to capture high-level semantic interactions between visual and textual modalities, enabling the model to identify coarse-grained cross-modal inconsistencies. To further reduce the influence of irrelevant objects, GLDA introduces a local gated cross-modal attention mechanism that dynamically filters redundant visual and textual features and refines multimodal representations at a finer granularity. In addition, a multi-branch learning objective is designed to jointly optimize unimodal and multimodal representations. Experimental results demonstrate that GLDA effectively improves sarcasm detection performance and achieves competitive results compared with existing approaches. The code will be available at this https URL.
[LG-67] Mixed Precision Scaling Laws for Quantized Pretraining
链接: https://arxiv.org/abs/2610.17637
作者: Jinuk Kim,Rajarshi Saha,Muhammed Fatih Balin,Youngsuk Park
类目: Machine Learning (cs.LG)
*备注: Preprint
Abstract:Since scaling laws were introduced for modern LLMs (Kaplan et al., 2020), several works have extended them to account for quantization, tracking the emerging low-precision data types supported on AI accelerators. We develop a more accurate scaling law for quantized pretraining with mixed precision across the forward and backward passes, grounded in optimization theory and validated empirically. Our formulation separates the effects of quantization on model capacity and on optimization efficiency, and explicitly models the persistent optimization error from forward and backward quantization identified through convergence analysis. To select the regimes for fitting and evaluating our law, we first decompose existing 4-bit training recipes into their forward and backward components, isolating those that most affect loss. Our Qwen3 experiments under selected 4-bit regimes, together with published Llama-style experiments, show that our law predicts loss with up to 11\times lower MSE than existing laws as model size and token count scale. Finally, we extend our law to layer-wise mixed precision and identify the optimal quantized-pretraining recipe, i.e., model size, training tokens, and per-layer forward/backward precision, under a hardware-aware compute budget. Unlike prior precision scaling laws, our law predicts that the optimal precision increases with the compute budget.
[LG-68] FedLoRD: Personalized Federated Learning based on Low-Rank Structures of Model Parameter Differences
链接: https://arxiv.org/abs/2610.17621
作者: Tsutahiro Fukuhara,Junya Hara,Hiroshi Higashi,Yuichi Tanaka
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注:
Abstract:We propose FedLoRD, a personalized federated learning (PFL) method based on the low-rank structure of the differences between the global and personalized model parameters. PFL is a branch of federated learning that aims to learn models tailored to the local data of each client under biased local data distributions. In its local training, each client learns a personalized model by regularizing the difference between its model parameters and the global ones sent from the server. The regularization function needs to be designed with priors on the model parameter difference; however, the existing PFL methods rely on empirical choices (e.g., \ell_1 - and \ell_2 -norms), which do not take into account the actual characteristics of the difference. The resulting mismatch of priors can lead to suboptimal personalized models. To address this issue, we propose a PFL method whose regularizer is built on an observed characteristic of the model parameter differences: they are approximately low-rank at every layer. In our method, we first formulate a local training problem with nuclear norm regularization on the layer-wise differences. We further rewrite this problem via the factorized representation of the nuclear norm so that it can be efficiently solved by a plain stochastic gradient descent method. Image classification experiments on three benchmark datasets demonstrate that FedLoRD outperforms several existing PFL methods.
[LG-69] raining Generalist Navigation Policies via Action Self-Distillation
链接: https://arxiv.org/abs/2610.17593
作者: Bowen Feng,Keqin Wang,Yulong Yang,Amogh Joshi,Christine Allen-Blanchette,Dhruv Shah,Felix Heide
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: Accepted to the 10th Conference on Robot Learning (CoRL 2026)
Abstract:Language provides a natural interface for humans to specify and reason about goals, but training language-conditioned robotic policies and reasoning typically requires expensive manual annotations. We propose Action Self-Distillation (ASD), a method for learning language-conditioned reasoning for robotic navigation without human-annotated language or reasoning labels. ASD uses an open-source VLM, which contains broad world knowledge and reasoning capabilities, as both a teacher network and a student policy. During training, the self-teacher receives privileged trajectory feedback and infers the underlying navigation goal together with goal-oriented reasoning. The student is then trained to condition on the inferred navigation goal, predict future ego-frame trajectory, and provide a description of the scene semantics, plan, and reasoning, from onboard observations alone. Across 13 challenging indoor and outdoor real-world navigation tasks, our method improves success rate over the strongest baseline by 30% in indoor settings, 30% on multi-stage tasks, and achieves 95% success rate on outdoor tasks, validating that privileged trajectory feedback can enable self-distillation for scalable generalist navigation policies without human annotations.
[LG-70] ATTUNE: Agent ic Throughput Tuning for LLM Training
链接: https://arxiv.org/abs/2610.17528
作者: Leo Fan,Hongyang Wang,Wenbo Chen,Jingguo Liu,Xian Li,Xuan Wang,Ruochen Wen,Yiming Yan,Hao Wu,Zheshen Wang,Benoit Dumoulin
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注: "21 pages, 2 figures, 14 tables
Abstract:Training recipes are tuned by hand for one model, cluster and workload, and every new model, accelerator, framework version or dataset means tuning again. Automated tuners search a fixed space of knobs and check only that a job runs, yet a configuration can run to completion while it silently changes training. ATTUNE is a single coding agent that tunes any working recipe for the workload and stack at hand, unattended, within a GPU-hour budget the user sets. It diagnoses each run and changes one lever at a time, guided by a catalog of lessons from past campaigns, and ships a recipe only if it passes a correctness gate: a configuration check against objective-changing settings and known stack defects, and a convergence check of its loss curve against the baseline’s. On four held-out models at 64-128 GPUs, ATTUNE raises mean throughput over the strongest public recipe run on our stack by 6.7% on Llama-3.1-405B (best campaign 11.9%) and 33.9% on GLM-4.5 (54.7%), whose public recipe was written for H100. It keeps Mixtral-8x22B’s recipe, where nothing it finds clears the noise band. Under a common GPU-hour budget it beats OptiCo, Bayesian optimization and random search on GLM-4.5 and MiniMax-M2; on GLM-4.5 its mean gain is more than four times the best alternative’s at less than half the GPU-hours, and on both models its tuning pays for itself within a day of training. The gate rejects no legitimate run and catches 16 of 24 faulty runs, missing only ones that barely move the 20-step loss. Ablations tie its best campaigns to one of its search rules: retrying an out-of-memory layout with recomputation. We release ATTUNE’s code and TrainingThroughputBench, a leaderboard of the fastest known recipes for MoE models up to 744B parameters on up to 256 GPUs, each with its throughput, cost per million tokens and a passed correctness check, at this https URL.
[LG-71] Merit: Making every modality earn its contribution in multimodal attention decoding
链接: https://arxiv.org/abs/2610.17507
作者: Ali Alavi,Donald S. Williamson
类目: Machine Learning (cs.LG)
*备注:
Abstract:More than 1.5 billion people have hearing loss. When several people talk at once, hearing aids amplify every voice alike. Auditory attention decoding (AAD) aims to solve this by reading a listener’s brain signals to find the attended talker, letting the device amplify only them. A real device will not see the brain alone; smart glasses and earbuds also track eyes, head, and scene. These behavioral signals help, as people often look at who they listen to. However, they show where a listener looks, not whom they hear. A decoder relying heavily on them fails when a listener cannot move their head or listens without looking - exactly when a brain-based device is needed. A multimodal decoder must use these signals without letting them replace the electroencephalogram (EEG). Accuracy alone cannot show whether it does. We present MERIT, a decoder and evaluation method asking: how much accuracy does each input earn? MERIT shuffles each input stream across test examples, measures the accuracy drop, and splits the result across streams using Shapley values. It also blocks two shortcuts: candidate audio is checked before training so it cannot reveal the answer, and the scoring function cannot beat chance if it ignores the brain. On a public dataset with EEG, eye tracking, head motion, and scene video, MERIT improves on the published benchmark, reaching 0.82 four-way accuracy within listeners and 0.74 on unseen listeners (chance is 0.25). The measurement reveals what accuracy hides: adding behavioral streams raises overall accuracy but almost halves the brain’s share of it. On two other datasets, the same tests find a label imbalance in a widely used benchmark and flag results that accuracy alone would report as brain decoding.
[LG-72] Longitudinal Learner Participation Profiling in Online Learning: Heterogeneous Consensus Clustering under Unequal Observation Windows
链接: https://arxiv.org/abs/2610.17494
作者: David Otoo-Arthur,Terence L. van Zyl
类目: Machine Learning (cs.LG); Computers and Society (cs.CY)
*备注: Accepted for presentation at SACAIR 2026 and publication in the conference proceedings
Abstract:Learning management system traces are clustered to identify participation profiles, but unequal observation windows may cause profiles to reflect duration rather than participation alone. Using 16,448 complete learner–course sequences from 4,599 learners across 152 courses at a South African public university, we evaluated eight participation measures. All remained related to observation-window length even when expressed as rates or shares. Course-grouped adjustment reduced these relationships. Five retained measures were evaluated using K-means, Gaussian mixtures, Ward agglomerative clustering and K-medoids, with standardised measures and principal-component summaries. Evidence supported a broad two-profile grouping and a more detailed adjusted four-profile consensus: Gradually declining, Early-concentrated, Cyclical variable and Sustained active. Profile membership accounted for 2.87% of the variation in duration in the primary consensus, versus 38.18% using original measures. Mean assignment support was 0.739; 8.35% of assignments fell below 0.50, with greatest uncertainty in Sustained active. Two week-by-week sequence analyses agreed strongly with each other but only modestly with the four-profile consensus, indicating that the profiles combined multiple aspects of participation rather than distinct sequence patterns. Profiling should therefore assess duration effects, compare clustering methods and participation summaries, and report assignment consistency.
[LG-73] A Time-Synchronous Prequential Evaluation Framework for Next-Week Student Disengagement Prediction from Moodle LMS Logs
链接: https://arxiv.org/abs/2610.17490
作者: David Otoo-Arthur,Terence L. van Zyl
类目: Machine Learning (cs.LG)
*备注: Accepted for publication in the proceedings of SACAIR 2026 (Springer CCIS)
Abstract:Retrospective evaluation of early-warning models may not reflect information available when predictions are issued and outcomes become observable. We developed a time-synchronous prequential framework for online prediction of next-week student disengagement, operationalised as inactivity onset, from Moodle logs. Ordered by calendar time, models were updated only with available outcomes and remained unchanged for all predictions at each prediction cutoff; development, calibration and evaluation were chronologically separate. We compared stochastic gradient descent logistic regression (SGD-LR), online sequential extreme learning machine (OSELM) and echo state network (ESN) across predictive performance, fixed, frozen and rolling threshold policies, calendar-time variation and four behavioural feature sets. Held-out evaluation comprised 49,329 predictions with 37.68% inactivity-onset prevalence. Discrimination was similar (average precision 0.665-0.679; ROC-AUC 0.755-0.774). OSELM had the highest recall, SGD-LR the highest precision and specificity, and ESN the highest Matthews correlation coefficient and balanced accuracy. Lower thresholds increased recall and alert volume, whereas rolling adaptation did not improve MCC over frozen thresholds. Weekly performance and calibrated-probability distributions varied without sustained deterioration. During model selection, cumulative-and-temporal features achieved the highest individual-set AP, with modest gains from combining all feature groups. These findings indicate that streaming early-warning evaluation should consider calibration, threshold-dependent performance, alert burden and calendar-time variation alongside aggregate discrimination.
[LG-74] Structure-preserving fidelity diagnostics for partial differential equations
链接: https://arxiv.org/abs/2610.17483
作者: Kyriakos Flouris
类目: Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
*备注: Preprint
Abstract:Machine learning surrogates of partial differential equations (PDEs) are commonly trained on data generated by numerical solvers. Therefore, the training data obey not the PDE but the modified equation of the numerical scheme, and the test loss, computed on data from the same solver, cannot distinguish a model that learned the physics from one that learned the discretization error. Here, we show that the difference is nevertheless visible in the trained model. We propose a suite of structure-preserving fidelity diagnostics, simple scores that measure whether a learned model respects one known property of the physics, such as a conserved quantity, a symmetry or a Hamiltonian structure, without access to the training data or the governing equation. Over five systems, the diagnostics track the error with respect to the true dynamics with a rank correlation of up to 0.94 , while the test loss is uncorrelated or anti-correlated with it. For the Burgers, Korteweg-de Vries and Navier-Stokes equations, the diagnostics separate numerical schemes that the test loss ranks as equivalent, including schemes of the same order. order.
[LG-75] From Orbit Matching to Docking: A Hierarchical Reinforcement-Learning Approach to Multiphase Rendezvous Trajectory Design
链接: https://arxiv.org/abs/2610.17468
作者: Mariko Storey-Matsutani,Anatoly Zlotnik,Lee Holguin,Zigfried Hampel-Arias,Richard Linares
类目: Robotics (cs.RO); Machine Learning (cs.LG); Systems and Control (eess.SY)
*备注:
Abstract:As the complexity of space missions increases, autonomous spacecraft guidance must accommodate transitions between mission phases with distinct dynamics, objectives, and constraints. Rendezvous and proximity operations (RPO) are particularly challenging because they span multiple operating regimes, from orbital-scale matching and far-range approach to close-range maneuvering and terminal docking. Existing reinforcement-learning approaches generally address only a subset of this sequence or rely on manually defined switching conditions between controllers. This study presents a hierarchical reinforcement-learning framework for autonomous, three-degree-of-freedom RPO that spans the complete trajectory from orbit matching to docking. Five specialized policies are trained for orbit matching, far-range approach, mid-range approach, close-range approach, and docking. A learned high-level manager policy then selects among these phase-specific controllers based on the evolving spacecraft state. The hierarchical approach is evaluated over the full end-to-end rendezvous sequence and compared with a fixed sequential-invocation heuristic to assess trajectory performance and transition behavior. Training and inference metrics are also analyzed for the individual phase policies and the complete hierarchical system.
[LG-76] BrowseSafe: Benchmarking Prompt Injection Detection Across Web Contexts Languages and Browser Views
链接: https://arxiv.org/abs/2610.17457
作者: Yuetian Chen,Kaiyuan Zhang,Yuntao Du,Thomas Li,Mark Tenenholtz,Kyle Polley,Jerry Ma,Denis Yarats,Ninghui Li
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注:
Abstract:Browser agents need to make decisions and execute actions on behalf of users while consuming untrusted webpages, creating opportunities for malicious instructions embedded in webpages to redirect legitimate tasks. We introduce BrowseSafe++, which extends BrowseSafe from 14,719 pages to 38,675 pairs of pages. Most of its prompt injections were observed in the wild on production websites. We keep each injection unchanged and rewrite the sensitive parts of the page around it to protect privacy. Each attacked page has a clean copy of the same page as its control. Both are provided as HTML, Markdown, an accessibility tree, and a screenshot. Every pair includes a user task and its expected answer. The benchmark includes twelve types of websites, from news articles to shared documents and forums, with 24 attack types appearing in 18 different locations on a page. We use these paired pages to train BrowseSafe-Guard, two 0.8B-parameter detectors for text and screenshots. Both achieve higher F1 scores than the detector based on frontier models, while keeping false-positive rates on benign pages and inference cost low. BrowseSafe-Adapt revises benchmark injections from browser-agent feedback, and attacks that succeed on only a few percent of tasks when replayed directly reach up to 95.41% success once adapted.
[LG-77] Effect of Spherical Parameter Space on Number of Non-Forgetting Tasks and Capacity of Continual Learning
链接: https://arxiv.org/abs/2610.17453
作者: Nazanin Mirhosseini
类目: Machine Learning (cs.LG); Information Theory (cs.IT); Machine Learning (stat.ML)
*备注:
Abstract:We propose a continuous learning formulation for T non-forgetting tasks with the spherical parameter space \mathbbS^d-1(\sqrtd) in asymptotic regime d\rightarrow \infty . Each task i has an optimum \bm\Theta_i , minimizing its corresponding loss function R_i(\bmw) on sphere. We reduce each task i to an acceptable loss region modeled by a geodesic ball B_\bm\Theta_i(r_i(d)) on \mathbbS^d-1(\sqrtd) centered at \bm\Theta_i with radius r_i(d) . To model task dependencies (similarities), task optima \bm\Theta_i are distributed by a Neyman-Scott clustered point process \Phi . Cluster centers \bmY_j_j=1^N_p form a homogeneous Poisson point process on sphere, while intra-cluster task optima follow a von Mises-Fisher distribution with mean \bmY_j and concentration \kappa . Each task t\in \1, …, T\ learning algorithm is considered to be a channel Q_t(\mathrmd\bmw_t | \bmw_t-1, \mathcalK_t-1(\bm\theta_1:t-1), \bm\theta_t) , where \bmW_t is the learned parameter of task t and \mathcalK_t-1(\bm\Theta_1:t-1)=\bigcap_i=1^t-1B_\bm\Theta_i(r_i(d)) . Since in continual learning, the learner does not need to reach \bm\Theta_t exactly to successfully learn task t , to avoid forgetting as well as reliable learning, it only needs for the learned parameter vector \bmW_t to enter B_\bm\Theta_t(r_t(d)) \cap \mathcalK_t-1(\bm\Theta_1:t-1) . Using this model, we derive an upper bound on T ensuring \mathbbP[\mathcalK_T(\bm\Theta_1:T) \neq \emptyset] \rightarrow 1 as d\rightarrow \infty and \kappa/d \rightarrow \alpha , where \alpha0 is a constant. Finally, for T tasks sampled uniformly from \Phi , we derive an upper bound for total information theoretic capacity under the non-forgetting constraint for each task.
[LG-78] On Sleep-Wake Disorder Identification from Polysomnographic Representations
链接: https://arxiv.org/abs/2610.17451
作者: Alvise Dei Rossi,Luigi Fiorillo,Giuliana Monachino,Alessia Calzoni,Julia van der Meer,Markus H. Schmidt,Claudio L.A. Bassetti,Silvia Santini,Francesca Faraci
类目: Machine Learning (cs.LG)
*备注:
Abstract:Self-supervised polysomnography foundation models (PSG FMs) aim to capture clinically useful information beyond sleep staging, yet are rarely compared with sleep-staging encoders pretrained at comparable scale and modality coverage. Using a common whole-night, missingness-aware state-space downstream modeling framework, we compared frozen representations from prominent large-scale, multi-cohort sleep-stage-supervised encoders and self-supervised PSG FMs for multilabel identification of six sleep-wake and circadian disorder groups in 7,906 PSGs from an unseen registry. Although PSG FMs outperformed an EEG-only, single-cohort supervised baseline, this ordering reversed when supervised pretraining approached comparable scale and modality coverage. Transfer learning from neural sleep scorers achieved higher overall performance while exhibiting greater downstream data efficiency. These findings underscore the need for scale- and modality-matched pretrained supervised baselines when assessing PSG FMs’ downstream potential.
[LG-79] Formal-Verification-Guided Bayesian Optimization for Non-Intrusive Last-Mile Coverag e Closure
链接: https://arxiv.org/abs/2610.17449
作者: Chengjia Liu,Priyadarsi Mishra,David Kebo Houngninou,Michael Quinn,Aakash Tyagi,Jiang Hu
类目: Hardware Architecture (cs.AR); Machine Learning (cs.LG)
*备注: 21 pages, 3 figures, 7 tables. Accepted for publication in ACM Transactions on Design Automation of Electronic Systems (TODAES)
Abstract:Last-mile functional coverage closure for modern chip designs remains costly because a mature UVM environment exposes many controls while direct hits on the remaining bins are sparse. We present a non-intrusive workflow that closes pre-defined bins in an existing, human-authored functional coverage model; it does not synthesize verification objectives from the design specification. The workflow first uses abstract-syntax-tree (AST) parsing and cone-of-influence (COI) tracing to map each residual target to existing testbench knobs. Under a matched 20-simulation budget, Bayesian optimization (BO) is compared with continued constrained-random verification (CRV) and random search over the same AST/COI-reduced space. Formal reachability then removes only targets supported by complete unreachability proofs, after which retained targets are grouped by formal difficulty and processed by group-specific BO under a second matched 20-simulation budget. Cleanup and grouping never add numerator bins; only simulator coverage artifacts do. On OpenTitan AES, first-pass reduced-space BO reaches 241/271 bins; formal cleanup removes two denominator bins; the integrated post-cleanup stage - difficulty grouping, policy routing, and group-specific BO - adds 20 simulator-confirmed hits, yielding 261/271 raw coverage (96.31%) and 261/269 effective coverage (97.03%). On Entropy Source, first-pass reduced-space BO reaches 2117/2277 bins; formal cleanup removes ten denominator bins; the integrated post-cleanup stage - difficulty grouping, policy routing, and group-specific BO - adds 93 simulator-confirmed hits, yielding 2210/2277 raw coverage (97.06%) and 2210/2267 effective coverage (97.49%). These results characterize the integrated diagnosis-and-search workflow; the present evaluation does not isolate the causal contributions of difficulty classification, policy routing, and post-cleanup BO.
[LG-80] Monarch-Muon: Structure-Aware Muon for Fast and Memory-Efficient Pretraining
链接: https://arxiv.org/abs/2610.17429
作者: Alexander Miasnikov,Ruslan Kabirov,Gleb Molodtsov,Dmitrii Vlashchuk,Vladimir Aletov,Nail Bashirov,Daniil Tikhonov,Maxim Krutikov,Aleksandr Shestakov,Aleksandr Beznosikov
类目: Machine Learning (cs.LG)
*备注:
Abstract:Training large language models is constrained by both wall-clock time and memory footprint, and recent matrix-aware optimizers sharpen this tradeoff: Muon can improve optimization but its dense polar update remains expensive. We propose Monarch-Muon, a structured matrix-aware optimizer for weights parameterized by Monarch factors. The resulting structure reduces weights, gradients, and optimizer state, and exposes small independent update subproblems in place of a single dense orthogonalization; we formalize this through a blockwise trust-region interpretation. In quality runs, Monarch-Muon improves over Monarch-AdamW at fixed structure: 3.06 vs. 3.09 in the dense 257M model and 2.68 vs. 2.77 in the 1B MoE model; the two-block MoE variant also matches dense AdamW while using 15% fewer active and 25.4% fewer total parameters. In 6.89B-parameter systems measurements, Monarch-Muon reduces peak memory from 68.9 to 52.9 GB, Muon optimizer state from 14.60 to 6.77 GB, and the Muon optimizer step from 1288 to 304 ms.
[LG-81] When Sparsity Breaks Connectivity: Functional Reachability and Guided Topology Dynamics in Extreme Sparse Training
链接: https://arxiv.org/abs/2610.17425
作者: Mohammad Pasande,Mateusz Jaworski,Jeroen Rook,Elena Mocanu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Dynamic Sparse Training (DST) periodically prunes and regrows connections, inducing a discrete dynamical mechanism over network topologies. At extreme sparsity ( \ge 99% ), this dynamic tends to destroy input-output information pathways, and performance with standard magnitude/random updates degrades sharply. We formalize forward, backward, and functional reachability on layered sparse graphs, derive closed-form expectations for reachable neurons and edges, and show that Monte Carlo realizations match the theory. Guided by this reachability-based structure, we bias pruning toward non-functional edges and regrowth toward functionally consistent candidates. We introduce this reachability-driven modification of topology evolution as Reachability-guided in DST, referred to as guided DST. On CIFAR-10/100 and Tiny ImageNet, guided DST recovers large accuracy gaps versus unguided DST (up to +14.4 points at 99.9% sparsity on CIFAR-100), and on MLP-Mixer guided DST remains trainable at 99.99% sparsity while unguided random regrowth degrades sharply. These results cast extreme-sparse training as a connectivity-preserving dynamical process. Our implementation is publicly available at this https URL.
[LG-82] Position: Beyond Training Time Test-Time Coordination is Essential for Cooperative MARL NEURIPS
链接: https://arxiv.org/abs/2610.17409
作者: Dongsu Lee,Sooraj Sathish,Joonkyung Kim,Woojun Kim,Amy Zhang
类目: Machine Learning (cs.LG)
*备注: NeurIPS Position Track 2026
Abstract:We argue that cooperative multi-agent reinforcement learning (MARL) should treat test-time coordination as part of the problem definition rather than a deployment artifact. Our focus is on long-horizon, centrally monitored cooperative systems such as warehouse robotics, automated storage-and-retrieval systems, and cloud-resource scheduling, where teams share a reward, operate under shifting objectives, and already run with global telemetry at deployment. The dominant paradigm in this regime, centralized training with decentralized execution (CTDE), uses global information only during training and provides no coordination channel at execution. This is sufficient when agents are weakly coupled or the environment is stationary, but it fails when the team must realign after conditions change. The limitation is structural rather than algorithmic, and more training cannot add a channel that does not exist at execution time. We call this the recoverability gap. To address it, we name the underexplored design space of intermittent test-time coordination as centralized training with mixed execution (CTME), in which a coordinator observes global state and emits high-level signals while agents act locally. CTME is not a particular learning algorithm, but a cooperative MARL problem setting that makes test-time information structure explicit. It specifies team-level information, high-level signals, and update frequency after training, with CTDE recovered as a limit case. We do not argue that CTDE is obsolete. We argue that removing the test-time channel is a modeling choice, not a necessity, and the wrong default when global telemetry is available and teams must recover from drift.
[LG-83] JustOPD: On-Policy Distillation from Autoregressive to Diffusion Language Models
链接: https://arxiv.org/abs/2610.17397
作者: Zikun Zhang,Jiayuan Sheng,David D. Yao,Wenpin Tang
类目: Machine Learning (cs.LG)
*备注: 24 Pages, 1 figure
Abstract:On-policy distillation (OPD) is a post-training approach for large language models (LLMs) that enables a student model to learn a teacher model’s distribution from sequences sampled from the student’s own policy. Despite its promising capability, OPD for diffusion language models (DLMs) remains underexplored and existing studies primarily focuses on DLM-to-DLM distillation. The limited availability of strong teacher DLMs and the flexible generation order of DLMs impose additional technical challenges as well. However, it remains unclear how to effectively distill knowledge from a strong autoregressive language model (ARM) into a weaker DLM, given the mismatch between the left-to-right decoding of ARMs and the flexible decoding order of DLMs. Inspired by Ni et al. (2026), we propose JustOPD, an ARM-to-DLM on-policy distillation method that uses autoregressive rollouts during training to align the student’s generation process with the ARM teacher’s left-to-right decoding. This alignment enables effective distillation from ARM teachers while preserving the DLM’s flexible generation-order capability at inference time. Experiments using Qwen3-14B as the teacher and Dream-7B as the student demonstrate that JustOPD consistently improves mathematical reasoning performance over the base DLM across multiple divergence choices.
[LG-84] A Method for Exact Enforcement of Boundary Conditions on General Curved Triangular Domains: Algorithm and Application to Scientific Machine Learning
链接: https://arxiv.org/abs/2610.17392
作者: Suchuan Dong,Yuchuan Zhang
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
*备注: 47 pages, 10 figures, 6 tables
Abstract:We present a method for enforcing exactly Dirichlet, Neumann, and Robin conditions on general triangular domains with arbitrarily curved boundaries. The algorithmic constructions developed here complements those in a previous work for general quadrilateral domains. Our method combines (i) the exact mapping between a general curved triangular domain and the standard triangle, (ii) TFC constrained expressions, and (iii) Barnhill transfinite interpolations, to provide explicit closed-form trial ansatze that mathematically satisfy the imposed boundary conditions exactly. Corner compatibility induced by the boundary conditions is pivotal, and enforcing exactly the imposed boundary conditions hinges on the exact enforcement of the induced compatibility constraints. The corner compatibility associated with the acute vertices of the standard triangle and the coupling between gradient components when the hypotenuse corresponds to a Neumann/Robin boundary pose particular challenges. Two strategies are critical to resolving these issues, by reformulating conditional corner constraints into equivalent unconditional forms and by introducing a separate TFC-style auxiliary edge construction together with an associated linear solve on the hypotenuse. We analyze the corner compatibility constraints and present formulations for the general form of functions that exactly satisfy the imposed Dirichlet, Neumann, or Robin conditions on curved triangular domains. The proposed method has been implemented using extreme learning machine type randomized neural networks for solving PDE problems. We present ample numerical experiments with linear/nonlinear PDE problems on a variety of domains with complex boundary geometries. Numerical results show that the current method has enforced the Dirichlet, Neumann, and Robin conditions on curved domain boundaries to the machine precision.
[LG-85] Measure Neural Mappings
链接: https://arxiv.org/abs/2610.17384
作者: Ricardo Baptista,Edoardo Calvello,Bohan Chen,Andrew Stuart,Urbain Vaes
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 65 pages, 6 figures, 7 tables
Abstract:We introduce and define measure neural mappings (MNMs), which combine measure-dependent neural transport with ensemble consistency for learning operators on probability measures. Ensemble consistency requires outputs from growing ensembles sampled from an input law to converge to the specified population output in probability. We establish quantitative continuum limits for attention on noncompact laws under strong sub-Gaussian and tail-integrability assumptions. These laws accommodate unbounded state spaces in scientific machine learning, including Bayesian inference and uncertainty quantification. For fixed self-attention transformers, the expected output Wasserstein error is bounded by the empirical input Wasserstein error plus an additional N^-1/2 attention term. These self-attention transformers are MNMs under the stated assumptions: pre-norm architectures follow from compact stability after normalization, while post-norm architectures are covered under the stated tail conditions. We prove corresponding cross-attention estimates and extend the results to multihead attention and grouped transformer evaluation, with convergence as the group size grows. For Set Transformers viewed as measure-to-vector mappings, we obtain an N^-1/2 bound on the expected Euclidean output error under the stated assumptions. Experiments on seven operators in dimensions d=1,2,3,5,10 support the predicted particle-count dependence, using one-dimensional Wasserstein computations and Sinkhorn approximations in higher dimensions. The observed Wasserstein and sliced Wasserstein decay rates are close to their respective theoretical sampling benchmarks.
[LG-86] Variable Input Masked Autoencoders for Wrist-Worn Accelerometry
链接: https://arxiv.org/abs/2610.17383
作者: Harish Haresamudram,Wanting Mao,Aditya Radhakrishnan,Maxwell A. Xu,James M. Rehg
类目: Machine Learning (cs.LG)
*备注:
Abstract:Large-scale repositories of wearable sensor data have enabled the development of foundation models for human activities and behavior. However, existing models are pretrained using fixed input durations and sampling rates, despite substantial heterogeneity in recording conditions, downstream tasks, and deployment constraints. We introduce VarMAE, a variable input masked autoencoder that learns a single representation across sampling rates and temporal contexts. During pretraining we construct two views of the same signal: a long duration high frequency view, and a short duration lower frequency view, obtained via random cropping and downsampling. We perform masked reconstruction within each view and introduce a cross-resolution masked reconstruction objective that uses visible tokens from the long view to predict masked tokens from the short view. Cropping and downsampling impose complementary information bottlenecks, encouraging the model to learn representations that remain effective across input durations and sampling rate changes. We pretrain VarMAE on the large-scale NHANES dataset, producing a single pretrained encoder matching or outperforming controlled foundation model baselines on eight benchmark datasets. With only 3.82M parameters, VarMAE is parameter efficient and shows that a single foundation model can achieve broad transfer capabilities across sensing configurations.
[LG-87] Robustly Reliable Learning for Regression
链接: https://arxiv.org/abs/2610.17381
作者: Avrim Blum,Anh Tuan Nguyen
类目: Machine Learning (cs.LG)
*备注: 36 pages, 1 figure
Abstract:We introduce the problem of robustly reliable learning for the regression setting in adversarial environments. Given a test point x_test , a training dataset S of n data points, a corruption budget b \ll n , and a real-valued hypothesis class \mathcal H , our goal is to output the optimal prediction range I^_x_test(S, b) , such that any L -Lipschitz predictor consistent with all but at most b points in S must predict a value h(x_test) \in I . We then instantiate the analysis for L -Lipschitz hypothesis classes, a natural and widely studied class in the regression setting, focusing on two key scenarios. First, for the restricted case where the corrupted dataset is guaranteed to be L -Lipschitz consistent, we propose a simple and efficient b^th -best heuristic algorithm and prove its correctness in computing the tightest possible range I^_x_test(S, b) . We then establish its computational complexity of O(n) and show that it is computationally optimal in this scenario. Building on this foundation, we address the general scenario where the dataset S is no longer L -Lipschitz consistent. We demonstrate that computing the optimal range in this setting reduces to analyzing a violation graph, which models the pairwise L -Lipschitz violations among data points. By leveraging the structural properties of this graph’s kernel, we propose a Fixed-Parameter Tractable (FPT) algorithm that computes the exact optimal range in O(n^2 + b^4) time. Furthermore, under the Strong Exponential Time Hypothesis (SETH), we establish a computational lower bound of \Omega(n^2 - o(1)) , proving that our algorithm is computationally optimal. Finally, we discuss open directions for the problem of robustly reliable learning for regression.
[LG-88] HarnessTTT: Learn to Optimize Harness at Test-Time for Scientific Discovery
链接: https://arxiv.org/abs/2610.17370
作者: Yingzi Ma,Yulong Cao,Nanxi Li,Fang Wu,Yizhe Zhang,James Zou,Yejin Choi,Chaowei Xiao
类目: Machine Learning (cs.LG)
*备注: 31 pages, 9 figures, 10 tables. Project page: this https URL
Abstract:Scientific discovery with language models requires finding the best valid solution within a finite computational budget. We study how feedback from this search can improve the procedure used to conduct it. We introduce HarnessTTT, a test-time learning framework that trains a harness proposer while keeping the executor frozen. The proposer generates executable harnesses comprising system prompts, reusable skills, tools, middleware, and execution controls. Each harness guides a budgeted discovery rollout, whose outcome provides feedback for updating the proposer. A discovery-oriented policy-gradient objective combines gap-normalized rewards, leave-one-out advantages (RLOO), and max-sharpening to favor high-performing proposals. Generated components are validated before execution, and the search retains the best valid solution and selected harnesses across rounds. Across fourteen discovery tasks in mathematics, systems, GPU kernels, heuristic programming, astrodynamics, and biology, HarnessTTT with a frozen 9B executor improves on OpenEvolve with the same executor on all eleven benchmark tasks (e.g., from 1.17 to 2.64 on circle packing), improves on the discovery-fine-tuned Finch-9B by up to 92%, holds the best open-weight sub-10B result on nine, and reaches 0.380919 on Erdős minimum overlap, better than the best human reference of 0.380927. Scaling the proposer and frozen executor to 27B improves on eleven of fourteen tasks, and transferring the learned harnesses to GPT-5.5 matches or exceeds the strongest reported system of any backbone on nine of fourteen tasks. Together, these results show that harness search can itself be learned at test time: a trained proposer lets a small frozen executor search effectively, and the harnesses it finds carry over to stronger executors.
[LG-89] COMPASS: Membership Inference Attack Against Vision Large Language Models via Adaptive Noise Calibration
链接: https://arxiv.org/abs/2610.17332
作者: Damien Lo,Hong Kyu Lee,Ruixuan Liu,Li Xiong
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
*备注:
Abstract:Vision large language models (VLLMs) are deployed in privacy-sensitive domains, raising training-data exposure concerns. Existing gray-box membership inference attacks (MIAs) commonly rely on entropy statistics or generated descriptions, requiring costly generation with limited membership separation. Perturbation-based methods instead measure response changes but focus on large perturbations, leaving VLLM membership behavior under smaller perturbations poorly understood. We systematically characterize VLLM membership behavior under controlled Gaussian perturbations across noise levels at fine resolution. We find that optimal perturbation magnitude and membership-signal direction vary across model-dataset pairs. In many settings, a previously unstudied low-noise regime yields stronger separation than large perturbations, with members exhibiting lower divergence than non-members, reversing the conventional score direction. Other settings retain the conventional direction, demonstrating that neither a fixed perturbation magnitude nor a predetermined score orientation is universally reliable. We propose COMPASS, an adaptive gray-box MIA that calibrates both the perturbation magnitude and membership-score direction using a loosely distribution-matched, known non-member reference set. We instantiate COMPASS using Renyi-normalized KL divergence and Renyi divergence, with max-k aggregation over the most informative image-logit positions. Across LLaVA v1.5-7B, MiniGPT-4, and HuluMed-32B, spanning pretraining, instruction-tuning, and medical-image datasets, COMPASS outperforms baselines. On general-domain data, COMPASS achieves AUCs up to 0.88. On the distribution-matched medical evaluation, it reaches 0.64 AUC, while the strongest baseline remains at chance level. These results establish full-range adaptive perturbation analysis as an effective approach to VLLM membership inference.
[LG-90] GPU Goldmines: Tunable WebGPU Kernels for Edge LLM Inference
链接: https://arxiv.org/abs/2610.17321
作者: Rithik Sharma,Reese Levine,Zheyuan Chen,Ananthajit Srikanth,Yanwen Xu,Tyler Sorensen
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG); Performance (cs.PF)
*备注: 11 pages, 5 figures, 5 tables
Abstract:As demand for LLM inference grows and consumer GPUs grow more capable, edge deployment is becoming increasingly attractive. WebGPU is compelling in this setting because it offers cross-platform portability and browser accessibility. However, these same strengths also make high performance difficult to achieve, as an implementation that performs well in one context may not perform well in another. To address this, we present GPU Goldmines, a framework for specifying, executing, and tuning WebGPU kernels. We instantiate GPU Goldmines with kernels for key LLM building blocks: matrix multiplication, matrix-vector multiplication, and flash attention. We characterize performance across diverse devices, shapes, and browsers, and use the resulting database to accelerate subsequent autotuning and configuration selection. On shapes drawn from Llama and Qwen models, our matrix multiplication kernels outperform ONNX Runtime and this http URL WebGPU kernels by 7.0x and 1.41x geometric mean, respectively, and deliver end-to-end improvements of 1.48x in the prefill stage of this http URL for chat workloads.
[LG-91] Near-Optimal Regret Bounds for Monotone Stochastic Optimization under Semi-Bandit Feedback
链接: https://arxiv.org/abs/2610.17320
作者: Tao Hu,Rohan Ghuge
类目: Machine Learning (cs.LG); Data Structures and Algorithms (cs.DS); Machine Learning (stat.ML)
*备注: 68 pages, 20 figures, 1 table
Abstract:Stochastic optimization models sequential decision-making under uncertainty, typically assuming known input distributions. When these distributions are unknown, \citetagarwal2026semibandit give an online-to-offline reduction for monotone stochastic optimization under semi-bandit feedback, where only probed variables are observed. Its expected regret relative to the offline guarantee is \widetilde O(n\sqrt T) , where n is the number of variables and T the horizon. For optimal stopping and Pandora’s box, \citetliu2025improved achieve \widetilde O(\sqrtnT) using problem-specific algorithms. \emphCan a general online-to-offline reduction achieve this improved rate? We give such a reduction for strongly monotone stochastic optimization, answering the corresponding open question. Strong monotonicity is defined relative to the chosen offline algorithm. Our oracle-efficient reduction achieves \widetilde O(\sqrtnT) expected regret while preserving approximation guarantees for oracles satisfying the same condition. It combines Bernstein-type optimistic distribution estimates with a Hellinger stability bound for adaptive policy transcripts, aggregating squared marginal errors weighted by the executed policy’s probing probabilities. The rate is optimal up to logarithmic factors for T\ge n , even under full feedback. Under monotonicity and uniformly bounded coordinate-wise variation, the same algorithm achieves \widetilde O(n^3/4\sqrt T) expected approximation regret, improving the previous linear dependence on n . The exponent 3/4 is sharp for its ordinary regret over instances and exact oracle choices. Our framework applies to prophet inequality, Pandora’s box, best-of- K selection, stochastic knapsack, stochastic orienteering, and stochastic submodular maximization. Comments: 68 pages, 20 figures, 1 table Subjects: Machine Learning (cs.LG); Data Structures and Algorithms (cs.DS); Machine Learning (stat.ML) MSC classes: 68W27, 68Q32, 90C15, 68W25 Cite as: arXiv:2610.17320 [cs.LG] (or arXiv:2610.17320v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.17320 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-92] A Geometrical Approach to Constructing Neural Networks for Perfect Training-set Classification
链接: https://arxiv.org/abs/2610.17312
作者: Akos Hajnal
类目: Machine Learning (cs.LG)
*备注: 8 pages, 1 figure, 2 tables
Abstract:This paper proposes a method for constructing a neural network that achieves perfect classification of all samples in a given training set, without taking regularization into account. The network is constructed on-demand, with its layer sizes and neuron parameters determined through a geometrical analysis of the input data, in which separating hyperplanes are created and mapped to neurons with step activation functions. The proposed architecture consists of two hidden layers with distinct roles: inter-class separation and transformation of the data into a linearly separable representation. Inter-class separation requires every bi-chromatic pair of samples belonging to different classes to be separated by at least one hyperplane. We show that when this separation is achieved through a hierarchical partitioning of the data, the leaf nodes of the oblique decision tree, obtained as a byproduct, directly define the neurons of the linear separability layer, whose size is only one neuron greater than that of the inter-class separation layer. Experiments on the MNIST dataset demonstrate that the proposed method achieves perfect training-set classification, and compact models. The direct correspondence between the decision tree and network neurons enhances interpretability and explainability, while the construction procedure offers opportunities for parallel implementation and improved scalability.
[LG-93] Which Geometry on Which Layer? Measured Sign-Spectral Assignment in Mixed-Optimizer Training
链接: https://arxiv.org/abs/2610.17285
作者: Hiroki Naganuma,Laura Gomezjurado,Mahdi Ghaznavi,Ganesh Talluri,Atsushi Nitanda,Seng Pei Liew,Ryuichiro Hataya,Ioannis Mitliagkas
类目: Machine Learning (cs.LG)
*备注: 35 pages, including appendices
Abstract:Modern neural-network training uses mixed recipes, assigning AdamW-like or sign-based updates to embedding and output layers and spectral updates to hidden matrices. These assignments are usually fixed by layer type, without a clear criterion for selecting the appropriate geometry. We turn sign-versus-spectral assignment into a measurable, layerwise decision problem. For ideal memoryless steps, we derive a finite-batch ratio of conservative descent-lemma lower bounds for spectral and sign updates. Its closed form separates signal efficiency from a geometry-dependent global curvature factor. In the positive-signal regime, the same smoothness assumption gives two-sided bounds that determine which step has the larger conditional expected decrease when the ratio crosses explicit noise-dependent thresholds. We then derive a local directional quadratic score from gradient alignments and Hessian-vector products and use it in Generalized-Scion (G-Scion) to control layerwise routing. In a controlled input-rank experiment, including directional curvature matches the measured same-iterate ordering in all seven aggregate settings, whereas the curvature-free proxy selects the opposite geometry. On 31M GPT-2, the output score routes the output layer from spectral warmup to sign, and G-Scion improves validation loss over fixed Scion in memoryless and matched-momentum comparisons. ViT-B/16 results further show that a spectral edge assignment can outperform fixed Scion outside language modelling. These results suggest that optimizer geometry need not be hard-coded by layer type and are a first step toward understanding how it should be assigned.
[LG-94] he Surprising Expressivity of a Single Neuron with Autoregressive Chain-of-Thought
链接: https://arxiv.org/abs/2610.17283
作者: Roey Magen,Nirmit Joshi,Gal Vardi
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:The success of autoregressive language models trained with Chain-of-Thought~(CoT) supervision raises a fundamental question: How simple can a next-token predictor be while still enabling the efficient simulation of general computation through its autoregressively generated CoT? Surprisingly, we show that a single neuron suffices. For a broad family of activations, including threshold and ReLU, a single autoregressive neuron can simulate any feedforward circuit with the same activation. Specifically, a size- s circuit on n inputs can be simulated by a neuron of dimension O(ns+s^2) using O(s^2) generation steps, with both bounds independent of circuit depth and tight under natural assumptions. For threshold activations, this improves upon the simulation of Joshi et al. (COLT 2025), whose bounds scale exponentially with circuit depth. Consequently, autoregressive threshold neurons can simulate polynomial-time Turing machines, while remaining efficiently learnable from CoT supervision. Our main expressivity result relies on a novel connection to Sidon sets from additive combinatorics. As an experimental illustration, we efficiently train a single autoregressive ReLU neuron to mimic a neural network teacher with nontrivial test accuracy on CIFAR-10.
[LG-95] Protocol Choices Matter More Than the Aggregation Algorithm: Federated Learning for In-Hospital Mortality Prediction in Care-Unit and Multi-Hospital Federations
链接: https://arxiv.org/abs/2610.17265
作者: Rodrigo Tertulino,Ricardo Almeida,Laércio Alencar
类目: Machine Learning (cs.LG); Computers and Society (cs.CY); Distributed, Parallel, and Cluster Computing (cs.DC)
*备注: 23 pages, 6 figures, 10 tables. Code: this https URL
Abstract:Federated learning allows hospitals to train a shared mortality model without pooling their records, but published comparisons of aggregation strategies rarely control the evaluation protocol. This study compared seven strategies (FedAvg, FedProx, SCAFFOLD, FedAdagrad, FedAdam, FedYogi and FedCluster) on two federations built with one pipeline: 12 care units of a hospital in MIMIC-IV and 49 hospitals in eICU-CRD. Each strategy was tuned and selected on validation data by the same rule, and every comparison was repeated across 10 seeds with Friedman and Holm-corrected tests. SCAFFOLD ranked first in both federations (AUC-ROC 0.886 and 0.819) and in 17 of 18 variants, but its margin over the runner-up never exceeded 0.008, and a control sharing its local optimizer attributed most of its margin over FedAvg to drift correction. On the multi-hospital federation, federated training matched centralized training. In the care-unit federation, three protocol choices changed discrimination by more than the choice of strategy did: discharge-coded predictors (+0.038 in AUC-ROC), client-level SMOTE (up to -0.074), and evaluating the last instead of the best validation round (up to -0.084). Platt scaling on validation data repaired the risks distorted by imbalance corrections without loss of discrimination. Studies should therefore fix the protocol before ranking algorithms.
[LG-96] Value-Preserving Action Quantization
链接: https://arxiv.org/abs/2610.17252
作者: Max Muchen Sun,Jueun Kwon,Benjamin Riviere,Todd Murphey
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: 8 pages, 6 figures
Abstract:Search-based policy learning, such as Monte Carlo tree search (MCTS), offers the flexibility to trade runtime computation for data efficiency and adaptation to novel environments. For robot learning, however, applying search requires discretizing the continuous action space and associated dynamics, typically through action quantization. While reconstruction-based quantization is sufficient for representing the continuous action space alone, policy learning additionally requires the quantized actions to preserve their effects on future states and values. We show that such requirement is not necessarily captured by action reconstruction error. Instead, we prove that the error in the value function resulting from action quantization is tightly bounded by the reconstruction error of state reachability under the quantized actions. Based on these results, we propose value-preserving action quantization (VAQ), which quantizes actions by preserving state reachability. We further combine VAQ with MCTS (VAQ-MCTS) as a value-preserving, search-based policy learning framework for behavioral cloning. Using a wide range of object pushing tasks as a controlled benchmark, we evaluate VAQ-MCTS against state-of-the-art behavioral cloning methods. VAQ-MCTS matches or exceeds their performance under standard settings, while substantially improving data efficiency and robustness to heterogeneous demonstrations, while enabling zero-shot adaptation to novel environments. Hardware experiments further demonstrate the effectiveness of VAQ-MCTS, completing all 30 real-world object pushing trials with only 10 minutes of demonstration data.
[LG-97] ght End-to-End Accounting for Differentially Private Synthetic Data
链接: https://arxiv.org/abs/2610.17249
作者: Georgi Ganev,Vadym Doroshenko,Arun Ganesh,Juan Felipe Gomez,Ryan McKenna,Bogdan Kulynych
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Differentially private (DP) pipelines often combine heterogeneous mechanisms, yet their guarantees are best analyzed using different notions of privacy (e.g., bounded range, zCDP, ADP). When composing them, we have to analyze the guarantees through a reduced common representation that loses accounting precision. This issue is particularly relevant in DP synthetic data generation, where data preprocessing, generation algorithms, and DP-SGD could appear within the same pipeline. We develop novel tools for tight end-to-end accounting of such compositions using privacy loss distributions (PLD), f -DP, and Gaussian DP (GDP). In particular, we derive PLD and GDP analyses of bounded-range mechanisms such as the Exponential mechanism. Building on these theoretical results, we introduce AIM-GDP, a GDP-native variant of AIM (a state-of-the-art DP synthetic data mechanism) that enables easy composability with complex pipelines via PLD, attack-aware risk calibration, and tight end-to-end auditing while consistently achieving better utility; and MixCal, an algorithm for calibrating adaptive mixed compositions using PLD accounting. Overall, our tools enable tighter end-to-end accounting and thus less noise at the same privacy level in DP synthetic data pipelines.
[LG-98] LLM -to-Flow Distillation: Teaching Natural Language Goals to Scientific Generators
链接: https://arxiv.org/abs/2610.17246
作者: Riccardo De Santi,Federico Di Gennaro,Angelo Gnazzo,Sophia Tang,Pranam Chatterjee,Fanny Yang,İsmail İlkan Ceylan,Michael M. Bronstein,Andreas Krause
类目: Machine Learning (cs.LG)
*备注:
Abstract:Generative discovery in science requires coupling high-level scientific reasoning with the ability to generate samples respecting the structure of a design space. Yet these capabilities remain largely separated: foundation LLMs encode broad chemical and biological knowledge, while domain-specific flow- and diffusion-based generators capture the priors needed to produce valid sequences and geometries. We study how to align such scientific generators to high-level natural-language discovery goals without requiring an explicit proxy reward, which is often unavailable or poorly captured by learned predictors in experimental sciences. We introduce LLM-to-Flow Distillation (LFD), where a frozen LLM acts as an adaptive teacher for a domain-specific generator. Given a goal such as ‘generate peptides with promising anticancer activity’, the teacher constructs a curriculum of intermediate sub-goals aware of the flow’s current capabilities, and progressively directs them toward the original discovery goal. At each round, the LLM judges sampled candidates under the current objective. These judgments induce preferences that are iteratively distilled into the generator through DPO. This procedure provides dense learning signals even when the initial model places negligible probability on designs satisfying the final goal, enabling adaptation toward poorly covered, out-of-distribution regions. Experiments on therapeutic peptides and small-molecule structure design demonstrate that LFD enables a new form of generative control: aligning arbitrary domain-specific scientific generators to scientific discovery tasks expressed via high-level natural-language goals.
[LG-99] MIRRA: Fast High-Fidelity Molecular Generation with Chemistry-Informed Discrete Flow Matching
链接: https://arxiv.org/abs/2610.17244
作者: Ge Sun,Juan J. de Pablo
类目: Machine Learning (cs.LG); Chemical Physics (physics.chem-ph); Computational Physics (physics.comp-ph)
*备注:
Abstract:Fast, high-fidelity molecular generation requires accurate chemical choices and efficient sampling. Chemical building blocks reduce graph size, but make each node a choice among tens of thousands of related identities. We introduce MIRRA (Molecular Identity Representation, Refresh and Assembly), a discrete flow-matching generator that learns from chemical knowledge and evaluates identity probabilities only when needed, while keeping every identity in its vocabulary available. Chemical features share information across related blocks, and learned identity corrections distinguish them in molecular context. By deciding which positions will refresh before computing their probabilities, the sampler avoids unused predictions while preserving each sampling transition. Against the strongest baselines in each task, MIRRA reduces chemical and graph distribution errors on ZINC250k by 29-31% and 53-60%, respectively, with generation up to 8.6 times faster. Property targeting yields an estimated 7.8-20.8 times more distinct novel hits per second. On a held-out polymer test set, MIRRA achieves 100% periodic validity with known chain topology, versus 57.0% for a polymer variational autoencoder. These results demonstrate fast, high-fidelity generation with the full chemical vocabulary available throughout sampling.
[LG-100] CodeLift: Your Test Suite Is Secretly a Dense SWE Curriculum
链接: https://arxiv.org/abs/2610.17224
作者: Junjielong Xu,Huachao Zhu,Alyssa Wu,Zhichao Hu,Pinjia He
类目: Machine Learning (cs.LG); Software Engineering (cs.SE)
*备注:
Abstract:The pass-all-test binary reward has become a standard recipe for training software engineering agents, but its sparsity eliminates difference of partially correct patches. The test pass-rate is a natural alternative. However, we find that it is not a reliable indicator of patch quality. A functionality being tested can be covered by either one challenging test or one hundred trivial tests, where the majority might be already satisfied pass-to-pass tests. We introduce CodeLift, a reinforcement learning method that dynamically calibrates pass-rate rewards, turning the test suite into an adaptive curriculum. At each iteration, CodeLift groups tests that kill the same patches and patches that pass the same tests, then gives more weight to test groups that fewer patch groups pass. This reduces the impact of majority simple tests and focuses more on what is still challenging. Within each rollout, CodeLift applies the same scoring rule to repository snapshots, assigning credit to useful actions even when later ones introduce regressions. Experiments on SWE benchmarks show that CodeLift consistently improves pass@1/2/4/8 over binary and pass-rate baselines. Analyses show that CodeLift provides dense, stable learning signals and unlocks the inherent curriculum in the test suite.
[LG-101] Density-Matrix Reinforcement Learning
链接: https://arxiv.org/abs/2610.17214
作者: Chiuman Ho
类目: Machine Learning (cs.LG)
*备注: 46 pages
Abstract:What information does a reinforcement learning policy omit? At each state, a standard policy stores one probability per action. It has no place to record how actions relate to each other, such as which pairs go together and which conflict. We ask when this missing structure matters, and we give exact answers. We store the policy at each state as a density matrix, a positive semidefinite, unit-trace matrix borrowed from quantum information theory. We use it here as a purely classical data structure, computed on ordinary classical hardware. The diagonal holds the usual action probabilities. The off-diagonal entries, called coherences, are new parameters. This gives density-matrix reinforcement learning (DMRL). Coherences change outcomes in two settings. First, suppose a policy is regularized toward a coherent reference policy using matrix relative entropy. The optimum then has coherences that no diagonal reference policy produces, and its action probabilities can depend on the reference policy’s coherences. This is relevant to fine-tuning of large language models (LLMs) with a Kullback–Leibler penalty. Second, suppose the agent chooses a unit-vector action and the reward is a quadratic function of that vector. This could arise when an LLM judge scores a weighted consolidation of reasoning paths, or when coupled robotic sensors each respond to a mix of actions. Then some coherent policies earn strictly more expected reward than every classical policy with the same action probabilities, and we give the margin in closed form. In a two-action example, both policies assign probability 0.5 to each action, yet the classical one earns expected reward 0 while the coherent one earns 1 , the maximum. Outside these two settings, DMRL matches or transfers the classical results. This is a theory paper. All results are proved, and we report no empirical evaluation.
[LG-102] CartoVec: General-Purpose Location Embeddings from Open Map Data
链接: https://arxiv.org/abs/2610.17190
作者: Furkan Gursoy
类目: Machine Learning (cs.LG); Geophysics (physics.geo-ph)
*备注: 16 pages, 4 figures
Abstract:Geographic coordinates are common inputs in applied machine learning, but latitude and longitude describe position rather than the surrounding environment. Image-grounded location encoders learn this context from satellite or ground-level imagery; structured open map data offers another source of supervision whose utility for continuous coordinate encoders remains less explored. We introduce CartoVec, a coordinate encoder trained to predict Overture-derived land cover, land use, transportation, place and connectivity targets at multiple spatial scales, without image inputs or downstream-task labels. We evaluate frozen embeddings on Census outcomes, Airbnb prices and Philadelphia crime in the US Northeast using spatial block cross-validation. CartoVec has the highest mean performance on most Census outcomes and the highest average across the full suite, with several comparisons close across folds. The findings support open map data as a practical source of reusable location representations with a coordinate-to-vector interface that requires no map lookup at inference. The open-weight model is available at this https URL.
[LG-103] ModRes: A modification of the ResPlan dataset ready for floor plan analysis and graph machine learning
链接: https://arxiv.org/abs/2610.17178
作者: Emma Shelley,Wassim Jabi,Thomas Woolley,Katerina Kaouri
类目: Machine Learning (cs.LG)
*备注: 24 pages, 23 figures
Abstract:Floor plan datasets underpin computational analysis of building layouts, spatial connectivity and graph machine learning. However, publicly available datasets often require substantial preprocessing before they can be used reliably. We present ModRes, a cleaned and enriched version of the ResPlan dataset, containing 8,431 valid and unique residential floor plans. Invalid and duplicate floor plans are removed through a dedicated cleaning workflow, producing a ready-to-use dataset. For each floor plan, ModRes provides two-dimensional geometry, images, three-dimensional building representations and three complementary dual graphs in CSV and JSON formats with graph-, node- and edge-level features. Continuous and categorical graph attributes support machine learning workflows. We characterise the dataset using graph-based statistics and demonstrate its utility through a range of case studies. ModRes provides structured geometric, topological and semantic representations needed to automate spatial analysis and support machine learning-based workflows for building design and performance evaluation.
[LG-104] FunDM: Cheap Label-Free Decision Models from Frozen Language Models
链接: https://arxiv.org/abs/2610.17175
作者: Ujjawal K. Panchal
类目: Machine Learning (cs.LG)
*备注:
Abstract:Decision models such as Jev score candidate answers or actions. A small decision head trained label-free can recover about 99% of a frozen language model’s multiple-choice accuracy while using only two-thirds of its transformer layers. It learns the full model’s answer probabilities with KL divergence, without correct-answer labels for training. One backbone pass reads the question and all options together; the head returns probabilities for every option. We call this model FunDM (Fun Decision Model). Using 24 of Qwen3-8B’s 36 layers and skipping the LM head, its +1T variant reaches 75.0% mean accuracy over eight benchmarks, versus 75.7% for the full model. On MMLU-Pro, both heads achieve 1.56 to 1.57 times batched vLLM throughput with about one-third lower peak GPU allocator memory. Both heads score higher than released Contrastive Language Models (CLM), fine-tuned CLM, and CLM-shaped heads trained from scratch on the eight multiple-choice benchmarks and coding-agent selection. Separately trained agent heads learn judge probabilities without pass/fail labels and select passing runs for 74.4 to 74.6 of 100 held-out tasks on average, versus 74 for the full-model judge and 52.0 for a CLM-shaped action-trained head. Code is available at: this https URL
[LG-105] Neuromorphic Decoding and Dynamical Image Reconstruction from Populations of Spiking Retinal Ganglion Cells
链接: https://arxiv.org/abs/2610.17154
作者: Ning Zhang,Arto Nurmikko
类目: Emerging Technologies (cs.ET); Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注:
Abstract:Decoding visual stimuli from the spiking activity of retinal ganglion cells (RGCs) informs both our understanding of the retinal code and the design of visual prostheses. Natural vision is dynamic, yet state-of-the-art decoders mostly reconstruct single frames with artificial neural networks that process dense, continuous-valued signals rather than the retina’s sparse, asynchronous spikes. We asked whether dynamic scenes can be reconstructed and tracked entirely within the spike domain, and at what computational cost. We built an end-to-end spiking neural network (SNN) whose neurons integrate the recorded RGC spikes over time. Populations of up to 1836 RGCs were recorded ex vivo from mouse retina on high-density microelectrode arrays while flashed images, moving bars, drifting gratings and randomly moving handwritten digits were presented. For moving digits, one shared spiking encoder fed two readouts, one reconstructing the image and one decoding the digit’s position. Compute energy was estimated from counted synaptic operations. The SNN decoded what and where at once. It reconstructed the stimuli with a structural similarity (SSIM) of up to 0.93, refining the reconstruction as spikes accumulated, and tracked a randomly moving target with a mean error of 1.33 pixels on a 32\times32 grid. Its fidelity nearly matched that of the best of 17 larger spiking and artificial decoders (5.9 million to 1.7 billion parameters, versus 5.4 million), and at a mean firing rate of 14.2% it needed an estimated \sim 70-fold less compute energy than a dense artificial network of the same architecture. Dynamic visual scenes can thus be decoded from retinal population activity in the spike domain, with a compact network and at a compute cost compatible with implantable and wearable neural interfaces, supporting spike-native decoders for visual prostheses.
[LG-106] OpenDGO: Benchmarking Multi-Agent Organizations Across Diverse Application Domains
链接: https://arxiv.org/abs/2610.17143
作者: Xunkai Li,Miao Zhang,Zekai Chen,Xuaner Chen,Yu liu,Rui Qian,Xun Wu,Hongchao Qin,Rong-Hua Li
类目: Machine Learning (cs.LG)
*备注:
Abstract:Large language model (LLM)-based multi-agent systems are increasingly applied across diverse domains, yet different applications require different forms of agent organization. Existing studies typically adopt predefined organizations, such as Star, Chain, and Tree, or learn communication graphs to improve performance on specific tasks. However, these methods are largely designed and evaluated within a limited set of application domains, providing little understanding of whether an organization effective in one domain remains suitable for others with different collaboration requirements. We present \textbfOpenDGO, a controlled benchmark and analysis framework covering six real-world domains and three collaboration settings: Sequential Dependency, Parallel Decomposition, and Information Integration. Under matched tasks, agents, tools, and resource limits, OpenDGO compares fixed and learned organizations while separately examining communication graphs, role assignments, input assignments, resource budgets, and component failures. Our results show that there is no ``one-organization-for-all’’ solution: preferred organizations vary across applications, and even the same communication graph can yield different outcomes under different role and input assignments. These findings highlight the need for application-dependent organization design, aiming to understand \emphwhich organization is suitable for a given application, under what conditions, and why.
[LG-107] Masked Attention Pooling Is Invariant to Padding and Permutation: A Formal Proof in Lean
链接: https://arxiv.org/abs/2610.17132
作者: Yusuf Khaled Sadek,Enes Yilmaz
类目: Machine Learning (cs.LG)
*备注: 9 pages, 1 figure, 2 tables. Code: this https URL
Abstract:Attention-based multiple instance learning (MIL) pools a variable-size bag of instance embeddings into one vector. In batched training, bags are padded to a common length, and the pooled output is expected not to depend on how much padding was added or on the order in which instances are stored. This expectation holds only if padded rows are removed from both the numerator and the softmax normalizer. Common implementation idioms meet it only under side conditions. Zeroing padded features alone does not work. Masking logits with -\infty fails on empty bags and on nonfinite padding. Masking with a large negative constant is exact only after floating-point underflow. We define masked attention pooling with a guarded normalizer and prove in Lean 4 with mathlib that it is invariant under any bijective reindexing of rows and under appending any finite number of masked rows with arbitrary real features and logits. We also prove that values at masked positions are irrelevant and that an all-masked bag maps to zero. The theorems hold for arbitrary finite index types and feature dimensions. Using the jax-lean certificate pipeline, we connect these theorems to the Jaxpr traced from the JAX kernel itself. Lean checks translation certificates for two concrete shapes, identifies the generated functions with the generic specification, and transfers both invariances to the imported Jaxpr’s real-arithmetic semantics. All audited theorems depend only on Lean’s three standard axioms. The results concern ideal real arithmetic. Floating-point execution, XLA, and the JAX tracer and importer remain outside the proof.
[LG-108] Can Rényi Accountants Beat Lossless Accountants for DP-SGD?
链接: https://arxiv.org/abs/2610.17120
作者: Jinho Bok
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR); Data Structures and Algorithms (cs.DS); Information Theory (cs.IT); Machine Learning (stat.ML)
*备注: 41 pages
Abstract:Recent frameworks in differential privacy (DP), such as privacy loss distributions and f -DP, provide lossless representations of the privacy leakage of mechanisms. These approaches are often considered to yield tighter (\varepsilon, \delta) -DP guarantees than those obtained from relaxed notions such as Rényi DP. But is that always the case? We study this question for differentially private stochastic gradient descent (DP-SGD) with fixed-size batches under replace adjacency. Our key observation is that in this setting, privacy accounting based on Rényi DP can use a pair of worst-case distributions (i.e., a dominating pair) that differs from the one used in the lossless frameworks. In particular, we characterize the worst-case distributions as two-dimensional Gaussian mixtures and derive an improved Rényi DP bound for DP-SGD. Numerical and asymptotic comparisons show that across various hyperparameter settings, a Rényi DP accountant based on the new bound can yield better privacy guarantees than those obtained from the lossless frameworks.
[LG-109] Masked Diffusion for Travel Diaries Synthesis
链接: https://arxiv.org/abs/2610.17109
作者: Pascal Jutras-Dubé,Raziq Zaman,Fabian Bastin,Cinzia Cirillo,Ruqi Zhang
类目: Machine Learning (cs.LG); Social and Information Networks (cs.SI)
*备注:
Abstract:Because household travel surveys combine household characteristics, variable-length person rosters, vehicle rosters, and trip chains, existing synthesis pipelines split them across separate models or flatten them into cropped fixed-length vectors. Representation, rather than modeling, becomes the bottleneck: splitting severs cross-segment dependence, and cropping truncates the household-structure distribution. We instead tokenize the diary and keep the object whole: each household-day becomes one token sequence with count tokens for vehicles, persons, and trips, and distinct symbols for padded, hidden, unobserved, and inapplicable positions. We introduce DiaryBridge, a masked diffusion model that bridges the fully hidden sequence to the distribution of complete diaries by unmasking positions in any order. One trained model, conditioned on any observed subset, performs synthesis and partial-diary completion alike. Experiments on real data extracted from the 2017/2018 Regional Travel Survey of the Baltimore region show that DiaryBridge generates travel reflecting vehicle availability and employment status, recovers shared trips using other household members’ observed diaries, and better reproduces aggregate demand across modes, purposes, and departure periods than the learned baselines.
[LG-110] LocalTrain: Integer-Only Local Learning for Energy-Efficient CNN Training on Edge FPGAs
链接: https://arxiv.org/abs/2610.17099
作者: Shiqi Ding,Xuanyu Liang,Xinyang Liu,Boyang Li,Zhiqiang Que,Guosheng Hu,Jiayang Li
类目: Hardware Architecture (cs.AR); Machine Learning (cs.LG)
*备注: 7 pages, 2 figures, 5 tables
Abstract:Local learning reduces cross-layer training dependencies, but an efficient accelerator must still retain operands, accumulate gradients, and select quantization scales. We present an FPGA system that trains a CNN and its regional auxiliary classifiers using a shared integer matrix engine. Compact errors are expanded on chip, gradients are folded tile by tile, and historical scales permit direct output conversion. At 200 MHz, the KV260 prototype achieves 76.57% accuracy on CIFAR-10 and 91.85% on SVHN at approximately 4.20 W mean module power. Effective matrix throughput is 31.22 and 31.57 GMAC/s, respectively. Against the same integer method on Jetson Orin Nano, measured training energy is 40.4% and 42.1% lower, with 1.86\times and 1.82\times longer runtime; measurements use the respective platform power domains. Selected state checks and matching complete-run prediction counts support functional agreement with the software reference.
[LG-111] Nearest Neighbor Functional Curve Completion for Intraday Probabilistic Revisions of Renewable Generation Forecasts
链接: https://arxiv.org/abs/2610.17075
作者: Guillermo Terrén-Serrano,Michael Ludkovski
类目: Machine Learning (cs.LG)
*备注:
Abstract:We propose a nonparametric approach that fuses deterministic initial forecasts and a partial realization into a probabilistic forecast using the framework of functional curve completion. Our motivation is to construct intra-day forecasts for renewable energy generation, which is a key operational task in modern power grids. The aim is to combine the day-ahead (DA) forecast and the morning realized generation profile to create trajectories for the rest of the day, quantifying operational uncertainty. Starting with a large database of complete forecast-actual pairs across many assets, we adopt a functional point of view and construct a weighted neighborhood based on similarity metrics for the observed partial curve, the DA forecast, the day, and the geographical location. The selected complete nearest-neighbor curves provide a full predictive distribution, including functional focal envelopes and predictive bands based on functional depth. The hyperparameters governing neighborhood size, distances, and fusion rates are tuned across forecast update times and directly target probabilistic calibration in terms of the predictive probability integral transform. Three case studies using synthetic wind and solar assets, as well as real-life regional wind generation, showcase that our algorithm is able to achieve excellent calibration.
[LG-112] ILE: Bridging LLM s and Table Foundation Models for Tabular Learning with Strings
链接: https://arxiv.org/abs/2610.17065
作者: Félix Lefebvre,David Holzmüller,Gioia Blayer,Myung Jun Kim,Marine Le Morvan,Gaël Varoquaux
类目: Machine Learning (cs.LG)
*备注:
Abstract:Real-world tables are often heterogeneous, combining numerical features with free-form string columns. Table foundation models (TFMs) excel on numerical tables but require strings to be encoded numerically, while large language models (LLMs) hold semantic knowledge but are not designed for tabular predictions. Most existing bridges use LLMs as plug-in encoders for cell strings, yet how to extract representations effectively for tabular learning remains underexplored. We show that the interface between LLMs and tables matter as much as the LLM itself. Applied cell-wise, larger LLMs do not give better features. With our interface, the same LLMs improve steadily with size. TILE (Table-Interfaced LLM Embeddings) combines four ingredients: row serialization to provide cross-column context, intermediate-layer extraction to avoid layers specialized for next-token prediction, columnar pooling to remove mean pooling’s length bias, and by-pass features to route numbers and categories past the LLM. Paired with TFMs, TILE sets a new state of the art for tables with strings. Across 100 datasets, it beats the strongest baseline on 89% of splits, with gains across 7 LLMs and 10 tabular learners, and ranks first on the 108-dataset STRABLE benchmark.
[LG-113] LLM Evaluation is a Forecasting Problem
链接: https://arxiv.org/abs/2610.17018
作者: Rahul Thomas,Adam Block,Micah Goldblum
类目: Machine Learning (cs.LG)
*备注:
Abstract:Standard procedure in machine learning is to randomly split train and test data, emulating IID draws from one underlying data distribution. In practice, models are trained on past data and deployed in the future, which may come from distinct distributions. In this paper, we find that web text is indeed highly non-stationary, as text distributions shift over time. We further observe that model performance changes continuously over time across diverse benchmarks. Consequently, future model performance is better predicted by measurements made on recent evaluation data, and is best predicted by time series forecasters applied to historic performance rather than by historic performance itself. Choosing which model to deploy based on forecasted accuracy enables more reliable selection of performant models, thus improving downstream accuracy in the regime practitioners care about: deployment. Taken together, our experiments suggest that time should be treated as a fundamental variable in LLM evaluation. We release our code and data.
[LG-114] Mixture of GFlowNets
链接: https://arxiv.org/abs/2610.17007
作者: Omar G. Younis,Joseph D. Viviano,Chirayu Anant Haryan,MD Vasimuddin,Sanchit Misra,Yoshua Bengio
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
*备注: 30 Pages, 21 Figures, 14 Tables, under review. Omar G. Younis and Joseph D. Viviano are co-lead authors
Abstract:Generative Flow Networks (GFlowNets) learn a policy that samples compositional objects in proportion to an unnormalized reward R(x) . This makes them useful wherever many diverse, high-reward candidates are needed and oracle queries are expensive. In practice, a single GFlowNet often fails to cover all modes of a complex distribution, and the obvious ways to scale compute saturate: larger batches and data-parallel gradient averaging draw every trajectory from one, possibly biased, policy, so extra rollouts rediscover known modes instead of finding new ones. We explore whether this mode-bias problem (mode collapse in the extreme) can be addressed with a population of asynchronously trained policies which divide the state space between them. Mixture of GFlowNets (MoG) replaces the single forward policy with a community of K independent GFlowNets coupled through one jointly trained soft classifier, which assigns each member its own region of the state space through reward shaping. Mixing the members under their learned partition functions provably recovers the target p(x)\propto R(x) . MoG exchange (MoGX) also lets the classifier route sampled trajectories: a trajectory the classifier assigns to another member is sent to that member and trained on there, with no importance sampling. We evaluate both methods by keeping the environment-interaction budget fixed while scaling the community size, turning parallel compute into mode coverage without extra oracle calls. On HyperGrid, across six reward landscapes and communities of up to K=64 , MoG and MoGX keep finding new modes as K grows, while data-parallel training saturates. On a genomic ideotype-design task, MoGX covers 39% of the 1024 optima at K=64 , against 31% for independent agents and 20% for data-parallel training. MoG also has a lower per-iteration synchronization cost than data-parallel training.
[LG-115] Multi-Agent Deferred-Decision Trajectory Optimization with Anytime-Valid Target Retention for Moving Candidates
链接: https://arxiv.org/abs/2610.17003
作者: Farbod Siahkali,Mengxue Hou,Vijay Gupta
类目: ystems and Control (eess.SY); Machine Learning (cs.LG); Robotics (cs.RO)
*备注: 8 pages, 3 figures, 3 tables. Submitted to IEEE Robotics and Automation Letters
Abstract:We study multi-agent terminal interception when one of several moving candidates is the unknown target and both target identity and candidate terminal locations are uncertain. We develop a receding-horizon deferred-decision trajectory optimization (DDTO) controller that combines conformal terminal-location prediction, sequential candidate rejection, capacity-constrained assignment, and candidate-contingent planning. Under exchangeability conditions for terminal and identity residuals, with predictable querying, we establish fixed-time terminal-location coverage and anytime-valid target retention. At the final replan, when one control step remains, we bound the joint probability that a tightened-set DDTO plan passes nonlinear rollout verification but capture fails. The closed-loop controller applies fallback guidance to any agent without a verified DDTO plan and, across 100 paired scenarios, captures 79% of targets under the strict control-effort budget and 85% when that budget is relaxed.
[LG-116] A physics-coupled graph neural operator for mid-plane six-component stress tensor field prediction in concrete penetration
链接: https://arxiv.org/abs/2610.16996
作者: Wenpu Du,Peng Zhou,Yunlong Xia,Sinuo Xin,Congcong Zhang,Boyang Zhang,Yi Zhang,Wenzheng Xu
类目: Machine Learning (cs.LG)
*备注:
Abstract:The penetration-resistant design of concrete structures requires the spatiotemporal evolution of the stress field within the target, whereas in design only the strength grade, the geometry and the projectile parameters are known, and the aggregate arrangement is not. Empirical formulae and scalar-output surrogates serve design screening, and mesoscale finite element models resolve the aggregate-scale response, but one solution cannot cover random aggregate samples at scale, and no framework links mesoscale heterogeneity to field-level prediction. A physics-coupled graph neural operator (PC-GNO) is proposed, which learns in function space the evolution of the mid-plane six-component stress tensor field. The training data are generated by full-scale, aggregate-resolved finite element models and comprise 359 cases at the impact velocities of 100, 135 and 165 m/s, corresponding to compressive strengths of 14, 23 and 48 MPa; the data source is verified by laboratory measurements. The coupling strength of message passing is constructed from the impedance contrast across the material interface, and the update is constrained by spectral normalization. Under the single-step primary setting, the mean accuracy of the three levels is 80.16% (91.45%, 77.80% and 71.22%), i.e. 27.81, 42.46 and 34.16 percentage points above the preceding-frame field of the same window; with 20 664 parameters it exceeds the 87.9% of the best Fourier neural operator configuration (about 812 times as many parameters) and the 71.0% of the closed-form linear regression. The output is the complete six-component tensor field rather than a fixed engineering quantity, so that strength criteria and energy indicators follow from one set of weights. For field-level response prediction in concrete penetration, this result provides a data-driven route, with conditions of applicability defined by the training and the evaluation.
[LG-117] Intrinsic Geometry Matching: learning observation-equivariant process geometry
链接: https://arxiv.org/abs/2610.16995
作者: Manuel Lecha,Michael Bronstein,Iolo Jones
类目: Machine Learning (cs.LG); Differential Geometry (math.DG)
*备注:
Abstract:Real data are often arbitrary, indirect observations of an unknown underlying process, and those observations can distort its geometry. Existing geometric methods equip data with the induced geometry of the ambient space, but this entangles the true geometry of the process with the distortion of the observation. To address this, we introduce Intrinsic Geometry Matching (IGM): a scalable approach to learn the intrinsic geometry of the underlying process. Specifically, we learn the carré du champ operator of the process, allowing us to recover its diffusion geometry. Our method uses the fact that data often come with natural perturbations, like steps in a time series, to recover the intrinsic carré du champ while transforming naturally with changes in observation: it is observation-equivariant. Furthermore, since this equivariance relates the geometry of different spaces, we can train a model using additional sources of data, even when that data lies in another space. We evaluate IGM on molecular dynamics, image, and video data and find better performance and scaling than existing neural and graph-based geometric tools.
[LG-118] Constant-Depth Rational-Weight GNNs and their Connection to Schanuels Conjecture
链接: https://arxiv.org/abs/2610.16993
作者: Daumantas Kojelis,Carsten Lutz
类目: Machine Learning (cs.LG); Logic in Computer Science (cs.LO)
*备注:
Abstract:A landmark result states that the expressive power of graph neural networks (GNNs) is bounded from above by the 1-dimensional Weisfeiler-Leman (1-WL) test. We ask whether a constant-depth GNN can match the expressive power of 1-WL while using only rational weights. We prove that this is the case when Schanuel’s conjecture, a famous open problem from transcendental number theory, is taken to be true. Our GNN construction works even with integer weights, and more generally with a broad range of rational weights. As the activation function, it may use logistic sigmoid, hyperbolic tangent, or related functions. It is uniform across all graph sizes. We also prove a partial converse: if a GNN with exponentiation, hyperbolic sine or hyperbolic cosine as activation matches the expressive power of the 1-WL test in a slightly more expressive setting where graphs may contain certain auxiliary vertices, then this implies Schanuel’s conjecture restricted to the numbers that arise in computations of the GNN. We also conduct experiments showing that the constructed GNNs indeed match the 1-WL test.
[LG-119] Generative Models as Adaptive Trust Regions for Discrete Bayesian Optimisation
链接: https://arxiv.org/abs/2610.16992
作者: Michael Dodds,Erik Bodin,Sebastian W. Ober,Tom Diethe,Carl Henrik Ek,Henry Moss
类目: Machine Learning (cs.LG)
*备注: 33 pages (14 main text and references, 19 supplementary material), 8 figures, 16 tables. Code available at this https URL
Abstract:Bayesian optimisation (BO) is a powerful framework for sample-efficient optimisation of expensive black-box functions, but its application to high-dimensional discrete scientific design remains challenging. Local trust-region BO methods such as TuRBO address high dimensionality by restricting search to adaptively sized neighbourhoods. However, a direct discrete analogue based on edit distance can be poorly suited to structured discrete objects: neighbourhoods grow combinatorially, and metric proximity does not capture the dependencies that determine whether a design is plausible or useful. Motivated by the growing availability of pretrained generative models, we introduce GenTuRBO, which replaces the geometric trust region with an adaptively scaled proposal distribution defined by a pretrained generator. The generator provides a learned prior over discrete transformations, encoding coordinated changes without explicitly enumerating edit-distance neighbourhoods. GenTuRBO performs strongly across all three domains and achieves the highest mean final scores on Ehrlich sequence optimisation and Sokoban coordination.
[LG-120] Learning to Design SR Solvers: Harness Optimization for Symbolic Regression
链接: https://arxiv.org/abs/2610.16966
作者: Qian Li,Yuxiao Hu,Qihao Duan,Jinyue Yan,Yuntian Chen
类目: Machine Learning (cs.LG)
*备注:
Abstract:Symbolic regression (SR) seeks explicit equations that explain observed data, and recent methods increasingly employ large language models (LLMs) and agents for equation discovery. However, these systems typically rely on manually designed workflows and prompts. We argue that the SR agent harness should itself be an optimization target: its design shapes search priors and how numerical feedback guides subsequent equation refinement. We therefore propose HarnessSR, a framework for jointly optimizing prompts and workflows to construct reusable SR agents. During both harness optimization and agent execution, HarnessSR supports either PySR or an LLM-based equation generator through tool-specific interfaces, allowing the framework to optimize search guidance for different equation-discovery mechanisms. HarnessSR first optimizes role prompts through equation-discovery feedback in basic workflows, then jointly searches prompt assignments and workflows with brief adaptation to align the two. Finally, it refines role instructions through reflection on residual patterns, candidate equation structures, and search outcomes, improving how these signals guide subsequent equation search. On LSR-Synth in LLM-SRBench, HarnessSR achieves a mean ID \mathrmAcc_0.01 of 61.33% across three backbones, a 45.0% relative improvement over the next-best baseline.
[LG-121] Learning to Design Convex Relaxations for Nonconvex Optimization
链接: https://arxiv.org/abs/2610.16948
作者: Jinrun Liu,Bai Yan,Tengfei Liu,Qunfeng Liu,Jian Yang,Shi Cheng,Yuhui Shi,Qi Zhao
类目: Machine Learning (cs.LG); Mathematical Software (cs.MS)
*备注:
Abstract:Convex relaxations make nonconvex optimization tractable by replacing hard nonconvex problems with convex surrogates that give valid bounds. Yet their design is still largely manual: a human expert must decide where to relax, which relaxation to use, and how much formulation growth is worth paying. We show that this design process can be learned. The method proposed in this paper, AutoConvexRelax, represents the symbolic formulation of a nonconvex problem as a tripartite graph linking variables, symbolic terms, and their objective or constraint positions. It then learns a policy that composes symbolic rewrites into problem-specific relaxation trajectories as the formulation evolves. We prove that finite action sequences generated by the learned policy preserve relaxation validity under explicit domain, occurrence-context, and representation-consistency conditions. We further establish computable error bounds for mixed relaxations and sufficient conditions for feasible recovery and global-optimality certification. As a first instantiation, we apply the method to nonconvex quadratic problems and fractional problems. On 241 benchmark quadratic instances, the learned policy strengthens McCormick lower bounds by 13.52%. Its mean lower bound is 3.70% below strengthened semidefinite relaxations on the normalized comparison metric, while it uses substantially smaller formulations. The learned policy retains this trade-off on fractional instances and reaches near-semidefinite bounds on beamforming and sensor-network localization instances. These results suggest that convex relaxation design can move from a fixed handcrafted choice to an automated learning process that adapts to problem structures. Code is available at this https URL.
[LG-122] Making Tensor Decoupling More Robust and Less Parametric
链接: https://arxiv.org/abs/2610.16937
作者: Maxime Rochkoulets,Joppe De Jonghe,Mariya Ishteva
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注:
Abstract:Tensor decoupling algorithms recover compact representations of multivariate functions by combining linear factor matrices with univariate internal nonlinearities. We identify two limitations of recent methods. First, their performance is not invariant to mismatched output or gradient magnitudes, and second, their spline internals require manual tuning of their degrees of freedom. We propose two lightweight fixes. The first is a preprocessing step that rescales the input Jacobian tensor, removing both the bias towards larger magnitude outputs and the ill-conditioning of certain factor updates caused by such a mismatch. The second replaces B-spline internals by penalized splines coupled with generalized cross-validation, making the method less dependent on the user selecting the right number of control points. Finally, we release an efficient open-source JAX implementation of our algorithm.
[LG-123] me Window Optimization for Motor Imagery Classification via Temporal Structure Embedded Reinforcement Learning
链接: https://arxiv.org/abs/2610.16921
作者: Huan Luo,Na Lu,Xiaopeng Wang,Ziyang Shang
类目: Neural and Evolutionary Computing (cs.NE); Machine Learning (cs.LG)
*备注: 12 pages, 4 figures. Code available at this https URL
Abstract:Online motor imagery-based brain-computer interfaces (MI-BCIs) typically decode continuous EEG using fixed-length sliding windows that may be misaligned with neural events. Existing window optimization methods often evaluate windows independently, overlooking temporal dependencies and potentially producing fragmented or physiologically implausible selections. We propose Temporal Structure Embedded Reinforcement Learning (TSE-RL), a framework that incorporates prior knowledge of MI event structure into a constrained Markov decision process for adaptive window optimization. First, TSE-RL assigns each window an explicit semantic role through Discard, Onset, Sustain, or Offset actions, extending binary retain/discard decisions to event-level alignment. Second, a finite-state machine masks invalid actions to enforce the temporal order of MI phases and promote physiologically plausible window sequences. Third, a Transformer-based history encoder captures dependencies across windows, improving training stability and computational efficiency compared with recurrent architectures. TSE-RL is plug-and-play and can be integrated with diverse MI classifiers. Experiments on multiple benchmark datasets demonstrate consistent improvements in decoding performance in both cross-session and cross-subject settings. Visualization analyses further suggest that the learned window selection policy aligns with neurophysiological principles.
[LG-124] ML Engineering Agents - A Data Management Perspective
链接: https://arxiv.org/abs/2610.16919
作者: Arnab Phani,Elias Strauss,Tim Schuster,Sebastian Schelter
类目: Databases (cs.DB); Machine Learning (cs.LG); Software Engineering (cs.SE)
*备注: 6 pages
Abstract:Machine learning engineering (MLE) agents are emerging as a new paradigm for automating end-to-end ML pipeline development. Using large language models for code generation and predictive performance as a machine-gradable feedback signal, these agents iteratively synthesize, execute, and refine pipeline code. Yet current systems are largely evaluated on pre-built, single-table benchmarks that abstract away key data management challenges. This tutorial introduces the foundations of MLE agents and highlights research opportunities for the database community. We cover how agents combine program synthesis with iterative and evolutionary optimization, and discuss open challenges in efficient pipeline execution, data discovery and augmentation over heterogeneous sources, reliability and debugging, as well as fairness and compliance. We connect these challenges to existing data management techniques and conclude with a hands-on session using contemporary MLE agents.
[LG-125] AI-Assisted Tumor-Growth Law Discovery from Sparse Data: Refitting and LLM -Based Ranking
链接: https://arxiv.org/abs/2610.16891
作者: Yitian Lu,Weinan Wang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Inferring tumor-growth laws from sparse, noisy observations requires more than accurate curve fitting. We study an AI-assisted workflow involving symbolic regression, mathematical admissibility checks, coefficient refitting, and large language model (LLM) ranking. Synthetic Gompertz data allow trajectory fit, forecasting, and recovery of the governing growth law to be assessed separately. A dense noiseless control recovers the generating family, but sparse noisy sampling frequently leads to alternative forms. Among retained candidates in the tested configurations, physics-informed neural network refitting often increases growth-law error, whereas classical nonlinear least squares usually reduces it. A separate exploratory analysis of archived LLM rankings shows no consistent improvement over selection by training error and identifies errors in mathematical reasoning. Plausible saturation, accurate trajectory fits, and convincing explanations can therefore coexist with incorrect growth laws. Mathematical validity and law recovery require evaluation alongside predictive performance.
[LG-126] FORGE: Procedural Weight Generation for Sparse MoE Inference ASPLOS2027
链接: https://arxiv.org/abs/2610.16890
作者: Chanyoung Moon,Travis Pence,Daisuke Yamada,Jiaqi Mo,Yanxin Zhang,Vikas Singh,Karthikeyan Sankaralingam
类目: Hardware Architecture (cs.AR); Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注: 16 pages. Accepted to ASPLOS 2027
Abstract:We introduce FORGE (Fast On-chip Rotor Generation for Experts), a procedural weight generation system for Sparse Mixture-of-Experts (MoE) inference. Rather than storing and loading every expert weight tensor from HBM, we show how a compact accelerator inside each Streaming Multiprocessor can regenerate them on-the-fly from O(\log^2 d) rotor parameters per projection. Per replaced layer, gigabytes of stored expert weights are approximated with under a megabyte of seeds. This addresses the dual memory bottleneck of autoregressive Sparse MoE decode, where per-expert FFN operations are HBM-bandwidth-bound at small routed-token counts and the aggregate expert footprint forces multi-node Expert Parallelism (EP) with all-to-all traffic. We realize the approach as the Rotor Generation Unit (RGU), a fixed-function datapath that generates FFN weight tiles without competing with Tensor Core execution. With 30% of the MoE FFN layers replaced across five Sparse MoE models, we estimate 1.3-2.8x higher system throughput at matched GPU count and 1.10-1.15x lower per-user decode latency at short context. In exchange, downstream accuracy on three model families falls below that of the same model fine-tuned without rotor replacement, and the replacement ratio controls this tradeoff. FORGE also expands KV cache capacity by 1.2-2.6x. It is agnostic to the bit width at which weights are stored, so it can compose with quantization. The RGU is fully implemented/verified in SystemVerilog at ~0.9% SM area overhead (0.032 mm^2/SM, 7nm).
[LG-127] Learning Operators for Symmetry Discovery: From Lie Groups to Local Symmetry Geometry
链接: https://arxiv.org/abs/2610.16884
作者: Pavan Karjol,Scindhiya Laxmi,Prathosh AP
类目: Machine Learning (cs.LG)
*备注:
Abstract:Continuous symmetry discovery is typically formulated as a separate estimation problem for each function or dataset. We instead investigate whether the map from a function to its symmetry structure can be learned across a family of functions from finite observations. Using the infinitesimal differential characterization of continuous symmetries, we formulate this map as an inverse symmetry operator, analogous to an inverse PDE problem. For global Lie-group symmetries, we establish a representation-theoretic population characterization: the symmetry Lie algebra is the nullspace of a positive-semidefinite operator constructed from representation coefficients and the infinitesimal group action. We also show that the inverse symmetry operator is equivariant under coordinate changes through the adjoint action. This characterization separates the geometric symmetry map from finite-sample estimation, motivating a hybrid approach that amortizes statistically difficult components while retaining analytic structure. On SO(5) benchmarks, amortized equivariant estimation of harmonic coefficients improves symmetry recovery from sparse observations, whereas ordinary least squares suffices for SL(2,R). We extend this viewpoint to local symmetries, where symmetry directions form smoothly varying tangent distributions rather than a fixed Lie algebra. Under the Lie-bracket consistency condition, Frobenius’ theorem guarantees local integral submanifolds along which observables remain unchanged. We recover this geometry using a hybrid estimator that learns observation weights while preserving analytic differential and geometric recovery. Across global, non-compact, local, and physically motivated synthetic benchmarks, our results support learning the statistically difficult finite-sample components while preserving the known geometric structure of symmetry recovery.
[LG-128] Sequence Composition in EHR Foundation Model Pretraining
链接: https://arxiv.org/abs/2610.16878
作者: Joshua Placidi,Inigo Stibbe,Alfred J. Balston,Marek Rei,A. Aldo Faisal
类目: Machine Learning (cs.LG)
*备注: 26 pages, 11 figures, 10 tables
Abstract:Autoregressive electronic health record (EHR) foundation models predict the next event token in patient trajectories: sequences of structured medical events. A typical pretraining strategy is to concatenate all trajectories into a global stream and sample token windows from it. This gives patients with longer sequences more training exposure and allows tokens to attend across patient boundaries. We argue that these choices create a misalignment between pretraining and downstream clinical tasks. To explore this, we investigate how the composition of pretraining sequences affects clinical prediction performance. Using the global-stream approach as a baseline, we propose alternative sequence composition methods that control how patient trajectories contribute to pretraining sequences. Across four MIMIC-IV v2.2 tasks, our best method improves macro AUROC, AUPRC, F1, and balanced accuracy, with relative gains of 9.3% in F1 and 2.7% in AUPRC under a fixed training recipe. We find that these gains generalize across four additional evaluation settings, including the INSPIRE and eICU-CRD datasets, with improvements in macro AUROC, AUPRC, and F1 score. Our results show that simply changing how EHR data are presented to the model during pretraining can improve downstream performance, without changing the model or collecting additional data.
[LG-129] From Parallel to Sequential: Small-Batch Noise Restores Simplicity Bias of Muon
链接: https://arxiv.org/abs/2610.16868
作者: Tom Jacobs,Rebekka Burkholz
类目: Machine Learning (cs.LG)
*备注: 27 pages, 13 figures
Abstract:Simplicity bias, which can arise from sequential feature learning during training, has been proposed as a general explanation for the success of deep learning. An important driver of this phenomenon is saddle-to-saddle behavior, in which the gradient flow trajectory remains close to a sequence of saddle manifolds while learning one feature at a time. However, Muon, a state-of-the-art optimizer based on spectral descent, does not exhibit such dynamics in the full-batch setting and instead learns features in parallel. How is it then that Muon is so successful in deep learning settings? We show that stochastic gradient noise fundamentally changes this behavior: small-batch noise restores sequential feature learning, providing a mechanism for simplicity bias that is absent from full-batch Muon. We explain this transition through stochastic smoothing of the spectral descent operator underlying Muon. When gradients are small relative to the noise scale, the expected spectral update becomes approximately proportional to the actual gradient, which restores gradient-flow-like dynamics. We formalize this mechanism for deep linear models by deriving the effective continuous dynamics of noisy spectral descent. Experiments on low-rank matrix reconstruction, vision transformers, and large language models (LLMs) support these findings.
[LG-130] Efficient Resource-Shiftable RL Post-training with Generative Rewards
链接: https://arxiv.org/abs/2610.16866
作者: Yanfeng Zhao,Haoyang Li,Sheng Lin,Yuming Zhou,Jiawei Jiang,Bin Cui,Fangcheng Fu
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
*备注:
Abstract:Reinforcement learning (RL) post-training with generative reward models (GenRM) has become a central recipe for improving large language models (LLMs), and it is also increasingly resource-intensive: policy rollout, GenRM-based reward evaluation, and policy training all compete for GPU resources. However, existing colocated and disaggregated systems rely on static resource allocation for these roles. Such allocation cannot adapt as each role’s workload changes during RL and therefore leaves a role’s resources underutilized. We present PivotRL, an RL post-training system that dynamically shifts resources among different roles according to their runtime demands. PivotRL introduces a resource-shiftable runtime built on a Global Parameter Storage, together with a unified mechanism that efficiently shifts role-specific context. A shift-aware scheduler enables sequence migration upon a shift, while a planner considers the aggregate throughput benefit and the shift overhead to determine when and how to shift. We evaluate PivotRL on a 128-GPU cluster using four model groups spanning dense and mixture-of-experts models. PivotRL enables resource shifting within seconds and improves end-to-end throughput by 1.81-2.05 times over colocated deployment and 1.50-1.95 times over disaggregated deployment, while preserving RL convergence. Subjects: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC) Cite as: arXiv:2610.16866 [cs.LG] (or arXiv:2610.16866v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.16866 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-131] GpFlash: Maximizing Speed and Minimizing Memory for Gaussian Process Training and Prediction
链接: https://arxiv.org/abs/2610.16861
作者: Andre Marchildon
类目: Machine Learning (cs.LG); Mathematical Software (cs.MS); Optimization and Control (math.OC)
*备注:
Abstract:This paper presents GpFlash, a Gaussian process library that is focused on fast training and posterior evaluations, minimal memory usage, and scalability to high-dimensional problems. GpFlash was compared to the popular GP libraries scikit-learn, GPy, GPflow, GPyTorch, and SMT on a MacBook Pro and the supercomputer Trillium. GpFlash was found to use less memory and to be faster for training and posterior evaluations relative to the other GP libraries for all cases considered. For example, when using \nx = 10^4 training points in a 10-dimensional space GpFlash required 9.6 to 35 times less memory and was 2.6 to 54 times faster than the other GP libraries for the evaluation of the marginal log likelihood and its gradient. The memory reduction was achieved by only using one matrix of size \nx \times \nx to store different symmetric matrices in its lower and upper triangular entries. The computational speedup was the result of designing algorithms to access memory efficiently, using BLAS and LAPACK methods extensively, and factoring kernel equations so that expensive exponential calculations could be reused. GpFlash was more than 440 times faster than all of the other GP libraries considered when evaluating the posterior mean together with its gradient for the case with 10^4 training points and 10 dimensions. GpFlash is available at this https URL.
[LG-132] Radial Tail Repair: Post-Training Correction for Diffusion and Flow Models
链接: https://arxiv.org/abs/2610.16851
作者: Hamza Cherkaoui,Hélène Halconruy,Antonio Ocello
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 30 pages
Abstract:Diffusion and flow-matching models may struggle to accurately represent distributional tails, especially for heavy-tailed data. Existing approaches often use heavy-tailed noise or modify the generative dynamics, requiring dedicated retraining and more difficult estimation. We take a simpler approach: rather than relearning the full distribution, we correct only the model’s behavior in the extreme regime. Our method builds on the radial-angular decomposition of multivariate heavy tails, where radius and direction become asymptotically independent. This allows us to preserve the directional structure learned by the pretrained model and correct only its radial tail. We estimate the tail onset from reference and generated samples, then rescale generated extremes so that their radial survival matches the target distribution. The resulting method is a lightweight post-training correction applied to a frozen diffusion or flow-matching model, requiring neither retraining nor changes to its sampling dynamics. Experiments on challenging synthetic heavy-tailed distributions and real weather and wildfire datasets show that this simple correction substantially improves extreme-tail coverage using only a small number of fitting samples.
[LG-133] Mean Accuracy Is the Wrong Estimand for a Release Decision
链接: https://arxiv.org/abs/2610.16825
作者: Jaewon Choi,Jihye Kang,Namhyuk Ahn
类目: Machine Learning (cs.LG)
*备注:
Abstract:A release decision about a fine-tuned model is an operational commitment about the \emphnext rerun of its training recipe. Yet standard benchmark reporting gives only the mean accuracy of past seeds. These target fundamentally different questions: release decisions depend on catastrophic tail risk, whereas standard evaluation reports only central tendency. Because fine-tuning induces bimodal mode collapse, high average accuracy frequently conceals severe failure rates. Auditing 2,124 fine-tuned models across eleven language models, we show that shipping models based on mean accuracy violates a 10% deployment failure budget 48.0% of the time. Common heuristics like \bar b - 2\sigma prove equally dangerous: by underestimating variance on small samples, they become most permissive when evidence is weakest. We introduce \textbfrecipe cards, an evaluation framework that directly certifies rerun tail failure rates using exact finite-sample binomial bounds. Recipe cards reduce held-out budget violations to 0.6% , enforce a mathematical sample floor ( N \ge 22 ) that prevents certification on thin evidence, and explicitly refuse unverified claims under domain and task shifts.
[LG-134] DASS: A Solver-Adaptable Dynamic Auxiliary Search Strategy for Symbolic Regression NEURIPS2026
链接: https://arxiv.org/abs/2610.16812
作者: Yuxiao Hu,Qian Li,Ding Wang,Yekun Zheng,Jinyue Yan,Yuntian Chen
类目: Machine Learning (cs.LG)
*备注: Accepted at NeurIPS 2026
Abstract:Symbolic regression (SR) aims to discover interpretable mathematical expressions from data and plays a key role in scientific discovery. However, existing methods face a common bottleneck: the enormous search space and severe combinatorial explosion make it difficult to achieve a favorable trade-off between accuracy and algorithmic efficiency. We argue that, in addition to designing increasingly complex solver-specific SR strategies, an equally promising direction is to equip them with a shareable auxiliary strategy that can provide reliable search guidance. To this end, we propose DASS, a solver-adaptable Dynamic Auxiliary Search Strategy that improves SR by constructing and refining a high-potential auxiliary search subspace. DASS represents the auxiliary search space as a functional subspace spanned by a set of high-potential basis function terms. It estimates term utility through multi-environment evaluation, filters unstable guidance, and updates term potential scores with a Gibbs posterior, thereby calibrating the subspace to provide more reliable guidance for downstream solvers. The high-quality solutions produced by the solver are used to update the set of basis terms. This forms a virtuous closed-loop optimization process: the subspace guides the downstream solver, and the solver feedback reshapes the subspace for subsequent exploration. DASS can be easily integrated with various types of SR solvers. Extensive experiments on LLM-SRBench show that DASS significantly improves the accuracy of original solvers with a manageable increase in runtime.
[LG-135] Predicting Audio Popularity: A Comparative Study of Metadata and Acoustic Features Across Music and Sound Effects
链接: https://arxiv.org/abs/2610.16809
作者: Ikram El Hadi,Nihad Elbari,Kawtar Tlemçani,Théo Mariotte,Meysam Shamsi
类目: ound (cs.SD); Machine Learning (cs.LG)
*备注:
Abstract:This study addresses the prediction of audio popularity on an online sharing platform by leveraging both metadata and acoustic feature representations. Two distinct categories of audio content are considered: music and sound effects. Popularity is formulated as a supervised classification task based on two target variables: the number of downloads and the average user rating. We conduct a comparative evaluation of multiple machine learning models, including Random Forest, XGBoost, and a Neural Network model. Experiments are performed across different feature sets, including metadata, platform-provided descriptors, and eGeMAPS acoustic features. Results indicate that metadata consistently provide the most informative representation for predicting popularity. Nevertheless, acoustic features alone achieve performance significantly above chance level, demonstrating that intrinsic properties of the audio signal contribute to popularity prediction.
[LG-136] One-Layer Transformers Provably Learn In-Context K-Nearest Neighbor Prediction with Chain-of-Thought
链接: https://arxiv.org/abs/2610.16798
作者: Lyumin Wu,Yuan Cao
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 107 pages, 2 figures
Abstract:Chain-of-Thought (CoT) enables transformers to solve complex reasoning tasks by generating intermediate reasoning steps. Despite strong empirical success, the mechanisms by which such reasoning abilities emerge during training remain poorly understood, particularly from a theoretical perspective. In this work, we study this question in a stylized yet fully analyzable setting: in-context \mathrmK -nearest-neighbor ( \mathrmK -NN) prediction with a one-layer transformer. Earlier work has demonstrated that one-layer transformers can be trained to perform 1 -NN without CoT (Li et al., 2024a) by having the softmax attention attend to the nearest neighbor in context. However, this result on 1 -NN cannot be extended to \mathrmK -NN for K1 , as softmax attention is much less naturally suited to attending to the 2nd through K -th nearest neighbors. We first give a rigorous negative result showing that this limitation already appears on a well-separated class of binary classification tasks for odd K1 . We then give an explicit construction of a one-layer transformer, and show that it can solve \mathrmK -NN via CoT reasoning. We further prove that, under a stylized training setup, gradient descent can recover this construction, thereby showing that transformers can acquire the capability to solve \mathrmK -NN through training. In addition, we show that the resulting trained model applies to a broader test class than that assumed in the training analysis. Our results shed light on how CoT expands the ability of transformers to learn and solve tasks that are otherwise hard to solve.
[LG-137] Efficient Preconditioned Optimization Methods with Frequent Directions
链接: https://arxiv.org/abs/2610.16792
作者: Sifan Yang,Yuanyu Wan,Yibo Wang,Lijun Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Adaptive optimization methods have received considerable attention in recent years due to their strong empirical performance. However, they generally require maintaining a preconditioning matrix, which incurs prohibitive computational overhead in large-scale applications. To address these limitations, we investigate the acceleration of preconditioned optimization methods through frequent directions (FD), a widely used matrix sketching technique. We first propose FTSL by integrating FD with adaptive subgradient methods under the primal-dual framework and adding back the cumulative information discarded by FD. FTSL obtains a tighter regret bound than existing state-of-the-art results. We then apply a doubling trick that further reduces the time complexity of FTSL, while maintaining the same regret guarantee. We also consider optimization over matrix-valued decision variables and exploit their structural properties to construct sketched preconditioners. We develop FTSL-Shampoo by adapting FD to Shampoo and give a novel analysis under the primal-dual framework. To further mitigate its computational complexity, we design one-sided variants of FTSL-Shampoo under both mirror descent and primal-dual frameworks. Our algorithms not only reduce the computational cost of FTSL-Shampoo, but also achieve tighter regret bounds. We extend our methods to offline stochastic convex optimization and exploit smoothness to achieve faster convergence rates. Finally, we introduce practical variants of our algorithms and empirically demonstrate their efficiency and effectiveness across a broad range of tasks.
[LG-138] NFMap: Topographic 2D Manifold Learning with Invertible Normalizing Flows
链接: https://arxiv.org/abs/2610.16750
作者: Héléna Alexandra Gaspar
类目: Machine Learning (cs.LG)
*备注: 20 pages, 10 figures. Code: this https URL
Abstract:We present NFMap, a dimensionality reduction method for 2D visualisation that extends Generative Topographic Mapping (GTM) with a bijective normalizing flow. Like GTM, NFMap places a fixed mixture grid prior over a 2D latent map, giving a parametric embedding, an explicit likelihood, and the ability to project new points onto an existing map; the likelihood is an effective out-of-distribution signal on chemical datasets. Unlike GTM, whose latent-to-data mapping is an RBF network of bounded flexibility, NFMap’s mapping is a normalizing flow, and is therefore fully invertible: any location on the map can be carried back to data space. To alleviate map crowding on large datasets, NFMap additionally supports heavy-tailed Student-t mixture components. NFMap maps data to a latent space of the same dimension, split into a 2-dimensional navigable map and an off-manifold residual. Because the flow is bijective and its objective penalises volume contraction, variance squeezed out of the residual must reappear in the navigable map. The residual scale is the main control between local neighbourhood fidelity and global distance fidelity; its effect depends on the dimensionality of the data and is coupled to the extent of the latent grid.
[LG-139] CopyQuad: Quadrupedal Skill Transfer from a Single Side-View Video Without a Depth Model
链接: https://arxiv.org/abs/2610.16737
作者: Mohamad H. Danesh,Raina Hu,Mengxue Xu,Hsiu-Chin Lin
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注:
Abstract:Video-to-robot pipelines for quadrupedal imitation insert a learned lifting stage, either a monocular depth model or a per-clip 3D reconstruction, to recover the one axis a side-view camera cannot see. We present CopyQuad, which omits that stage and pins the missing axis to the robot’s own stance width instead. The setting is side-view footage of sagittally dominated motion, where limb motion is almost entirely in the image plane. A frozen 2D keypoint detector and a fixed analytic retargeting then turn a single uncalibrated clip into a robot reference whose foot targets are reachable by construction, so no per-clip trajectory optimization is needed. One phase-indexed reward structure, whose only per-behavior freedom is three scalars from a shared sweep, learns a dog walk, a horse walk, a horse trot and two non-periodic gestures, all deployed zero-shot on hardware. We then measure what omitting the axis costs. In a training-free rigidity test, a video depth model varies the distance between the animal’s hips by 40 to 107% of its own mean and a single-image metric model by 22 to 58% , though that distance cannot move. Motion capture puts the true lateral paw travel of a dog walk at 4.5 cm peak to peak, a mean deviation near 1.4 cm against the 3.5 cm of lateral foot error our policies already leave, so the discarded axis sits below the pipeline’s resolution. Depth repaired to feasibility never beats the constant, and the unrepaired estimates put the robot on the ground.
[LG-140] Universal Distributed Feature Extraction Complementary to Side Information via H-Score
链接: https://arxiv.org/abs/2610.16707
作者: Yuan Xinjie,Dier Tang,Shaolun-Huang
类目: Information Theory (cs.IT); Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注:
Abstract:This paper considers a distributed estimation system where edge nodes transmit averaged features to a center node for parameter estimation. A key challenge arises when the center node possesses auxiliary side information: how to design local extractors that are universally optimal for any downstream task, while avoiding redundancy with the side information. To tackle this problem, we orthogonally decompose the joint function space and demonstrate in the weak-dependence regime that the modal decomposition of the canonical dependence kernel’s conditional-additive component yields optimal, task-independent features. We construct a distributed nested H-score to learn this decomposition and show that, under a limited feature budget, the mean-squared error (MSE) gap to the best task-specific features is determined solely by the task’s alignment with the discarded, least informative modes, and vanishes to leading order for every square-integrable task once the budget is sufficient.
[LG-141] Making LoRA Identifiable: Orthogonal Alignment for Continual Learning NEURIPS2026
链接: https://arxiv.org/abs/2610.16664
作者: Pratik Rakesh Singh,Mohammadi Zaki,Akash Saha,Aneesh Mukkamala,Pankaj Wasnik
类目: Machine Learning (cs.LG)
*备注: Accepted at NeurIPS 2026
Abstract:There is growing interest in applying Parameter-Efficient Fine-Tuning (PEFT) to Continual Learning, as it enables faster training, mitigates catastrophic forgetting, and facilitates better adaptation to new tasks. Prior approaches have explored a range of strategies, including extending new LoRA branches for each incoming task, training under orthogonal loss objectives across all task-specific LoRA modules, and employing gating mechanisms to integrate new and existing LoRA modules. However, these methods incur memory costs that grow linearly with the number of tasks, limiting their scalability. More recent approaches investigate training a single LoRA by leveraging asymmetries in its parameterization; however, they often suffer from insufficient representational alignment across the LoRA parameter space. To address these limitations, we propose Stiefel Optimization and Aligned Rotation (SOAR), a two-stage framework for continual LoRA learning. First, we constrain the A matrix to lie on the Stiefel manifold, thereby reducing coordinate misalignment to the space of orthogonal rotation matrices. Second, we introduce an Orthogonal Procrustes alignment procedure for the B matrix to estimate the optimal rotation, coupled with a per-rank retention merging strategy that quantifies directional agreement between representations of new and previously learned tasks. We conduct extensive experiments across multiple models and benchmarks, demonstrating that SOAR achieves strong continual learning performance without incurring additional memory overhead.
[LG-142] Compression Through Time: An Interplay of Separability Stationarity and Capacity
链接: https://arxiv.org/abs/2610.16660
作者: Nikolaus Kopp,Franz Pernkopf,Bernhard Geiger
类目: Machine Learning (cs.LG)
*备注:
Abstract:Information-theoretic and geometric compression have been widely studied as possible explanations for generalization. Yet their temporal dynamics in recurrent networks remain largely unexplored. Using discretized LSTMs, whose finite embedding space enables direct estimates of information quantities, we observe pronounced compression across the final timesteps of many datasets. We explain this effect through three competing mechanisms: (i) class separability permits stronger compression at the final timestep; (ii) shared recurrent dynamics propagate this pressure across timesteps with similar activation states, making its extent depend on temporal stationarity; and (iii) limited representational capacity opposes prolonged compression because it restricts information available for subsequent computation. We systematically test these mechanisms through controlled synthetic interventions and real-world sequence tasks. Our results identify class separability, temporal stationarity, and network capacity as key determinants of temporal compression, and suggest that compression acts as a regularizer-potentially beneficial when excess capacity is available, but harmful to generalization when it consumes memory needed for the task.
[LG-143] Guidance Without Labels: Flow Matching from Coupled Gaussian Mixtures
链接: https://arxiv.org/abs/2610.16654
作者: Stanislas Ducotterd,Bassam El Rawas,Michael Unser
类目: Machine Learning (cs.LG)
*备注:
Abstract:Flow matching models are typically trained to transport a Gaussian source to the data distribution, with source and target samples paired independently. We replace the source with a fitted Gaussian mixture and couple each data point to a mixture component through an entropic optimal transport problem. We show that the resulting transport reduces to the standard Gaussian-source problem up to a guidance-like correction, which enables the reuse of a pretrained velocity field. This correction is a closed-form function of the fitted mixture plus a term that we learn with a small correction network. We test our model on published checkpoints for CIFAR-10 and CelebA-HQ. It improves FID at every step count, and on CelebA-HQ it reaches the quality of the baseline with about half the steps. The closed-form term alone, with no training, already delivers most of this improvement; the small correction network accounts for the rest.
[LG-144] Counting with Attention and LayerNorm: Norm Scaling and Representation Geometry
链接: https://arxiv.org/abs/2610.16653
作者: Ayyub Shaffy,Manoj Dhakal,Safal Shrestha,Keith W. Ross
类目: Machine Learning (cs.LG)
*备注: 29 pages, 8 figures, 3 tables
Abstract:Counting is a simple task on which transformers can nevertheless fail at long sequence lengths. We study the cost of counting in a single attention layer, up to a maximum length L , focusing on how softmax attention and layer normalization shape the representation from which the count is recovered. We mainly focus on regression read-out but also consider classification read-out. For regression, we show unnormalized linear attention counts exactly at parameter norms independent of L . For softmax attention, regression cannot count exactly with finite parameters; it recovers the count after rounding, but only when the vector parameter norms grow with L . We measure this cost by B_\min , the smallest bound that can be placed on every trainable vector to ensure exact counting after rounding, and derive tight rates. At fixed model dimension, B_\min grows as L without normalization, as L^3/2 under bare post-normalization, and as L^3/4 with learned post-normalization. The last rate is unexpected: post-normalization discards the magnitude that carries the count and might be expected to make counting harder, yet with a learnable gain it counts at a smaller norm than softmax without normalization. Bare pre-normalization is different: it bounds the attention scores, so beyond a finite length no parameter norm suffices. A classification read-out keeps every norm bounded under every placement, but a dense head stores one output vector per count and costs more in storage. Experiments on the post-normalized model reproduce this split between the two read-outs. Together, our theoretical and empirical findings characterize how self-attention, layer-normalization placement, and read-out choice shape the representation geometry and parameter cost of counting in the minimal single-head architecture studied here.
[LG-145] IND-Eval: Aligning Evaluation and Simulation for Industrial Time Series Foundation Models NEURIPS2026
链接: https://arxiv.org/abs/2610.16652
作者: Vebjørn Haugland(1),Pavel Gulin(1),David Foss(1),Peet Cremer(1),Erlend Aune(1),Simone Casolo(1),Andreas Solbrå(1),Milad Mobarhan(1) ((1) Cognite)
类目: Machine Learning (cs.LG)
*备注: 19 pages, 5 figures. Accepted to the Foundation Models for Temporal Systems workshop at NeurIPS 2026. IND-Eval: this https URL
Abstract:We introduce IND-Eval, a public benchmark that organizes industrial time series across domains into source-specific forecasting tasks. Relative to established benchmarks such as GIFT-Eval, it emphasizes finer sampling resolutions and a greater prevalence of multivariate forecasting. Evaluating Chronos-2, TiRex-2, and TimesFM-3, we find that TimesFM-3 achieves the lowest source-balanced aggregate nMASE and nCRPS and leads seven of nine sources under each metric; source-specific exceptions reinforce the need for domain-representative evaluation for model selection. These findings also motivate improving domain alignment during pretraining, where real industrial data are often too scarce to close domain-specific gaps directly; we therefore investigate whether a domain’s statistical structure can guide simulator design. In a case study on two offshore well-monitoring regimes, we improve a general simulator’s coverage using an oscillator-augmented simulation pipeline. This lowers forecasting error in the target domain but reduces performance on the subset of GIFT-Eval with natively univariate targets. A 50/50 mixture of the general and augmented simulation recipes achieves the lowest error among the three training recipes for all four event-metric pairs in the case study while recovering most of the lost GIFT-Eval performance. Together, these results illustrate how domain-representative evaluation can guide model selection and the composition of simulated data while revealing a trade-off between specialization and generality.
[LG-146] Learning Deployable Causal Action Geometry under Temporal Non-Stationarity NEURIPS2026
链接: https://arxiv.org/abs/2610.16644
作者: Changjian Liu,Tianyu Wang,Yuwei Xu,Xiaoxuan Deng,Zhilin Zhang,Yong Gao,Chuan Yu,Jian Xu,Bo Zheng
类目: Machine Learning (cs.LG)
*备注: Accepted at NeurIPS 2026
Abstract:Learning a response law for continuous decisions from observational time series is central to many real-world decision systems, yet temporal non-stationarity poses a fundamental challenge: as outcomes drift over time, naively pooled models can conflate action effects with shifting baselines and even reverse action rankings. Even when the one-step response is identifiable, unrestricted cross-time generalization remains fundamentally ill-posed. We therefore seek the part of the response that remains stable for decisions even when outcome levels do not. We propose deployable causal action geometry: the action-response shape that remains after anchoring away action-independent level shifts on a selected link scale and profiling out positive amplitude variation. Building on this target, we develop a two-stage orthogonal estimator that first obtains cross-fitted pilot estimates of anchored responses and then performs honest profiled selection over link, representation, and shape classes. The resulting estimator admits oracle guarantees for the restricted target and, under approximate persistence, produces decision rules that remain near optimal in future periods. Experiments on synthetic data and Dominick’s support the proposed target and its transfer behavior. In a randomized Taobao A/B test, the deployed system improves GMV and purchase counts by (1.16%) and (1.70%).
[LG-147] ACED: Reference-Anchored Aggregation of Natural-Language and Executable Evidence for Mathematical Reasoning
链接: https://arxiv.org/abs/2610.16635
作者: Yiyi Lu,Jiaxiang Geng,Bing Luo
类目: Machine Learning (cs.LG)
*备注: 23 pages, including references and appendices. Yiyi Lu and Jiaxiang Geng contributed equally
Abstract:Natural-language self-consistency and executable programs provide complementary test-time evidence for mathematical reasoning, yet either channel may support the same semantic error. Thus we propose ACED, a reference-anchored framework that treats the self-consistency answer as the default and executable outputs as supporting evidence rather than interchangeable votes. ACED converts source-specific answer counts into smoothed supports, combines them through a weighted logarithmic pool, and overrides the reference only when a calibrated relative-evidence margin is exceeded. The same rule supports both candidate selection and admission of executable-only answers. Across five math word-problem benchmarks, ACED matches or outperforms the corresponding baselines. On GSM8K, ACED-CANDIDATE recovers 89.7% of the attainable candidate-set gain with only three damaging overrides, while ACED-ALL reaches 86.9% accuracy versus 82.6% for direct Program-of-Thought aggregation. These results show that preserving evidence provenance enables more reliable integration of natural-language and executable reasoning.
[LG-148] DR-APO: Decoupled-Representation Advantage-Guided Policy Optimization from Continuous Control to Embodied Navigation
链接: https://arxiv.org/abs/2610.16618
作者: Runhua Zhang,Yingjie Niu,Chao Li
类目: Machine Learning (cs.LG); Robotics (cs.RO)
*备注: 9 pages, 4 figures,
Abstract:Policy representations in reinforcement learning have expanded beyond Gaussian policies to diffusion and flow-matching based generative policies. However, existing RL methods remain representation-specific, lacking a unified optimization interface for emerging generative policies. In particular, extending likelihood-based objectives to multi-step generative policies couples optimization to their internal generation process, where accumulated likelihood ratios can weaken the update signal. Meanwhile, generative-policy-specific methods often rely on multi-candidate generation and critic-based selection, improving overall return while providing limited improvement to the policy itself. We introduce DR-APO (Decoupled-Representation Advantage-Guided Policy Optimization), which decouples policy improvement from policy representation through advantage-guided target improvement and representation-compatible direct action fitting, providing a shared optimization principle for Gaussian, diffusion, and flow policies. Across five MuJoCo-v5 tasks, DR-APO remains competitive with SAC for Gaussian policies and achieves the best aggregate policy-only performance for both diffusion and flow-matching policies. NavDP post-training improves the Success Rate (SR) from 70.33% to 89.55% in simulation and from 25% to 65% in real-world navigation. These results demonstrate a unified policy-improvement principle across Gaussian, diffusion, and flow policies, from continuous-control learning to embodied-policy post-training.
[LG-149] DEFlow: Deep Equilibrium Flow Maximum-Entropy Reinforcement Learning
链接: https://arxiv.org/abs/2610.16597
作者: Ruslan Ageev,Huaze Tang,Runqing Miao,Jiarong Li,Wenbo Ding
类目: Machine Learning (cs.LG); Robotics (cs.RO)
*备注:
Abstract:Expressive flow policies can represent multiple high-value actions, but sampling them with few integration steps while retaining tractable action densities remains challenging. We introduce DEFlow, a deep equilibrium flow policy that replaces explicit Euler sampling with implicit Euler integration and solves all intermediate actions jointly as an equilibrium system. We combine implicit differentiation with a discrete change-of-variables formula for the final-action density, enabling off-policy training within soft actor-critic without backpropagating through solver iterations. DEFlow achieves an average relative improvement of approximately 10% in mean successful coverage over the strongest baseline across four Multi-Goal docking variants. It also achieves the highest mean time-averaged return on four of five manipulation tasks, while remaining broadly competitive across five MuJoCo tasks. These results support implicit flow policies for tasks requiring both diversity and precision, at the cost of additional nonlinear and linear solves. Primary
[LG-150] Prorsa: Unleashing the Potential of Prefill-Decode Disaggregation in Agent ic Reinforcement Learning
链接: https://arxiv.org/abs/2610.16592
作者: Chiheng Lou,Xiangwei Deng,Bangjun Xiao,Xiaodong Ji,Yikai Zhao,Xin Jin,Hailin Zhang
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注:
Abstract:Agentic reinforcement learning (RL) is commonly used to enhance the capabilities of large language models (LLMs) in solving complex multi-turn tasks. During rollout, agentic RL exhibits a workload pattern distinct from that of traditional RL: multi-turn interactions repeatedly introduce prefill requests, causing prefill and decode workloads to coexist throughout rollout. This workload makes prefill-decode (PD) disaggregation, a design widely used for online LLM serving, promising for speeding up agentic RL rollout. This paper shows that PD disaggregation can improve rollout efficiency by (a) mitigating interference among data-parallel attention modules, (b) enabling stage-specific parallelism, and © reducing end-to-end trajectory completion time. However, a fixed prefill-decode allocation may fail to sustain these benefits as workload composition varies across tasks and over time. We therefore present Prorsa, an agentic RL framework for prediction-guided prefill-decode resource allocation. Prorsa observes the lifecycle of every trajectory to forecast short-horizon prefill and decode demand. These forecasts then guide worker allocation decisions and trigger role switching of rollout engines, aligning prefill and decode capacity with anticipated workload demand. Across synchronous and asynchronous agentic RL settings, Prorsa achieves end-to-end rollout speedups of 1.11x-4.05x over prefill-decode colocation.
[LG-151] HyBRA-EEG: Hyperbolic Global-Local Brain-Region Aggregation for Self-Supervised EEG Representation Learning
链接: https://arxiv.org/abs/2610.16589
作者: Xin Liu,Feng Lu,Wei Li
类目: Machine Learning (cs.LG)
*备注:
Abstract:Electroencephalography (EEG) representation learning is challenged by noisy signals, inter-subject variability, and heterogeneous electrode montages in different datasets and recording settings. Recent self-supervised EEG pretraining methods improve transferability by converting heterogeneous recordings into unified token or sequence representations, but channel-aware designs may still encode montage- or dataset-specific regularities. We propose HyBRA, a self-supervised framework based on hyperbolic global-local brain-region aggregation. HyBRA maps heterogeneous channel observations into a fixed regional token space by combining electrode coordinates, region priors, and feature-dependent hyperbolic distances. A channel-evidence injection branch preserves fine-grained discriminative cues during aggregation. The resulting tokens are processed by a Region-Time Transformer and pretrained with masked reconstruction of latent features and region-level signals. HyBRA outperforms the compared methods in average balanced accuracy on six benchmark datasets. Under a controlled six-fold evaluation on OpenBMI-MI, HyBRA matches CSBrain in balanced accuracy, improves AUROC by 1.84 percentage points and AUPRC by 5.23 percentage points, and uses 96.8% fewer parameters. Matched comparisons on TUAB and TUEV show higher balanced accuracy for hyperbolic matching than for Euclidean distance or cosine similarity.
[LG-152] OctMole: Multiscale 3D Molecular Generation with Hierarchical Spatial Flows
链接: https://arxiv.org/abs/2610.16586
作者: Yu-Tong Jin,Yi-Yang Tian,Qiwei Ye,Peng-Shuai Wang
类目: Machine Learning (cs.LG)
*备注: 28 pages, 22 figures, and 17 tables
Abstract:Generating accurate 3D molecules requires jointly determining a variable number of atoms, their identities, and geometry across spatial scales: global shape emerges coarsely, whereas chemical bonds demand fine local precision. Existing atom-based diffusion and flow models operate largely at a single scale, while dense spatial fields become costly at high resolution. We introduce OctMole, a multiscale octree flow framework for 3D molecular generation. A hierarchical occupancy flow progressively resolves sparse molecular support from coarse to fine, and a joint atom flow assigns atom types and continuous in-cell offsets to occupied leaves. This decomposition combines efficient global structure with subcell geometric precision, while cell corruption improves robustness to generated support. On QM9 and GEOM-DRUG, OctMole achieves strong chemical, distributional, and geometric quality, with particularly clear gains over Uni-3DAR on GEOM-DRUG and substantial efficiency gains over dense 3D generation. As an additional capability, intermediate occupancy states and selected regions provide spatial conditioning without task-specific retraining, supporting multiscale shape conditioning, regional editing, linker generation, and shape interpolation. Code and pretrained checkpoints are available at this https URL and this https URL, respectively.
[LG-153] Characterizing and Mitigating Distributional Bias in Physics-Informed Generative Models
链接: https://arxiv.org/abs/2610.16582
作者: Tanja Bien,Samir Darouich,Lluís Pastor-Pérez,Vinh Tong,Loay Mualem,Mathias Niepert
类目: Machine Learning (cs.LG)
*备注:
Abstract:Physics-informed generative models augment a data-fitting objective with residual penalties that encourage physical constraints to hold in generated samples. However, improving constraint satisfaction need not improve fidelity to the data distribution. We study this conflict in residual-regularized diffusion models and identify two possible consequences. First, the same prediction error can incur different residual costs near different feasible states, preferring states with lower residual sensitivity (avoidance). Second, this preference can extend to feasible states outside the data support, where the residual also vanishes (leakage). Motivated by this analysis, we propose a data-anchored penalty that weights prediction errors by the residual Jacobian at the clean training sample and normalizes the resulting magnitude. This removes the incentive to relocate to less sensitive states without changing the model architecture or sampling procedure. Experiments on a controlled synthetic system, three PDE systems, and molecular transition-state generation with flow matching demonstrate the benefits of anchoring. Code is available at this https URL.
[LG-154] LLM -HB: Language-Aware LLM -Guided Hotword Biasing for Code-Switching ASR
链接: https://arxiv.org/abs/2610.16579
作者: Yuxuan He,Genshun Wan,Pengcheng Li,Gongping Huang,Jianqing Gao,Yanmin Qian
类目: ound (cs.SD); Machine Learning (cs.LG)
*备注: 5 pages, 2 figures, 2 tables
Abstract:Code-switching (CS) speech, characterized by language alternation within a single utterance, is prevalent in multilingual communities and poses challenges to automatic speech recognition (ASR) due to rapid language transitions. We propose LLM-HB, a language-aware large language model-guided hotword biasing framework for code-switching ASR (CS-ASR). The framework incorporates a mixture-of-experts (MoE) adaptor and an auxiliary language prediction head to better capture and differentiate multilingual representations, while hotwords are injected through LLM-based prompting. Experiments on the ASRU2019 code-switching challenge test set demonstrate that LLM-HB achieves a 20.30% relative reduction in mixed error rate (MER) over the baseline, reaching 5.85% MER under a 15-distractor hotword setting. To the best of our knowledge, this is the first work to integrate LLM-guided biasing into code-switching ASR. We further release the extracted ASRU2019 hotword list to support future research.
[LG-155] he Language of Looking: Motion Tokens Reveal Domain-Dependent Local Structure in Gaze Sequences EMNLP26
链接: https://arxiv.org/abs/2610.16565
作者: Virmarie Maquiling,Enkelejda Kasneci
类目: Machine Learning (cs.LG)
*备注: 18 pages, 6 figures, EMNLP26
Abstract:Human gaze is the primary high-bandwidth interface through which we parse the semantics of our environment. While often modeled as a sequence, the degree to which gaze exhibits transferable local structure across domains remains underexplored. Using motion tokenization as a diagnostic lens across four diverse datasets, we investigate the transferability of local sequence structure. We demonstrate that while gaze can share a reusable motion-token vocabulary, identical motion tokens are organized into different local transition patterns depending on the dataset and task context. Our results show the limitations of naive data-pooling under low-semantic motion-token representations and provide a framework for evaluating behavioral sequences through information-theoretic diagnostics.
[LG-156] Optimal Transport PAC-Bayes Bounds
链接: https://arxiv.org/abs/2610.16560
作者: Armin Beck,Peter Ochs,Jalal Fadili
类目: Machine Learning (cs.LG)
*备注:
Abstract:PAC-Bayesian generalization bounds are among the strongest distribution-dependent guarantees in statistical learning and are typically expressed in terms of the Kullback-Leibler (KL) divergence between a prior and posterior distribution. While analytically convenient, the KL divergence requires the posterior to be absolutely continuous w.r.t. the prior, imposing support constraints that can substantially restrict the choice of posterior. Moreover, the KL divergence does not account for the underlying geometry of the probability space. In this work, we derive new PAC-Bayesian bounds replacing the KL divergence with general optimal transport (OT) distances, including the total variation and Wasserstein- p distances.
[LG-157] Scalable Identification of Physics-Based and Neural Dynamics from Partial Measurements
链接: https://arxiv.org/abs/2610.16559
作者: Yibo Zhou,Johannes Köhler
类目: ystems and Control (eess.SY); Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注: The code is available at this https URL
Abstract:We present a scalable method for identifying nonlinear dynamics from partial measurements, supporting physics-based, neural-network, and gray-box models. We use the Prediction Error Method (PEM) coupled with state estimation performed by a Moving Horizon Estimation (MHE) scheme, which can account directly for general nonlinear dynamics. This results in a bi-level optimization problem that scales poorly with increasing model parameters. To improve computational efficiency, we derive a reformulation that avoids explicit parameter-wide sensitivity matrices and exploits the temporal structure of the underlying optimization problem. Numerical comparisons show up to a 215\times speedup in hypergradient evaluation. Racing-vehicle simulations demonstrate accurate prediction and physical-parameter recovery in tire-force identification. Real-world data show that the gray-box model improves prediction accuracy with a neural residual while retaining physical interpretability.
[LG-158] DeepFlow: Two Convergent Differentiable Approaches for Pairwise Conditional Random Fields
链接: https://arxiv.org/abs/2610.16546
作者: John S. H. Baxter
类目: Machine Learning (cs.LG)
*备注: 28 pages, 15 figures, 4 appendixes
Abstract:Basic probabilistic models form the basis of modern deep learning largely by defining the final activation layer of a network and allowing the networks outputs to be interpreted through a probabilistic lens. By increasing the complexity of these final activation layers to include features such as spatial regularisation, allow for these models to improve in performance without requiring significantly more data. The challenge however is to ensure that these more complex probabilistic graphical models adapt well to being used in tandem with deep learning, that is, ensuring they converge and have meaningful derivatives. The goal of this paper is to show two algorithms with provable convergence properties to allow this type of regularisation for image and signal classification models. Through both rigorous mathematical analysis and in silico experiments, we can show the differing performance between these solution types.
[LG-159] Neural Network Preconditioning: A Case Study for the Solution of the Parametric Helmholtz Equation
链接: https://arxiv.org/abs/2610.16542
作者: Luc Giraud(1 and 2),Carola Kruse(2),Paul Mycek(2),Maksym Shpakovych(1 and 3),Yanfei Xiang(1, 2 and 4) ((1) Inria, Concace joint team between Airbus CRT, Cerfacs and Inria, (2) Cerfacs, Concace joint team between Airbus CRT, Cerfacs and Inria, (3) Current affiliation: Opticalp, (4) Current affiliation: IRMA, Université de Strasbourg, MACARON, Inria)
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG)
*备注:
Abstract:This work presents a hybrid numerical approach for solving linear systems arising from the discretization of the two-dimensional parametric Helmholtz equation. A convolutional neural network based on the U-Net architecture is trained in an self-supervised manner (i.e., without explicitly generating reference solutions) to approximate the inverse of the discretized Helmholtz operator, using a loss function involving the residual norm of the linear system. The trained network is used as a nonlinear preconditioner within the flexible GMRES (FGMRES) algorithm. Numerical experiments show that while the neural network is not accurate enough to act as a standalone solver, it significantly improves the convergence of FGMRES when employed as a preconditioner. The neural preconditioner demonstrates robust performance and generalization capabilities with respect to variations in the velocity field and the domain size. Comparisons with algebraic preconditioners, based on LU factorizations of approximations of the original system matrices, indicate the superior efficiency of the neural approach under equivalent conditions. We believe that the proposed method is not tied to a specific neural architecture and can be extended to other parametric PDEs.
[LG-160] Learning Compression and Meta-Complexity
链接: https://arxiv.org/abs/2610.16535
作者: Halley Goldberg,Mikito Nanashima,Igor C. Oliveira
类目: Computational Complexity (cs.CC); Machine Learning (cs.LG)
*备注:
Abstract:The idea that learning is closely tied to compression has long served as both a powerful principle and, at times, an informal analogy across machine learning and learning theory. Here we focus on settings in which that intuition can be made mathematically precise, surveying formal connections among learning algorithms, data compression, and computational complexity theory, with an emphasis on recent progress and current frontiers. We describe how techniques from complexity theory can guide the design of both supervised and unsupervised learning algorithms with provable guarantees, and conversely, how learning algorithms can yield new insights and results in complexity theory. A central theme is the role of data compression as a bridge between learning and complexity, especially through an area called meta-complexity, which studies the computational difficulty of compression problems and time-bounded variants of Kolmogorov complexity.
[LG-161] Reinforcement Learning with Hawkes-Structured Networks for Market Making
链接: https://arxiv.org/abs/2610.16524
作者: Fernando Acero,Philipp Plank,Yufei Zhang
类目: Machine Learning (cs.LG); Computational Finance (q-fin.CP)
*备注: Accepted to ACM ICAIF 2026
Abstract:Market making under realistic order-flow dynamics is challenging due to the clustered, self-exciting nature of order arrivals, which are naturally modeled by non-Markovian Hawkes processes. We propose a model-free reinforcement learning (RL) framework for market making with general Hawkes order-flow dynamics. We first establish non-asymptotic performance bounds that quantify the loss incurred by approximating the infinite-dimensional Hawkes state using a finite collection of exponentially weighted features. Guided by this analysis, we introduce the Hawkes-Structured Network (HSN), a novel neural architecture that encodes historical order flow through a learnable mixture-of-exponentials filter. HSN is trained end-to-end using model-free RL, jointly optimizing the feature representation and quoting policy directly from reward feedback. Experiments on single- and multi-asset market-making problems with mixture-of-exponentials, power-law, and regime-switching kernels demonstrate that HSN outperforms general-purpose neural architectures, achieving higher trading rewards while implicitly recovering the underlying order-flow intensities.
[LG-162] Learning Reachability-Based Safety Representations for Real-Time Robot Navigation in Unknown and Dynamic Environments
链接: https://arxiv.org/abs/2610.16522
作者: Bojan Derajić
类目: Robotics (cs.RO); Machine Learning (cs.LG); Systems and Control (eess.SY)
*备注:
Abstract:Safe robot navigation in unknown, dynamic environments requires more than geometric collision checking: a robot still far from an obstacle may be unable to avoid it because of its speed, orientation, or the obstacle’s motion. Hamilton–Jacobi (HJ) reachability analysis captures this in a value function whose sign separates states that admit a safe continuation from those where collision is unavoidable, but computing it online for every newly observed environment is too expensive. This dissertation trains neural implicit representations offline on reachability solutions to give optimization-based controllers this information in real time. Since the value function never exceeds the constraint, it equals a directly computed constraint, such as a signed distance, minus a nonnegative dynamics correction. Learning only this correction keeps represented obstacles outside the predicted safe set despite approximation errors. The first two of five methods use the learned function as a terminal constraint in model predictive control, for static local environments and predicted obstacle motion. The next two use it as a barrier candidate in control barrier function safety filters, for static environments and, through pairwise barriers in relative coordinates, for a varying number of moving obstacles. Simulations and hardware experiments on ground and aerial robots show better collision avoidance than geometric and hand-designed baselines, especially for short horizons and moving obstacles. The fifth method replaces the grid-based offline solver, which limits the first four to low-dimensional models. For control-affine dynamics with polytopic inputs, the HJ Hamiltonian is maximized at a vertex, so a parallelizable tree search over vertex inputs can generate training labels. With a loss from the HJ variational inequality, these labels yield barrier candidates for systems with up to 13 states.
[LG-163] Asymptotic Risk of Learning-Rate Schedules for Least-Squares Regression with Discretely Distributed Features
链接: https://arxiv.org/abs/2610.16518
作者: Noé Colomban,Francis Bach,Fabian Schaipp
类目: Machine Learning (cs.LG); Optimization and Control (math.OC); Machine Learning (stat.ML)
*备注: 42 pages, 9 figures
Abstract:We study the last iterate of stochastic gradient descent (SGD) on streaming least squares with discretely distributed features (Slock’s model), which decouples the dynamics across eigendirections. Under power-law source and capacity conditions, we derive the exact finite-time risk and asymptotic equivalents of constant and Warmup-Stable-Decay (WSD) schedules. With a tuned base learning-rate, the resulting scaling law decays as \Theta(T^-q) , where T is the horizon, and q is independent of the cooldown length c \in (0,1] but is strictly smaller for the constant schedule: any cooldown of fixed relative length c 0 , however short, improves the risk by a polynomial factor, while its length only affects the constant, which is minimized by linear decay ( c = 1 ). Finally, we show that the optimal risk depends on the batch size b and horizon T only through bT , up to a critical batch size.
[LG-164] rust-Calibrated VLA Policy Post-Training with Evolving Imagination
链接: https://arxiv.org/abs/2610.16517
作者: Wenzhan Li,Yiran Qin,Heng Zhou,Huirui Wang,Yulan Guo,Ruimao Zhang
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注:
Abstract:Vision-Language-Action (VLA) models are often pretrained via imitation learning, making reinforcement learning (RL) post-training important when deployment moves beyond demonstration coverage. Existing refinement paradigms expose a reliability–efficiency tension: model-free refinement avoids simulator bias but is interaction-intensive, whereas model-based refinement can exploit unreliable imagination even as its simulator adapts. We propose Model-Evolving Reinforcement Learning (MERL), a trust-calibrated framework that couples policy optimization with an online-updated simulator comprising a world model and a reward proxy head. Across stages, simulator reliability controls the imagined-data ratio and rollout horizon; within rollouts, chunk-level trust controls replay selection and optimization weight. For recursive imagination without exact grounded continuations, a residual predictor estimates trust without querying future ground truth. On perturbed LIBERO-PRO tasks, MERL reaches 79.7% average success, improves high-threshold learning speed by 10.9% over the matched model-free control, and outperforms same-protocol external baselines across three training seeds. Fixed-data adaptation from 50 demonstrations per task achieves 80.0% average success on two physical-robot tasks.
[LG-165] A temporal graph dissimilarity measure capturing epidemic spreading behavior
链接: https://arxiv.org/abs/2610.16456
作者: Salvatore Freni,Lorenzo Dall’Amico,Francesca Pistilli,Alain Barrat,Tatiana Tommasi,Ciro Cattuto
类目: ocial and Information Networks (cs.SI); Machine Learning (cs.LG)
*备注:
Abstract:Comparing temporal networks is a challenging problem in network analysis. Commonly adopted methods compare structural properties of temporal graphs and can be highly sensitive to the pre-selected features. However, temporal networks are the support of dynamical processes, and a process may take place in statistically indistinguishable ways on structurally different temporal graphs. In particular, a dissimilarity measure that compares temporal graphs based on how spreading processes unfold on them is highly desirable. In this work, we first define the Time-Shift DTW, a dissimilarity between epidemic curves that accounts for both differences in their shape and temporal shifts. We then introduce Epidemic-Aware Temporal Graph Dissimilarity (EATGD), a deterministic and differentiable dissimilarity between temporal graphs that combines an embedding-based temporal graph distance with a set of topological and temporal features. Its parameters are calibrated using stochastic SIR simulations to maximize the correlation with the Time-Shift DTW, after which graph comparisons require no further epidemic simulations. Using empirical human proximity networks from the SocioPatterns collaboration and synthetic temporal graphs, we show that EATGD captures epidemic-based graph dissimilarity and generalizes across heterogeneous network settings. Its differentiability and computational efficiency make EATGD suitable for integration into machine-learning pipelines and provide a potential loss function for training generative models to reproduce epidemic dynamics without requiring an exact reconstruction of fine-grained network structure.
[LG-166] ECHO: Correcting Time Series Foundation Model Forecasts with Historical Analogues and Errors
链接: https://arxiv.org/abs/2610.16433
作者: Boya Zhang,Xilin Dai,Yilin Wang,Qiang Xu,Luyao Dong,Xing He,Wen Wu
类目: Machine Learning (cs.LG)
*备注: 26 pages, 12 tables, 6 figures
Abstract:Time Series Foundation Models (TSFMs) provide general-purpose forecasting capabilities, yet their forecasts can remain misaligned with the temporal characteristics of a target domain. This motivates lightweight forecast correction even when only black-box access to the TSFM is available. Our key observation is that two forms of historical information provide complementary correction signals: historical analogues reveal how similar contexts evolved, while past forecast errors reveal a TSFM’s characteristic deviations in the target domain. Based on this observation, we propose ECHO, a black-box forecast correction framework that learns from both sources. ECHO uses AnalogE to assess retrieved historical analogues together with the current context and base forecast, learning transferable corrections through cross-domain pretraining. In parallel, RecentE learns target-specific residual corrections from past forecast errors and refines them as causally available feedback accumulates. ECHO dynamically combines the two corrections according to their observed forecasting performance. Across 5 representative TSFMs and 9 real-world datasets, ECHO reduces MSE in all 45 backbone–dataset settings, yielding an average relative reduction of 9.26% in macro-average MSE. Its gains also extend to long-horizon forecasting.
[LG-167] Resample the Answer: Where to Spend Test-Time Compute in Recursive Reasoning
链接: https://arxiv.org/abs/2610.16427
作者: Maxime Illouz
类目: Machine Learning (cs.LG)
*备注: 23 pages, 4 figures, 16 tables
Abstract:Recursive reasoning models such as the Tiny Recursive Model (TRM) iteratively refine two latent states: an answer state y, decoded into the output, and a reasoning state z, which is never decoded. Many of their failures have been attributed to bad basins, regions of latent space that trap the recursion on wrong answers. Recent test-time approaches perturb y, z or both to recover these failures, but the impact of this choice on performance, test-time compute efficiency and scaling properties remains unclear. Through controlled experiments on TRM, we find that a failed trajectory carries its failure in its answer state. At equal compute, in depth and in width, resampling y recovers more failures than any perturbation of z we tested, and the performance gap widens as the test-time compute budget grows. Our approach, Gated Answer Resampling (GAR), pairs this resampling of the answer state with the model’s own halting head and restarts only the puzzles that the head rejects. With at most 64 restarts per puzzle and no training or hyperparameter search, GAR takes TRM from 85.9% to 99.4% on Sudoku-Extreme and from 68.8% to 85.6% on Pencil Puzzle Bench validation. On the same checkpoints, GAR also outperforms Probabilistic TRM (PTRM), which perturbs z at each step, even when PTRM uses 42 to 197 times as much compute.
[LG-168] Sampling from Neural Operator Priors to Solve Partial Differential Equations
链接: https://arxiv.org/abs/2610.16396
作者: Andrew Millard,Joshua Murphy,Zheng Zhao
类目: Machine Learning (cs.LG)
*备注:
Abstract:Guided sampling with sparse observations and PDE residuals has recently been combined with neural operator priors to solve partial differential equations in function space, their natural setting. We pair such a prior with pseudo-bootstrap sequential Monte Carlo (pBS SMC), a guided sampling framework that accommodates proposals without tractable densities, and introduce a new proposal by adapting the DPM++ algorithm to the guidance setting. Across five benchmark PDE systems this proposal attains the best error–time trade-off at every step count tested, and wrapping it in pBS gives the lowest error on most inverse-problem settings, with the largest gains when few sampling steps are used.
[LG-169] naspaclust: A hybridisation of spatial fuzzy clustering with nature-inspired optimisation
链接: https://arxiv.org/abs/2610.16356
作者: Bahrul Ilmi Nasution,Rezzy Eko Caraka,Robert Kurniawan
类目: Neural and Evolutionary Computing (cs.NE); Machine Learning (cs.LG)
*备注: 42 pages, 4 figures
Abstract:We present an overview of R package called \textttnaspaclust, a tool for spatial fuzzy clustering improvement by hybridising the Fuzzy Geographically Weighted Clustering (FGWC) with nature-inspired optimisation algorithms. There are seven optimisation algorithms in general, namely Artificial Bee Colony, Flower Pollination Algorithm, Gravitational Search Algorithm, Harris-Hawk Optimisation, Intelligent Firefly Algorithm, Particle Swarm Optimisation, and Teaching-Learning Based optimisation. The optimisation algorithms available do not only use the basic algorithms, but also the modified version of several parts such as randomisation terms and updating rules. Furthermore, we also provide additional options by providing more distance metrics options based on the \texttt rdist package. This study showed the implementation of the \textttnaspaclust package in various cases, as well as comparing the performance of the algorithms in the scope of basic algorithms. The \textttnaspaclust package allows to improve the performance of fuzzy clustering, mainly in spatial context.
[LG-170] Probabilistic Simulation of Gas-Liquid Interfaces with Coupled Flow Matching on Spherical Harmonics
链接: https://arxiv.org/abs/2610.16327
作者: Kiet Bennema ten Brinke,Rachna Ramesh,Douwe Orij,Ivo Roghair,Vlado Menkovski
类目: Machine Learning (cs.LG); Fluid Dynamics (physics.flu-dyn)
*备注: 15 pages, 8 figures, 3 tables
Abstract:Bubbly flows involve strongly coupled bubble motion, deformation, and hydrodynamic interactions, making high-fidelity simulation computationally expensive. We propose a neural surrogate that jointly predicts bubble trajectories and interface deformation in a compact, shared representation, replacing costly mesh-resolved simulation with fast generative inference. Because two-phase flow dynamics is nonlinear and chaotic, we treat forecasting as a conditional generative problem rather than deterministic regression, predicting entire future windows in a single pass. We train and evaluate the surrogate on BubbleSH, a Front-Tracking dataset of rising deformable bubble swarms, assessing performance using trajectory, shape, and distributional metrics over kinematic, morphological, and interaction-based properties. The surrogate reproduces key flow statistics while running several orders of magnitude faster than the underlying DNS solver.
[LG-171] Fast Quantized Johnson-Lindenstrauss Transforms
链接: https://arxiv.org/abs/2610.16289
作者: Angelos Korakitis,Vasileios Nakos
类目: Machine Learning (cs.LG); Data Structures and Algorithms (cs.DS)
*备注:
Abstract:Quantization has emerged as one of the primary techniques for reducing the memory footprint of large language models during inference. A prominent example is Quantized Johnson - Lindenstrauss (QJL), which compresses KV caches using 1-bit asymmetric quantization while approximately preserving the inner products required by the attention mechanism. Recent theoretical and practical work aims at replacing the dense Gaussian projections used by the aforementioned transforms with more structured embeddings - such as compositions of randomized Hadamard transforms - while retaining the same embedding dimension. In this paper, we introduce two constructions based on a composition of randomized Hadamard transforms, albeit with a Gaussian rather than Rademacher diagonal in the second stage. The first adds Gaussian dithering to smooth the quantizer, while the second inserts a random permutation to regularize the covariance structure of the resulting Gaussian measurements. We prove embedding-dimension and running-time guarantees analogous to those of Fast JL in the non-quantized setting. The first construction achieves m=O(\varepsilon^-2\log^2(d/\delta)) , while the second achieves m=O(\varepsilon^-2\log(d/\delta)) under a restriction on m . Both constructions can be applied in O(d\log d) time. Subjects: Machine Learning (cs.LG); Data Structures and Algorithms (cs.DS) Cite as: arXiv:2610.16289 [cs.LG] (or arXiv:2610.16289v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.16289 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-172] Spectral Considerations on Unsupervised Time-domain EEG Representation Learning with Deep Networks
链接: https://arxiv.org/abs/2610.16264
作者: David Aquilué-Llorens,Abel Puigseslloses,Angel Blanco,Claire Braboszcz,Aureli Soria-Frisch
类目: Machine Learning (cs.LG)
*备注:
Abstract:Electroencephalography (EEG) is a non-invasive, scalable source of quantitative biomarkers for the early diagnosis of brain disorders, and deep representation learning has become a powerful paradigm for extracting information-rich latent features from raw signals. However, neural networks tend to under-represent high-frequency information, a concern that is especially acute for EEG given its oscillatory nature and characteristic 1/f^\chi power decay, which risks masking disease-relevant high-frequency signatures. In this study, we systematically investigate the spectral constraints of unsupervised time-domain feature extraction across three autoencoder architectures (LSTMVAE, VAEEG, Transformer) and a state-of-the-art EEG foundation model. Using synthetic experiments and empirical validation, we show that while autoencoders capture macro-scale shifts such as alpha slowing, they remain biased against higher frequencies due to a magnitude-driven masking effect induced by the Mean Squared Error loss under 1/f-decaying spectra. Increasing latent dimensionality mitigates this attenuation but sacrifices latent compactness and generative stability. We argue that, as the field advances toward large-scale foundation-level feature extractors, it is imperative to move beyond simple time-domain reconstruction and to integrate spectral-domain loss terms and gradient-balancing techniques, so that the subtle high-frequency transients critical for diagnosing complex brain disorders are preserved in the latent representation.
[LG-173] From Simulated Fluids to Living Tissue: Zero-Shot Foundation Model Transfer to Perturbed Embryos
链接: https://arxiv.org/abs/2610.16236
作者: Louisa Cornelis,Haewon Jeong,Irina Espejo,Payel Mukhopadhyay
类目: Machine Learning (cs.LG); Biological Physics (physics.bio-ph); Quantitative Methods (q-bio.QM)
*备注: 19 pages, 13 figures
Abstract:Morphogenesis is driven by the interplay of genetic regulation and tissue mechanics. In vivo video data is limited, particularly under genetic mutations and environmental perturbations, making generalization beyond the most representative conditions in the dataset challenging. However, at tissue scales, the dynamics of Drosophila can be described as a continuum flow, motivating the usage of representations learned across other physical systems as priors for developmental dynamics. We investigate finetuning Walrus, a Continuum Dynamics Foundation Model pretrained across a broad set of simulated continuous systems, on tissue velocity fields from only 12 wild-type Drosophila blastoderm videos. We evaluate autoregressive prediction both in-distribution and zero-shot under unseen temperature and genetic perturbations. Zero-shot refers to evaluation on perturbation conditions never seen during wild-type finetuning, without any additional training. In-distribution, a Fourier Neural Operator model trained from scratch achieves best performance on VRMSE. However, Walrus (CRPS) is the strongest model under the most severe distribution shift of unseen samples of mutated embryos and temperature conditions. Interestingly, predictions capture the shape of the mean surface tissue velocity trajectory and its separation between 17°C and 27°C, including the timing of the peak at Germ Band Extension onset, a biologically meaningful milestone of embryo development. Our results provide preliminary support that representations learned from different dynamical systems can transfer to experimental morphogenesis in developmental regimes, provided they admit a continuum description. More broadly, our results motivate further study of physics foundation models as a new complementary tool for the simulation of morphogenesis in low-data regimes. Comments: 19 pages, 13 figures Subjects: Machine Learning (cs.LG); Biological Physics (physics.bio-ph); Quantitative Methods (q-bio.QM) Cite as: arXiv:2610.16236 [cs.LG] (or arXiv:2610.16236v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.16236 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-174] NetRL: Powering Network Decisions via NetGRPO-Driven Reinforcement Learning
链接: https://arxiv.org/abs/2610.16226
作者: Lingfeng Peng,Xiaohui Xie,Yang Bao,Yong Cui
类目: Networking and Internet Architecture (cs.NI); Machine Learning (cs.LG)
*备注: 10 pages
Abstract:Network decision-making tasks, including adaptive bitrate streaming, load balancing, and congestion control, require policies to process heterogeneous states and actions while balancing near- and long-term consequences. Existing solutions generalize poorly across network tasks, require substantial task-specific redesign, and lack reliable evaluation of long- horizon action consequences. Methods that learn through online interaction additionally face unstable optimization. We present NetRL, a two-stage LLM-based reinforcement learning framework that combines supervised fine-tuning on demonstration trajectories with online policy optimization. A universal mixed-embedding interface combines textual task descriptions with precise numerical states, while a network action head produces valid decisions. At its core, NetGRPO evaluates candidate actions over multiple future trajectories, directs deeper search toward uncertain alternatives through confidence-aware budgeting, and performs group-relative, clipped, and KL-regularized updates. NetRL achieves 23.6% higher QoE than NetLLM, 12.2% lower job completion time than Genet, and 24.1% higher network utility than Orca, yielding state-of-the-art performance among the evaluated methods. Confidence-aware budgeting reduces end-to-end training time by 27.1% without reducing validation QoE. NetRL also converges stably and generalizes to unseen network traces without retraining. Code and model weights are available at the repository linked in this paper. Comments: 10 pages Subjects: Networking and Internet Architecture (cs.NI); Machine Learning (cs.LG) Cite as: arXiv:2610.16226 [cs.NI] (or arXiv:2610.16226v1 [cs.NI] for this version) https://doi.org/10.48550/arXiv.2610.16226 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Lingfeng Peng [view email] [v1] Mon, 12 Oct 2026 06:03:33 UTC (583 KB) Full-text links: Access Paper: View a PDF of the paper titled NetRL: Powering Network Decisions via NetGRPO-Driven Reinforcement Learning, by Lingfeng Peng and 3 other authorsView PDFHTML (experimental)TeX Source view license Additional Features Audio Summary Current browse context: cs.NI prev | next new | recent | 2026-10 Change to browse by: cs cs.LG References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[LG-175] When Participant Disjoint Validation Is Not Enough to Establish Generalization in Parkinsons Gait Classification
链接: https://arxiv.org/abs/2610.16220
作者: Md. Sifat,Sania Akter,Akif Islam,Md. Ekramul Hamid
类目: Machine Learning (cs.LG)
*备注: 9 pages, 5 figures. Includes supplementary analyses integrated into the appendices
Abstract:Participant-disjoint cross-validation prevents direct participant overlap, but it may still produce optimistic estimates when acquisition site and diagnosis are structurally dependent. We examined Parkinson’s disease classification using 159 pressure-walkway summary features from 133 WearGait-PD participants, including 65 participants with Parkinson’s disease and 68 controls. The cohort was non-factorial: two sites contained only one diagnostic group, whereas VA-Seattle was the only site containing both groups. In repeated participant-level nested cross-validation, site alone achieved an area under the receiver operating characteristic curve (AUC) of 0.955 plus or minus 0.003, exceeding the gait-feature model (0.882 plus or minus 0.014). Under the evaluated logistic-regression pipelines, adding gait features to site, age, and sex produced no measurable improvement in pooled discrimination. More site-controlled evaluations produced lower and imprecise estimates: VA-Seattle same-site AUC was 0.668 (95% bootstrap confidence interval 0.477 to 0.836), and the JHOC + Bayview-to-VA-Seattle transport stress test achieved 0.576 (0.375 to 0.770). Supporting within-diagnosis site-prediction and site-conditioned permutation analyses showed that the gait representation retained substantial acquisition-site structure. These findings indicate that participant-disjoint pooled validation did not isolate disease-related discrimination from site-associated variation. They encourage careful examination of site-label dependence and, where possible, support for pooled results through within-site or independent validation before broader claims are made about generalization across institutions.
[LG-176] MSC: Multi-Stage Caster-aware Control Framework with Uncertainty-Aware NMPC and RL-based Execution Layer IROS2026
链接: https://arxiv.org/abs/2610.16199
作者: Jinho Park,Jeongtae Lee,Jeonghyeok Lim,Keun Ha Choi,Kyung-Soo Kim
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: 8 pages, 6 figures, IROS 2026
Abstract:Freely swiveling casters are widely adopted in industrial AMRs/SDVs for their mechanical simplicity and load capacity. However, caster misalignment during low-speed or high-curvature maneuvers induces strong contact nonlinearities, widening the planning-execution gap between commanded (v, \omega) and realized motion. Existing caster-aware strategies are highly sensitive to state-estimation reliability, causing unstable execution in uncertain transition regimes. To address this, we present Multi-Stage Caster-aware Control (MSC), a planning-execution integrated hierarchical framework featuring two core mechanisms. First, an uncertainty-aware NMPC (CU-MPC) dynamically schedules caster-aware costs and input regularization using predictive uncertainty to suppress overly aggressive maneuvers. Second, a reinforcement learning-based execution layer estimates and predicts caster states under partial observability, transforming high-level velocity commands into physically executable wheel speeds adapted to current contact conditions. Validations in Isaac Sim and real robots demonstrate that MSC offers a clear comparative advantage over existing decoupled compensation approaches by significantly improving caster-state estimation accuracy, minimizing misalignment, and enhancing tracking robustness in caster-sensitive maneuvers.
[LG-177] BARTER: Compact Dynamic Sparse Training with Standard Dense Kernels
链接: https://arxiv.org/abs/2610.16147
作者: Yueyuan Sui,Payal Mohapatra,Qi Zhu,Stephen Xia
类目: Machine Learning (cs.LG)
*备注:
Abstract:Dynamic sparse training (DST) aims to reduce training costs while maintaining accuracy. Existing DST methods mask out parameters but keep matrix dimensions unchanged. Training speedups therefore depend on specialized support, such as sparse kernels. We introduce BARTER, a structured DST framework that reduces training cost using standard dense kernels by keeping training tensors physically compact while dynamically reallocating capacity across layers. Unlike existing works that require computing gradients for all parameters, BARTER instead uses statistics collected only from active units during ordinary backpropagation to prune and clone attention heads and feed-forward channels under separate fixed global budgets. At each topology update, it resizes the affected parameter tensors and remaps their optimizer states, so subsequent training steps operate directly on the compact model. Training and fine-tuning experiments with Transformer-based vision models on two image classification benchmarks at densities from 0.1 to 0.7 show competitive Top-1 accuracy relative to full-tensor DST baselines. On an NVIDIA L40, BARTER achieves speedups of up to 6.29\times in measured training segments and reduces peak allocated GPU memory by up to 81.3% relative to these implementations. An iPhone 13 Pro Max training benchmark additionally shows a 2.78\times speedup over the full-tensor DST baseline.
[LG-178] XForecast: The KDD 2026 Finance Day Challenge on News-Grounded Financial Forecasting
链接: https://arxiv.org/abs/2610.16131
作者: Jaehoon Lee,Jun Seo,Seunghan Lee,Tae Yoon Lim,Dongwan Kang,Hwanil Choi,Minjae Kim,Sungdong Yoo,Junhyeok Kang,Sangjun Han,Soonyoung Lee,Yongjae Lee,Wonbin Ahn
类目: Machine Learning (cs.LG)
*备注: 6 pages
Abstract:Financial forecasting has traditionally relied on numerical time-series (TS) signals, while textual information such as news remains largely underutilized. We introduce XForecast, a benchmark and competition for news-grounded financial forecasting, where the task is to predict stock price movements four weeks ahead using both historical price data and paired news signals. XForecast combines stock price data with curated financial text generated using FinTexTS, a multimodal dataset generation method. It covers 100 large-cap stocks from 2019–2023, with training on 2019–2022 and evaluation on 2023. To better evaluate models that leverage textual signals, we adopt a weighted hit rate metric that places greater emphasis on instances where news information is expected to have stronger influence. The XForecast Challenge, hosted on Kaggle as part of KDD 2026 Finance Day, aims to benchmark methods for integrating financial TS and textual information.
[LG-179] Silence and Noise: Resolving Suppression and Overvaluation in Multimodal Entity Alignment
链接: https://arxiv.org/abs/2610.16106
作者: Xizhe Zhang,Jingfeng Zhang,Meng-Fen Chiang
类目: Machine Learning (cs.LG)
*备注: 26 pages, including appendices. Code: this https URL
Abstract:Multimodal entity alignment (MMEA) is fundamental to integrating heterogeneous knowledge graphs (KGs), yet existing paradigms remain critically vulnerable to incomplete modalities and cross-graph structural asymmetries. This paper exposes a fundamental oversight in current MMEA research: treating feature missingness and topological discrepancies as isolated preprocessing or denoising challenges, which overlooks their destructive coupling during multimodal fusion. We systematically formulate and evaluate two newly identified failure modes: 1) Detail Missing (DM)-induced overvaluation, where the absence of local modality details causes models to overemphasize uninformative, blurred representations; and 2) Context Imbalance (CI)-induced suppression, where topological discrepancies across KGs allow dense structural contexts to overwhelm weaker but highly discriminative modal signals, triggering high-level semantic drift. To break this cycle of degradation, we introduce ALMEA, a generative and calibration framework that establishes a bidirectional feedback loop between low-level feature reconstruction and high-level latent space alignment. Specifically, Cross-Modality Imputation (CMI) synthesizes missing modality details to mitigate overvaluation, while Latent Semantic Calibration (LSC) aligns cross-graph joint embeddings to suppress topology-driven semantic drift. Under sparse supervision, we develop Controlled Iterative Selection (CIS) to adaptively acquire representative and diverse pseudo-labeled pairs, optimizing latent calibration under constrained budgets. Extensive evaluations on standard benchmarks demonstrate that ALMEA consistently outperforms state-of-the-art baselines. At a 20% supervision rate, ALMEA yields substantial absolute improvements of up to 5.20% in MRR and 6.40% in Hits@1, maintaining robust gains across higher supervision thresholds.
[LG-180] Beyond Policy Rollouts: Embedding-Space Directional Exploration with Anchored Noise for LLM RLVR
链接: https://arxiv.org/abs/2610.16105
作者: Daijiao Liu,Huiyi Wang,Lina Yao,Dong Gong
类目: Machine Learning (cs.LG)
*备注:
Abstract:Reinforcement learning with verifiable rewards (RLVR) typically uses small rollout groups per prompt, where repeated sampling from the same policy can yield highly similar trajectories, reducing within-group reward variation and weakening group-relative advantage signals. Existing remedies such as higher sampling temperature, entropy bonuses, asymmetric clipping, and larger rollout budgets broaden sampling or modify policy updates but leave rollout conditioning unchanged. In this paper, we propose EDEN (Embedding-space Directional Exploration via anchored Noise), which intervenes at the rollout source through anchored input perturbations. EDEN explores beyond the policy’s own rollouts. Within the same rollout budget, it replaces a small fraction of standard rollouts in each group with rollouts that the same policy generates from the prompt plus a single appended, perturbed embedding position, rather than from the prompt alone. Gaussian noise drives variation, while a shared anchor provides a common conditioning direction within each prompt group. Selected perturbed rollouts contribute to GRPO updates with downweighted advantages, and the model architecture, discrete decoding, and test-time inference remain unchanged. Across model scales and mathematical reasoning benchmarks, EDEN outperforms GRPO and the exploration controls evaluated in the same pipeline, with the largest gains in sampled solution coverage. Analyses show that more prompt groups provide non-zero group-relative learning signals and that noise and anchoring play complementary roles. Intervening at the rollout source is thus an effective lever for exploration in RLVR.
[LG-181] SelectTTRL: Representation-Guided Data Selection for Efficient Test-Time Reinforcement Learning
链接: https://arxiv.org/abs/2610.16104
作者: Yi Li,Linan Yue,Weibo Gao
类目: Machine Learning (cs.LG)
*备注: 22 pages, 1 figure
Abstract:Test-Time Reinforcement Learning (TTRL) enables large language models to adapt directly on unlabeled test problems by constructing pseudo-labels from multiple sampled responses. However, existing TTRL methods typically apply the same expensive sampling and optimization procedure to all test problems, despite substantial differences in their training utility. In this paper, we study whether such problems can be identified before costly test-time training, without access to ground-truth answers. We uncover a stable relationship between internal response representations and problem-level empirical accuracy. Specifically, we define COS as the cosine similarity between a model’s response representation and a correctness-related direction constructed from an independent question-answering dataset. COS consistently correlates with problem solvability across different models, datasets, and experimental settings, and we further provide a theoretical analysis explaining this relationship. Building on this finding, we propose SelectTTRL, a representation-guided data selection method that ranks test problems using COS and applies TTRL only to a selected percentile interval. SelectTTRL leaves the standard TTRL pipeline unchanged for the selected problems, while avoiding unnecessary multi-response sampling and reinforcement learning updates on the remaining ones. Experiments across multiple reasoning benchmarks demonstrate that SelectTTRL substantially reduces test-time computation while maintaining or improving the performance of standard TTRL. Code is available at this https URL.
[LG-182] Memorization in image classification: test set behavior and scaling laws NEURIPS2026
链接: https://arxiv.org/abs/2610.16090
作者: Jeevesh Attri,Florentin Guth
类目: Machine Learning (cs.LG)
*备注: 11 Pages, 4 figures, Accepted at the NeurIPS 2026 Workshop on Neural Network Artifacts as a New Data Modality
Abstract:Deep neural networks are capable of both learning generalizing solutions on natural tasks and memorizing arbitrary mappings. We argue that understanding the former requires understanding the latter, which is much less studied. In what ways do networks that generalize or memorize differ in their behavior or their training dynamics? Do memorizing networks simply function as a lookup table where input-output mappings are stored independently? We investigate these questions in the context of CNNs trained on image classification tasks with randomized labels or pixels. We first study the behavior of such memorizing networks on unseen images. We find that independently initialized networks trained on the same random labels learn different predictive functions: they make different predictions on test examples, in contrast with generalizing networks. We then examine the effect of dataset size on memorization through the required network capacity and training time to interpolate the training set. Counter-intuitively, we find that the number of presentations of a given example required for complete memorization can decrease with dataset size even when input-output pairs share no common structure. Although memorizing networks do not learn a universal solution, they exhibit interesting behaviors which call for further investigations into the underlying mechanisms.
[LG-183] Interpretable Graph Partitioning for Spatial Prediction
链接: https://arxiv.org/abs/2610.16063
作者: Yuta Shikuri
类目: Machine Learning (cs.LG)
*备注:
Abstract:Spatial prediction estimates unobserved values from spatially distributed observations. Although flexible predictive models are increasingly available, their practical use remains limited when interpretability is essential. In this study, we formulate a graph partitioning problem to construct a global surrogate model that approximates black-box predictions with simpler representations. Specifically, we select connected subgraphs to minimize within-cluster residuals in vertex-level predictions. Although this formulation can be viewed as a variant of spatial regionalization, it operates on model-based predictions at individual locations rather than predefined areal units. With many locations, this point-level formulation may produce artificial connections between regions, reducing interpretability. We therefore introduce a constraint that excludes such connections while retaining flexible partition boundaries. We theoretically analyze how this constraint affects the resulting partition. We also develop a computational procedure for adjusting partition boundaries. Experimental results show that the proposed method yields spatially coherent partitions with flexible boundaries.
[LG-184] How the Classical Baseline Shapes Machine-Learning Speed Claims in Topology Optimization
链接: https://arxiv.org/abs/2610.16051
作者: Shaoliang Yang,Jun Wang
类目: Computational Engineering, Finance, and Science (cs.CE); Machine Learning (cs.LG)
*备注: 31 pages, 8 figures, 10 tables. Research article: tests of released code for eight topology-optimization methods and an exploratory audit of published speed claims
Abstract:Topology optimization places material so that a structure resists deformation under given loads. Machine-learning methods often report large speed-ups over classical, non-learned optimization, but a slow classical code or a less stiff learned design can exaggerate them. We re-ran released code for eight methods on one workstation and studied two in depth: Optimize Any Topology (OAT), a diffusion model that generates designs, and a graph-based neural optimizer. Speeding up only the classical code’s finite-element solver cut these two methods’ speed-ups by up to 20-fold. We then timed the classical code until its design was at least as stiff as the learned one; for OAT, we read this point off the completed runs. Published timings imply a 134x speed-up for OAT on a 256 x 256 grid; our matched single-design comparisons gave 1.5x to 3.5x. At matched stiffness, the graph-based optimizer was slower than the classical code in all six two-dimensional problems. Of the six other methods, two kept their advantage, one shrank, one reversed and two were not comparable. In an exploratory audit of published claims, language models judged that 8 of 28 speed comparisons in randomly sampled, openly accessible articles lacked evidence of equal design quality; our own review counted 11, so the count depends on who judges. We provide a comparison procedure and a minimum report: classical code, hardware, design quality, what each timing includes, and data-generation and training costs.
[LG-185] Backward Conformal Prediction under Covariate Shift
链接: https://arxiv.org/abs/2610.16048
作者: Amin Meghrazi,Yuanyuan Li
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 26 pages, 2 figures, 19 tables
Abstract:Backward conformal prediction (BCP) fixes the size of a prediction set and estimates its miscoverage, but assumes calibration and test data are exchangeable. We extend BCP to covariate shift through importance weighting, preserving its guarantees when the density ratio is known. We show that BCP’s tight miscoverage report depends only on the relative weight assigned to missed calibration examples, explaining why a better domain classifier can sometimes produce a worse estimate. Motivated by the fact that BCP depends only on nonconformity scores, we propose estimating density ratios on sorted score vectors rather than high-dimensional inputs. We prove that this approach recovers the same report asymptotically under an interpretable condition and characterize the resulting bias otherwise. Experiments with simulated and benchmark distribution shifts show that the proposed method improves BCP’s miscoverage estimates, while score-based weighting consistently yields more accurate estimates than embedding-based weighting.
[LG-186] CLNO: Conservative Local Neural Operator for Compressible Flow Prediction
链接: https://arxiv.org/abs/2610.16040
作者: Buxuan Wang,Rongzu Li,Qingsong Zou
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG)
*备注: 23pages, 15 figures
Abstract:We propose a conservative local neural operator (CLNO) for accurate and generalizable autoregressive prediction of compressible flows. Building upon the locality and translation invariance of the local neural operator (LNO), CLNO introduces a conservative evolution framework based on local subdomains of fixed physical size, separating local mean-state evolution from fine-scale reconstruction. The flow is encoded into subdomain-integrated conservative states and fine-scale features, while a shared local network predicts interface fluxes for finite-volume updates that enforce discrete conservation. A convolutional reconstruction network recovers fine-scale structures using zero-mean basis functions, preserving the updated local means and providing information for subsequent evolution. To accommodate different physical parameters, Reynolds–Mach number conditioning modulates the encoding, flux-prediction, and reconstruction networks, enabling a single model to learn parameter-dependent dynamics and generalize to unseen parameter combinations without retraining. CLNO supports prediction across spatial resolutions, computational domain sizes, and boundary conditions. Numerical experiments demonstrate improved generalization and competitive or higher accuracy compared with LNO, including order-of-magnitude error reductions in representative cases, at approximately three to five times the inference cost on periodic benchmarks. These results demonstrate accurate, generalizable, and physically consistent flow prediction through conservative local evolution and fine-scale reconstruction.
[LG-187] Certified Robustness via Sparse Second-Order Cone Relaxations
链接: https://arxiv.org/abs/2610.16016
作者: Mohammadreza Maleki,Daniel Sadig,Rushendra Sidibomma,Mahyar Fazlyab,Arman Adibi,Reza Samavi
类目: Machine Learning (cs.LG)
*备注:
Abstract:Certified robustness is critical for deploying neural networks in safety-sensitive settings, but computing exact certificates typically requires solving nonconvex optimization problems. Practical verifiers use convex relaxations: linear programming (LP) relaxations are scalable but often too conservative, whereas semidefinite programming (SDP) relaxations can be substantially tighter but are computationally expensive. We introduce a hierarchy of second-order cone programming (SOCP) relaxations that provide stronger certificates than LP while avoiding the cost of global positive-semidefinite constraints. Our construction combines convex-hull relaxations of ReLU activations, selective McCormick lifting over certified variable bounds, and second-order cone constraints induced by 2-by-2 principal minors. Adding these constraints progressively yields a monotone hierarchy of increasingly tight bounds with a controllable computation-tightness trade-off. On fully connected and convolutional networks on MNIST and CIFAR-10, SOCP-based verification improves certified robust accuracy over LP-based verification and is competitive with alpha,beta-CROWN, while remaining orders of magnitude faster than SDP-based verification. These results position selective second-order constraints as a practical middle ground between scalable LP-based and tight SDP-based certification. Finally, we introduce class-optimized robust evaluation (CORE), which applies tighter relaxations only to target classes not certified by cheaper ones. On the fully connected MNIST benchmark, CORE variants match SDP-cert’s certified robust accuracy while reducing average verification time by up to 86.30%.
[LG-188] A 4-Qubit Variational Quantum Classifier for Sequence-Specific Classification of DNMT3A and DNMT3B Flanking Methylation Preferences
链接: https://arxiv.org/abs/2610.15998
作者: Murat Isik,Yusuf Moinuddin,Humaira Gowher,Sabre Kais
类目: Machine Learning (cs.LG)
*备注: Accepted at IEEE International Conference on Quantum Artificial Intelligence (QAI) 2026
Abstract:DNA methylation is a critical epigenetic modification that regulates gene expression and genome stability, primarily catalyzed by the DNA methyltransferase 3 (DNMT3) family of enzymes. Distinguishing precise sequence-specific flanking preferences between DNMT3A and DNMT3B remains computationally challenging due to the complex and high-dimensional nature of biochemical sequence data. In this study, we propose a novel Quantum Machine Learning (QML) approach to classify DNA methylation preferences, leveraging a Variational Quantum Classifier (VQC). Using experimentally derived deep enzymology data, we constructed a dataset containing 4,096 unique NNNCGNNN motifs. We evaluated a four-qubit VQC using a custom NumPy exact-statevector simulator. The described two-layer circuit contains 24 trainable rotations, eight CNOT gates, and eight input-encoding rotations, totaling 40 gates before measurement. Across five initialization seeds, the revised VQC evaluation achieved a ROC-AUC of 0.661 pm 0.009, where variability denotes sample standard deviation. The held-out test partition is approximately balanced, containing 200 DNMT3A-labeled motifs and 210 DNMT3B-labeled motifs. An always-DNMT3B predictor achieves 51.2% accuracy and a balanced accuracy of 0.500 on this test partition. This proof-of-concept establishes a foundation for utilizing near-term quantum architectures in complex epigenetic classification tasks, paving the way for hybrid quantum-classical pipelines in bioinformatics.
[LG-189] Dissipative Lagrangian Neural Networks for the Interpretable Modelling of Passive Mechanical Systems
链接: https://arxiv.org/abs/2610.15996
作者: Rob Seabourne,Alice Cicirello
类目: Machine Learning (cs.LG); Computational Engineering, Finance, and Science (cs.CE)
*备注:
Abstract:In this paper, we propose the Dissipative Lagrangian Neural Network (dLNN), a continuous-time system identification technique that separately learns the conservative and dissipative dynamics of a mechanical system directly from data. A key contribution is a novel parametrisation of dissipative forces that (i) is structurally constrained to be dissipative, enforcing passivity of the learned model, (ii) cannot absorb conservative mechanics, and (iii) retains sufficient expressivity to represent a broad family of state-dependent damping and regularised friction forces. The model is validated on prediction tasks on two real-world mechanical systems: trajectory prediction of a chaotic double pendulum and inverse-dynamics modelling of a six-degree-of-freedom industrial robot. Across these tasks, the dLNN demonstrates improved predictive performance over existing Lagrangian architectures while remaining competitive with a black-box neural model. The imposed physical structure also improves generalisation to unseen states, with the benefits becoming particularly pronounced when training data is limited. Finally, we introduce a symbolic distillation procedure that extracts closed-form expressions from the learned physical components. Applied to the industrial robot, this produces interpretable symbolic models that retain high predictive accuracy while reducing the computational overhead associated with evaluating neural Lagrangian models.
[LG-190] Probabilistic Forecasting of Wildfire Size Using Conditional Flow Matching AACL ATC
链接: https://arxiv.org/abs/2610.15983
作者: Isaac Lee,Bryan Shaddy,Assad Oberai
类目: Machine Learning (cs.LG)
*备注: 18 pages, 7 figures, 1 table. Code and data: this https URL
Abstract:In this work we present a probabilistic model for predicting the temporal evolution of wildfire size using conditional generative algorithms, namely flow matching. Rather than predicting spatially resolved fire progression, the proposed model forecasts a time series of fire acreage, substantially reducing computational cost and enabling efficient integration with downstream applications such as early warning systems and risk assessment tools. Given estimates of the fire size at the current and previous forecasting steps, alongside spatially averaged weather conditions, terrain characteristics, and fuel composition, the model predicts the change in fire area over a 3-hour interval. Training data are drawn from 140 coupled atmosphere-wildfire simulations generated using WRF-SFIRE, paired with weather data from the North American Mesoscale (NAM) model. Sensitivity analysis demonstrates that the model captures physically-consistent responses to variations in wind speed, temperature, and relative humidity. Model performance is evaluated against reference solutions from WRF-SFIRE. For single-step predictions, the model achieves a symmetric mean absolute percentage error (sMAPE) of 0.39 relative to the reference WRF-SFIRE solutions. Recursive multi-step predictions over 24 hours are performed for 12 held-out test cases, where the flow matching model achieves a final fire area sMAPE of 0.39 while further providing meaningful quantification of forecast uncertainty.
[LG-191] Softmax-Free Attention in Spiking Transformers as the Zero-Temperature Limit of Two-Level Systems
链接: https://arxiv.org/abs/2610.15968
作者: Roberto C. Sotero,Pouya Taghipour,Jose M. Sanchez-Bornot
类目: Machine Learning (cs.LG)
*备注:
Abstract:Spiking Transformers achieve energy efficiency by discarding softmax normalization in favor of thresholded, binary attention. However, this removal breaks the thermodynamic framework used to analyze attention networks, as the normalized measure and partition function cease to exist. We restore this theoretical foundation by proving that softmax-free spiking attention operates exactly as the zero-temperature limit of a product of independent two-level systems. This correspondence places spiking attention strictly in the frozen ordered phase of the thermodynamic phase diagram for all inputs. Consequently, the entropic cost of organizing attention collapses, replaced by a Landauer-type physical floor governed by action-potential generation. Furthermore, this ordered-phase pinning mathematically explains the optimization difficulty caused by vanishing surrogate gradients. We validate this theory with forward-relaxation experiments, placing both standard and neuromorphic attention mechanisms on a unified thermodynamic axis.
[LG-192] Harmonic Representations in Symbolic Music Transformers are Localized and Controllable
链接: https://arxiv.org/abs/2610.15957
作者: Adi Silberschein,Megan Wei,Tamar Rott Shaham
类目: ound (cs.SD); Machine Learning (cs.LG)
*备注: 9 pages, 7 figures, 1 table. Accepted at ISMIR 2026 (27th International Society for Music Information Retrieval Conference)
Abstract:Transformer models trained on symbolic music have become remarkably capable. However, it is yet unclear how they represent music structure, and whether understanding their internal representations can be used to unlock methods for model control. We examine this through a case study of harmonic analysis: when a music transformer navigates a cadence, what does it internally “know” about its harmonic context, and can that knowledge be used to affect the generation? Using Aria, a LLaMA-based symbolic music transformer evaluated on Bach chorales, we find that harmonic decisions are causally localized progressively from middle layers onward. Linear probes show that mode becomes decodable in the same layer range where patching identifies growing causal effect, while key identity emerges later. We then confirm these representations are actionable for music generation control: steering via latent directions corresponding to harmonic changes reliably produces a deceptive or minor resolution, while preserving structural coherence and generation quality. This demonstrates that, at inference time, the model’s internal harmonic knowledge can be directly manipulated to control the harmonic output.
[LG-193] Instance-Adaptive Online Multiclass Multicalibration
链接: https://arxiv.org/abs/2610.15954
作者: Zhiming Huang,Jamie Morgenstern,Aaron Roth,Claire Jie Zhang
类目: Machine Learning (cs.LG)
*备注: 52 pages
Abstract:We study online multiclass multicalibration over T rounds, and give an algorithm with beyond-worst-case guarantees. With n possible outcomes, forecasts lie in a simplex of dimension d=n-1 . In the worst case, cumulative \ell_1 multicalibration error scales with \widetilde\Theta(T^(d+1)/(d+2)) , which becomes untenable once d exceeds a small constant. Our algorithm adapts to the geometry of the realized sequence: if the sequence of adversarially selected outcome means lies in an affine subspace of dimension s , it obtains error \widetilde O(n^1/(s+2)T^(s+1)/(s+2)) with high probability. The algorithm adapts to this structure without knowing the subspace or its dimension. For T\ge n\ge2 , this rate is minimax optimal up to polylogarithmic factors. For a stochastic instance with a fixed conditional mean, the bound specializes to \widetilde O(\sqrtnT) , giving a square-root dependence on the horizon.
[LG-194] Physics-guided machine learning for saturation control across variable hydraulic conditions in heap leaching
链接: https://arxiv.org/abs/2610.15947
作者: Sultan Alhamdan,Daniel Navia,Paulina Quintanilla
类目: ystems and Control (eess.SY); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Heap-leaching hydraulics vary between regions of an ore bed, motivating fast prediction models that accommodate different material properties. This study develops a parameter-conditioned physics-guided machine-learning (PGML) surrogate and demonstrates its use in simulated saturation control. A dataset of 4096 numerical Richards-equation trajectories was divided into 2867 training, 614 validation and 615 evaluation cases. The network predicts hourly saturation profiles from the current profile, irrigation rate and four hydraulic parameters, and is trained with an approximate Gardner mass-balance penalty. A daily predictive controller compared 200 constant irrigation rates over a 24-hour horizon. Across the evaluation cases, the median full-cycle tracking deviation was 4.9% of a 0.70 saturation target, equal to that of a tuned proportional-integral (PI) controller, while the PGML controller reached the target without overshoot in the median and applied approximately 4% less irrigation. Moderate physics guidance reduced long-horizon prediction error by approximately 30% relative to data-only training and reduced irrigation-bound activity, but slightly worsened closed-loop tracking. For a matched prediction workload, the surrogate was approximately 70 times faster than the numerical solver, and batched evaluation cost under 2 ms per column for 5000 columns. The results support a reusable physics-guided surrogate for large-scale saturation control in simulation.
[LG-195] Error-Based Bidirectional Predictive Coding Enables Deep 4-Bit Predictive Coding Networks
链接: https://arxiv.org/abs/2610.15933
作者: Garrett Kinman,Zeljko Zilic
类目: Machine Learning (cs.LG)
*备注:
Abstract:Predictive coding networks (PCNs) promise properties not offered by backpropagation-trained networks: learning from purely local information, a single network that both discriminates and generates, and flexible inference by energy minimization. However, their iterative inference is expensive, making hardware acceleration desirable, and a useful, efficient accelerator needs both depth and aggressive quantization. Canonical state-based PC delivers neither: its inference signal decays exponentially with depth, so deep models fail to train, and low precision makes it worse. We show that pairing two recent methods carries the full PCN property set into the deep, 4-bit regime: bidirectional PC (bPC), which unifies discrimination and generation, and the error-based reparameterization (ePC), which optimizes prediction errors rather than states during inference. Error optimization delivers the output signal to every layer in a single backward pass, removing the signal decay; one additional element, global-norm clipping of the error gradients, stabilizes the error dynamics at depth. The resulting ePC+bPC model trains stably up to at least eight layers and retains its full-precision accuracy and clean class-conditional generation under W4A4E4 quantization-aware training, across FP4 and three posit formats. On six-layer networks, where state-based bPC collapses to chance, ePC+bPC classifies more than 50 accuracy points above it after one epoch of training, with class-conditional generations 5-9x better in RMSE. 4-bit quantization is nearly free: full-precision and 4-bit accuracies differ by at most ~5 points, and by under two in the best-performing formats. We characterize the two failure modes the method overcomes, bounded signal decay in state-based PC and error divergence in unregularized error optimization, and discuss implications for a 4-bit-native PCN accelerator.
[LG-196] Mesh-Free Convolution: Learned Spectral Attenuation and Transport NEURIPS2026
链接: https://arxiv.org/abs/2610.15931
作者: Niloufar Zakariaei,Manav Manav,Eldad Haber
类目: Machine Learning (cs.LG)
*备注: Accepted at NeurIPS 2026
Abstract:Learning solution operators for partial differential equations (PDEs) requires models that can represent spatial structure across grids and scales. This becomes especially important in time-dependent problems, where the learned operator is often used autoregressively: each predicted state is fed back as input to predict future states. In this setting, small spatial errors can be repeatedly propagated through the learned dynamics, leading to instability and drift, vitiating long-term solution fidelity. We propose Mesh-Free Convolutions (MFCs), filters defined in continuous space and discretized only at evaluation time. This allows the same learned filters to be applied across resolutions. Rather than imposing a grid-dependent stencil or a fixed spectral cutoff, MFCs learn smooth scale-dependent filtering and include directional terms for transport-like behavior. Across fluid-dynamics and weather-forecasting benchmarks, MFC-based models produce stable long-horizon rollouts, transfer reliably across resolutions, and better preserve long-time statistics. These results suggest that continuous, mesh-free filtering is a useful inductive bias for PDE operator learning, especially in autoregressive prediction. Comments: Accepted at NeurIPS 2026 Subjects: Machine Learning (cs.LG) Cite as: arXiv:2610.15931 [cs.LG] (or arXiv:2610.15931v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.15931 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-197] Fast and Accurate: Improving Personalized LLM Performance by Semantic Concentration Inspired Workload Split between RAG and Fine-Tuning
链接: https://arxiv.org/abs/2610.15921
作者: Xingke Yang,Sicong Li,Lingfeng Yao,Rui Chen,Changqing Luo,Xin Fu,Miao Pan
类目: Machine Learning (cs.LG)
*备注:
Abstract:With ever increasing personalized large language model (LLM) services, the next question to answer is how to improve the quality of services/users’ experiences, which is indicated by personalized LLM generation accuracy/quality and inference latency. Although state-of-the-art designs, e.g., parameter-efficient fine-tuning (PEFT), retrieval-augmented generation (RAG) or Full PEFT+RAG, can help, they largely ignore the heterogeneous semantic concentrations of users’ historical data and cannot fully unleash the potential of personal data for personalized LLM performance enhancement. In this paper, we propose PEFT Union RAG (PUR), a novel fast and accurate'' approach that exploits the semantic concentration of users' personal historical data to split workload between PEFT and RAG, aiming to improve personalized LLM performance in terms of better generation accuracy/quality and lower inference latency. Briefly, PUR partitions user's personal historical data and assigns semantic concentrated core’’ information to PEFT for offline parametric fine-tuning, while retaining dispersed information for RAG. During the online LLM inference, PUR directly generates responses for core'' queries covered by the PEFT region and invokes RAG only for sporadic peripheral’’ queries. Experimental results show that, by appropriately splitting workload between PEFT and RAG, PUR consistently achieves the best generation accuracy/quality for various users across multiple tasks, while reducing inference latency by 14.4%–46.6% compared with the best performed method.
[LG-198] Sample-Optimal Multi-Group Learning with an ERM Oracle
链接: https://arxiv.org/abs/2610.15916
作者: Princewill Okoroafor
类目: Machine Learning (cs.LG); Data Structures and Algorithms (cs.DS); Computer Science and Game Theory (cs.GT)
*备注:
Abstract:In multi-group learning, a single predictor must perform nearly as well as the best hypothesis in a class on every group in a rich, possibly infinite and overlapping, collection of subpopulations. The known algorithms with the optimal sample complexity keep track of every pair of a group and a hypothesis, which is impossible for the large classes of interest, and the known oracle-based algorithms have a worse rate. We ask whether the optimal rate can be attained by a learner that accesses the classes only through an empirical risk minimization (ERM) oracle. We give an algorithm that uses \tildeO((d_G+d_H)/\varepsilon^2) samples and polynomially many calls to a weighted ERM oracle over pairs of a group and a hypothesis, where d_G and d_H are the VC dimensions of the group and hypothesis classes and \varepsilon bounds the excess risk on every group, weighted by its probability. We also show a matching lower bound of \Omega((d_G+d_H)/\varepsilon^2) samples for every learner even with unlimited computation, so oracle efficiency costs no additional samples. We also consider a natural special case in which every input belongs to at most k groups and these groups can be listed. In this case, an ERM oracle for the hypothesis class alone suffices, with \tildeO((d_H+\log k)/\varepsilon^2) samples in the agnostic setting and \tildeO((d_H+\log k)/\varepsilon) samples when every group has a hypothesis that is correct on it.
[LG-199] Lightweight Scheduling Policies for Continuously Batched LLLM Serving Across GPU Architectures
链接: https://arxiv.org/abs/2610.15899
作者: Ashman Mehra,Akshan Agarwal,Aryeh Carmi
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注:
Abstract:Large language model serving is increasingly limited not only by model size and GPU speed, but also by scheduling during autoregressive decoding, where continuous batching interleaves requests at token granularity as sequences arrive and finish. We study whether lightweight scheduling policies can improve latency-throughput tradeoffs without changing model architectures, kernels, or deploying learned schedulers. We implement a Python/PyTorch continuous-batching testbench with real Hugging Face execution and evaluate eight policies across 1,152 GPU-model-scheduler-workload runs using ShareGPT-derived traffic, six Qwen, Gemma, and Mistral models, and Modal L40S, A100, and H100 GPUs. Scheduler choice changes throughput, TTFT, completion latency, and tail behavior under fixed model, GPU, workload, batch size, and prefill budget. Aging shortest-prompt-first is the strongest practical baseline in many L40S and mixed-workload settings, while oracle remaining-length scheduling shows that job-size knowledge alone is insufficient because batching and hardware effects also shape performance. The custom dataset has been uploaded to Hugging Face.
[LG-200] A Hybrid Gaussian Process-Extreme Gradient Boosting Machine Learning Framework for Robust Knockdown Factor Prediction and Reliability-Based Design of Thin Spherical Shell Caps
链接: https://arxiv.org/abs/2610.15862
作者: Budhaditya De,Rohan Majumder,Devasmit Dutta,Aman Deep Gupta,Sudib Kumar Mishra
类目: Machine Learning (cs.LG)
*备注:
Abstract:Elastic buckling remains a critical concern in the design of deep spherical shell caps, given that experiments reveal a significant reduction of critical loads from those predicted by linearized eigen-buckling analysis. This discrepancy has been attributed to initial geometric imperfections and is often reconciled through a dimensionless parameter, namely the knockdown factor (KDF). However, conventional KDFs (NASA SP-8007; Eurocode 3) are highly conservative, as they represent the lower bound of an extensive body of experimental data. This study aims to narrow the conservatism in the KDFs for spherical shell caps through machine learning (ML) regression models trained on experimental data on spherical shell buckling. A hybrid model is proposed by combining Gaussian Process Regression (GPR) with Extreme Gradient Boosting (XGBoost), and its improved performance is demonstrated over conventional ML algorithms, i.e., the Artificial Neural Network (ANN), Support Vector Regressor (SVR), Random Forest Regressor (RFR), and Histogram-based Gradient Boosting (HGBoost). Each of these models uses six input features and a single output, i.e., the KDF. The most critical feature affecting the KDF is ascertained through a sensitivity analysis. The study concludes with a reliability-based design framework that relies on the predictive ML model to obtain safety-consistent, yet less conservative, KDFs.
[LG-201] GuidanceNFT: NFT-Based Reward Fine-Tuning of Guided Diffusion Models
链接: https://arxiv.org/abs/2610.15856
作者: Jiayin Liu,Zeyu Michael Li,Bingshuo Qian,Xiang Cheng
类目: Machine Learning (cs.LG)
*备注:
Abstract:DiffusionNFT (NFT) fine-tunes diffusion models toward task-specific rewards, but optimizing these rewards does not guarantee fidelity to the full prompt. Although classifier-free guidance (CFG) strengthens prompt alignment, NFT is formulated as a CFG-free method, and applying CFG to the fine-tuned velocity field can reduce the optimized reward. This motivates optimizing rewards over the guided output distribution while seeking to retain its prompt alignment. However, applying standard NFT directly to guided samples can cause reward degradation or training collapse. We trace this failure to a mismatch between the reference velocity used in the NFT loss and the score of the guided sample distribution, which biases the improvement direction. We propose GuidanceNFT, which estimates the matching reference velocity through online flow matching on guided samples and uses it to correct both regression terms. With an exact reference, the corrected objective recovers NFT’s intended update. In our experiments, the learned correction sustains reward improvement under both explicit and built-in guidance. On SD3.5-M, GuidanceNFT achieves higher optimized rewards than NFT with and without CFG across GenEval, text-rendering, and multi-reward objectives, together with better overall prompt alignment as measured by off-domain evaluation metrics. Experiments on models with built-in guidance, including Model Guidance and ELF, further demonstrate improved ImageNet generation, a better perplexity-entropy tradeoff on OpenWebText, and higher accuracy on GSM8K.
[LG-202] How Shared Representations Become Specialized: Formation Maintenance and Latent Susceptibility
链接: https://arxiv.org/abs/2610.15830
作者: Toru Takahashi
类目: Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
*备注: 23 pages including appendices; 5 figures
Abstract:When several predictions use the same internal representation, information about some targets can remain easier to recover than information about others. We study how this specialization forms, persists, and changes in minimal artificial learners. With reconstruction errors weighted equally during learning, changing which target is observed precisely selects the favored target and redirects an already specialized learner. Removing current support can reduce a formed bias, whereas state-following update feedback can maintain it. We then compare learners formed through different observation and update conditions but matched on specified reconstruction measurements. Both change under the same brief opposing intervention, yet their responses differ. Resetting Adam largely removes the ten-update difference; exchanging its second moment reverses which route is more susceptible. New interventions show that equalizing the mean second-moment scaling reduces but does not remove the difference: its coordinate-wise pattern also matters. With fresh Adam or calibrated SGD, smaller route differences emerge later, whereas 500 common opposing updates bring the selected allocation endpoints close. A further control produces response differences with identical acquisition and equal mean loss weights but jointly different weight variability and realized sequence. Acquisition-selected formation also extends to a symmetric three-target setting. These interventions distinguish specialization’s formation, present support, and time-dependent latent susceptibility.
[LG-203] Guarded Block Transfers for Polytope Distance: Linear Convergence with Explicit Constants and Gap-Certified Screening for SVM Training NEURIPS2026
链接: https://arxiv.org/abs/2610.15829
作者: Mayank Gupta
类目: Machine Learning (cs.LG)
*备注: Accepted as a poster at OPT 2026: 18th Annual Workshop on Optimization for Machine Learning (NeurIPS 2026 workshop). 19 pages (6 pages main text), 3 figures. Code: this https URL
Abstract:Hard-margin SVM training is the distance problem between two convex hulls, which the Triangle Algorithm (TA) solves with a certified duality gap. We propose \emphguarded block transfers: k Mitchell–Demyanov–Malozemov (MDM) transfers per scan are aggregated into one exact line search, guarded by a comparison with the single best transfer at a cost independent of the number of points n , and replaced by an away step when the leading transfer is capacity-clipped. On the Minkowski-difference polytope, where the objective is 1 -strongly convex, we prove global linear convergence with explicit constants and no combinatorial swap-step bounds; the per-step contraction is k -independent and only the drop-count bound grows with k , so aggregation is provably safe while the saving in scans is measured, not proved. Since the TA certificate is the Frank–Wolfe gap divided by the current distance, gap-safe screening applies, and we bound the iteration after which it has removed every zero-weight point off the optimal faces by O!\left(\tfrack\rho\log\tfrac1\tau\right) up to logarithmic factors in the geometry, \rho the rate constant and \tau the optimal-face margin. On five LIBSVM benchmarks, with fairer baselines than our submission, the solver matches the baselines to four or five digits wherever both converge, ties LIBSVM on sparse-support hard margins and beats our blended pairwise conditional gradients (BPCG). But our full-cache sequential minimal optimisation (SMO) code ties or beats it on kernel L_2 margins, LIBLINEAR’s dual solver is up to 4.5\times faster on the separable L_2 margin, and our solver is one to three orders of magnitude slower than LIBLINEAR’s Newton solver on dense-support L_2 margins and than LIBSVM on a nearly touching kernel hull. Its edge is the certificate, not speed.
[LG-204] Benchmarking Deep Learning Models for Anomaly Detection in Real-World Cloud Telemetry
链接: https://arxiv.org/abs/2610.15818
作者: Sumit Dhyani,Sougata Mukherjea
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
*备注: Published in 2026 11th International Conference on Cloud Computing and Big Data Analytics (ICCCBDA)
Abstract:Modern cloud platforms generate high-dimensional, noisy telemetry from thousands of microservices. Unlike classical anomaly detection benchmarks, cloud anomalies are short-lived, irregular, and largely non-periodic, making timely detection challenging. We present an empirical benchmark of deep learning models for anomaly detection using a real-world IBM Cloud Console dataset. We evaluate GRU, Transformer, TimesNet, and Autoformer architectures using the Numenta Anomaly Benchmark (NAB), a timing-aware metric that accounts for early detection and false alarms. TimesNet achieves the highest NAB scores due to strong early-detection sensitivity. Transformer models also detect anomalies early but incur higher false-positive rates. An improved GRU autoencoder, incorporating robust scoring and persistence rules, achieves competitive performance with a more stable false-alarm profile. Overall, the modest performance differences suggest that real cloud anomalies are dominated by short-term deviations rather than persistent periodic structure, emphasizing the importance of timing-aware evaluation and robust post-processing.
[LG-205] Grounding Deep Learning Framework Testing in Data Flow
链接: https://arxiv.org/abs/2610.15808
作者: Jiatao Ge,Yanzhou Mu,Yinglong Zou,Yibo Hu,Zhenyu Chen,Chunrong Fang
类目: oftware Engineering (cs.SE); Machine Learning (cs.LG)
*备注: 22 pages, 6 figures, 7 tables
Abstract:Data flow underlies deep learning (DL) framework execution, including automatic differentiation and operator-fusion optimization. Bugs involving data flow can cause incorrect results or unexpected behavior. Existing testing methods represent models primarily by operator topology, omitting runtime tensor values and properties and operator input requirements. They therefore cannot distinguish data flows in models with similar topologies, even when these differences affect framework behavior. We introduce the Tensor Flow Graph (TFG) to capture runtime tensor states and operator input requirements. Its nodes represent operators and their input requirements, while its edges record tensor values and properties. Building on the TFG, we propose FlowMuT, a mutation-based testing method grounded in data flow. Mutations are represented as constrained subgraph replacements, with mutation operators derived from prior work and an analysis of mutable data flow objects. In each round, FlowMuT builds a TFG for the seed model, pairs mutation operators with TFG sites, and filters illegal candidates using data flow constraints. It uses LinUCB to select a candidate based on coverage and impact, executes the mutant, updates selection feedback, and uses each accepted mutant as the next seed. We evaluate FlowMuT on PyTorch and MindSpore using representative models from diverse application scenarios. FlowMuT submitted 55 bug reports for cases missed by all baseline methods; developers confirmed 41 reports and fixed the bugs described in 7 reports. FlowMuT achieves 43.4% higher data flow diversity than the strongest baseline and a legal mutant rate of 96.7%, compared with 86.4% without constraints. These results show that testing from a data flow perspective is practical and effective.
[LG-206] A Dependency-Aware Multi-GPU Training Framework for Temporal Graph Neural Networks
链接: https://arxiv.org/abs/2610.15730
作者: Yang Zhao,Yue Dai,Liang Qiao,Buxin Tu,Bo Liu,Sen Zhang,Yi Zhang,Xiaoyu Hao,Jun Shi,Junshi Chen,Hong An
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
*备注:
Abstract:Temporal graph neural networks (TGNNs) achieve superior performance in many dynamic graph learning tasks. These models use node state vectors (i.e., memories) to capture and record graph changes (i.e., events), and then exploit these memories to assist prediction. However, this memory-centric design makes multi-GPU training challenging because predictions for later events often depend on memory updates from earlier events. To maintain model quality, existing multi-GPU TGNN training schemes strictly preserve all event dependencies, which limits parallelism and causes expensive cross-GPU memory synchronization. In this work, we observe that many of these dependencies can be handled more flexibly without harming training quality. Specifically, (a) dependencies in TGNN training have high locality in both time and graph topology, (b) the short-range dependencies missed within a large batch can often be recovered cheaply, and © many memory updates are not immediately needed. Based on these observations, we propose DepBloom to accelerate multi-GPU TGNN training with three techniques: (1) temporally bounded spatial partitioning to reduce unnecessary cross-GPU synchronization and improve parallelism; (2) intra-batch memory refinement to recover missing short-range temporal dependencies inside enlarged batches; and (3) critical-path optimization to prioritize urgent memory updates while deferring non-critical work during multi-GPU pipeline execution. Experimental results across representative models and diverse datasets show that DepBloom achieves an average speedup of 2.84 \times and up to 5.16 \times over state-of-the-art methods, while maintaining comparable or better model accuracy.
[LG-207] o Cache or Not to Cache: Understanding and Mitigating Safety Incompatibility in Cross-Model KV Cache Reuse
链接: https://arxiv.org/abs/2610.15721
作者: Zhiwei Lin,Jiahao Liu,Yan Liu,Chengpeng Wang
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR); Software Engineering (cs.SE)
*备注: 17 pages, 2 figures. Code: this https URL
Abstract:Cross-model KV cache sharing reduces prefill computation by allowing a receiver model to reuse prefix KV states generated by a related sender model. Existing methods optimize such reuse for latency and generation quality, but its safety implications remain unclear. In this work, we systematically evaluate this issue across diverse sender-receiver pairs and reuse configurations. We refer to deviations from the receiver’s native safety behavior as safety incompatibility. Our study shows that both the direction and magnitude of safety incompatibility depend on the model pair and reuse configuration, with value reuse causing substantially larger deviations than key reuse across all evaluated pairs. Furthermore, we propose VReparo, a technique that learns per-layer and per-KV-head affine maps offline to align reused sender value tensors with their receiver-native counterparts before decoding. Across two model pairs under full KV reuse and the state-of-the-art selective reuse setting, VReparo reduces the average HarmBench attack success rate (ASR) from 57.28% to 39.21% and the average StrongREJECT score from 0.4215 to 0.2163 upon our benchmarks, while retaining an average Time to First Token (TTFT) reduction of 85.62%.
[LG-208] DRanker: A Transformer-Based Multi-Task Ranking Model for the Disney Homepage
链接: https://arxiv.org/abs/2610.15696
作者: Priya Nirmal Singh Khokher,Daniel Nemirovsky,Benjamin P. Whitesell,Ethan Sukrae Lee,Eric Wulf,Tom Chee-Duffy,Nicholas Flores
类目: Machine Learning (cs.LG)
*备注:
Abstract:We present DRanker (Disney+ Ranker), a transformer-based multitask ranking model serving personalized recommendations on the Disney+ homepage. We describe the encoder-only architecture over enriched engagement-history sequences paired with a lightweight causal debiasing mechanism, and the inference stack that enables transformer-based ranking at scale. Through iterative development, we systematically evaluated more complex alternatives and found that simpler designs consistently outperformed them on offline metrics, qualitative behavioral evaluation, and in production. Deployed globally, DRanker drives over 1% lift in our key engagement metric and 0.5% lift in daily active users across all experiences powered by the ranker. We share practical lessons on data engineering, debiasing, and architectural choices that shaped our ranking system at scale.
[LG-209] Drift Then Spread: Learning One-Step Generative Policies for Online Reinforcement Learning
链接: https://arxiv.org/abs/2610.15662
作者: Chunyu Luo,Hongyu Zhao,Guangyi Ni,Yue Song,Lingkai Kong
类目: Machine Learning (cs.LG)
*备注:
Abstract:Generative policies offer expressive action distributions for online reinforcement learning (RL), but many diffusion- and flow-based policies require costly iterative sampling. We propose DriftSpread, a one-step generative policy based on Drifting for online maximum-entropy (MaxEnt) RL. To incorporate entropy without a tractable policy density, we build on an energy-based view of Drifting and derive critic-guided Drift and entropy-induced Spread from the negative MaxEnt objective. Our method learns these updates sequentially in the same actor, using action regression for value improvement and distribution matching to capture the effect of entropy. The actor updates require neither direct samples from the critic-defined Boltzmann target nor policy-density or score estimation. Under exact fitting with a fixed critic, we prove convergence to the Boltzmann target in the small-step limit followed by the long-time limit. On MuJoCo, DriftSpread achieves the highest aggregate normalized return among all evaluated methods, while outperforming the diffusion- and flow-based baselines in both training and inference efficiency.
[LG-210] sting Central Structure for Complex Networks
链接: https://arxiv.org/abs/2610.15659
作者: Qilong Wu,Yuxiao Li,Simeng Zhang,Xinying Liu,Yan Liu,Zengyou He
类目: Machine Learning (cs.LG)
*备注: 12 pages, 4 figures, submitted to Chaos, Solitons Fractals
Abstract:Centrality measures rank nodes even when the observed differences can be explained by ordinary structural variation. We therefore study a preceding question: whether a network contains a statistically identifiable central structure at all. We formulate this question as a network-level hypothesis testing issue using the adjacency spectral radius as the test statistic and estimate an empirical p-value from randomized networks generated under a specified null model. The framework is evaluated using G(N,M), Chung-Lu, and two implementations of the configuration model on simulated benchmarks and 55 real networks. Controlled experiments further vary the number and connection weights of designated central nodes while holding network size and density fixed. The results show a clear negative association between central-structure strength and the empirical p-value. They also demonstrate that rejection decisions depend substantially on the structural properties preserved by the null model: the detected network sets are not simply nested across models.
[LG-211] Context-Aware Precision Scheduling for Small Language Models
链接: https://arxiv.org/abs/2610.15640
作者: Mukund Athmaram,Manoj Rao,Liang Ma
类目: Machine Learning (cs.LG)
*备注:
Abstract:Small language models (SLMs) used as on-device agents must balance accuracy and latency as their context grows. Mixed precision helps manage this tradeoff by assigning different precisions to weight tensors according to their size and sensitivity. However, these choices depend on the operating conditions, so a configuration selected for a short context may no longer provide the best tradeoff at a longer one. We calibrate a timing model from measurements at different precisions and context lengths, and combine it with tensor sensitivities to choose configurations under a latency or accuracy budget. We argue that fewer bits do not always mean faster execution, so our predictor estimates latency from calibrated execution costs rather than weight bytes alone. Under a latency budget, reading the larger key-value (KV) cache leaves less time for weight operations, leading the allocator to select lower precisions. This gives a schedule of tensor changes that can be prepared before execution. Our runtime follows the schedule while retaining the session’s KV cache. We evaluate executed schedules on a laptop GPU, including a longer-context schedule with three weight formats and a tool-use benchmark measuring full response time and correctness. Under an accuracy budget, measured error from a quantized KV cache determines the budget left for weights. The allocator can raise or lower weight precision as this estimate changes. These experiments show where context-aware precision helps and where accuracy recovery remains limited.
[LG-212] MERIT: Scaling KG Foundation Models with Relation Transformers
链接: https://arxiv.org/abs/2610.15638
作者: Cosimo Gregucci,Daniel Hernandez,Antonio Vergari,Steffen Staab
类目: Machine Learning (cs.LG)
*备注:
Abstract:Knowledge graph (KG) foundation models (KGFMs) predict links on unseen graphs zero-shot. KGFMs have been pioneered by ULTRA which builds a relation graph from co-occurrences that match motif templates, and passes messages over it to obtain relation representations. More recent KGFMs such as MOTIF and TRIX increase the size of this relation graph, and FLOCK replaces message-passing with probabilistic random-walk ensembles. However, larger relation graphs mean more messages, and more walks mean more sequences to encode, so they improve performance at higher training and inference cost. This cost hinders their use in more complex reasoning tasks, such as complex query answering. To reverse this trend, we design MERIT, the first KGFM with a transformer relation encoder: every relation name attends to every other, and the motifs enter as a soft prior rather than as edges. Over 56 benchmarks, MERIT matches the zero-shot MRR of SoTA KGFMs while being 6x faster than TRIX and over 100x faster than FLOCK at inference. Fine-tuned for complex query answering like UltraQuery, MERITQuery improves over it at matched cost.
[LG-213] he Price of Precision: A Calibrated Cost Model for Mixed-Precision LLM Serving
链接: https://arxiv.org/abs/2610.15629
作者: Mukund Athmaram,Manoj Rao,Liang Ma
类目: Machine Learning (cs.LG)
*备注:
Abstract:Choosing a mixed-precision configuration for large language model (LLM) serving trades speed against accuracy. Allocators that use average bit-width or a memory budget as a speed proxy can miss changes in the resource that limits execution. We argue that served speed can be limited by compute, weight loading, or key-value (KV) cache access, depending on the phase, batch size, and context length. A byte-based proxy does not capture all of these costs, so fewer bits do not always mean faster serving. To address this gap, we build a cost model that accounts for the binding resource and use it in two ways. First, as a predictor: given a configuration and operating point, it estimates serving latency using constants calibrated for the model, GPU, and serving stack. We evaluate its accuracy across three hardware classes. Second, as an allocator: we combine modeled time costs with measured per-tensor accuracy costs and use a dynamic program to select a configuration under a latency or accuracy budget. We compare its configurations against byte-aware and sensitivity-only allocation.
[LG-214] Adaptive Role Assignment and Controlled Feature Reconstruction for Federated Image Semantic Communication
链接: https://arxiv.org/abs/2610.15623
作者: Yuxuan Chen,Jiao Bie,Hang Qi,Jingjin Wu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Hybrid federated semantic communication enables user equipment (UEs) to contribute model updates or semantic features under a shared training framework. However, heterogeneous role costs and the interaction between model aggregation and feature reconstruction (FR) remain key challenges in practical radio access network (RAN) deployment. To address these issues, we propose ARC-FedSC, a RAN-aware framework that coordinates UE role assignment and server-side controlled FR under a fixed communication budget. Within ARC-FedSC, RAN Aware Role-Bandit learns the historical contribution of UEs in different roles together with current uplink conditions, while Temporal Consistency Alignment (TCA) regulates FR according to recent model aggregation directions. Experiments on CIFAR-10 with time-varying RAN channels show that ARC-FedSC outperforms baselines across all five tested role configurations in the later training stage, with the best configurations achieving about 24.17 dB average PSNR over the last 20 rounds. Component level results further show that Role-Bandit maintains an average rolling UE coverage of 85.3%, compared with about 65.7% for baseline, while TCA reduces the probability of harmful FR events from 0.245 to 0.093, corresponding to a 62.0% reduction. Ablation and sensitivity studies further support the effectiveness and robustness of the proposed design.
[LG-215] ProML: Progressive Multi-label Learning
链接: https://arxiv.org/abs/2610.15614
作者: Yi Kou,Jieting Wang,Yuhua Qian
类目: Machine Learning (cs.LG)
*备注: 18 pages main text, 14 pages supplementary material
Abstract:Multi-label learning typically pursues model construction through the joint optimization of label fitting, correlation modeling, and structural regularization. While empirically effective, such a coupled paradigm imposes multiple learning objectives on a single predictive parameter matrix, forcing it to simultaneously fit observed labels and encode structural priors. This coupling may induce ill-conditioned optimization geometry, increase hyper-parameter sensitivity, and complicate capacity control. To address the these issues, this paper proposes a Progressive Multi-Label Learning (ProML) framework, which decouples direct label fitting from predictor learning via the introduction of an auxiliary predictive variable. This auxiliary variable admits a closed-form solution, yielding an equivalent profiled objective in which the supervised fitting term is scaled by a trade-off parameter. This result provides a unified view of ProML and conventional coupled learning, with the latter recovered as a limiting case. We theoretically analyze the optimization geometry of ProML and show that, under quadratic positive semidefinite structural regularization, ProML attenuates the dominant curvature of the fitting term, enlarges the stable learning rate range, and provides a controllable fidelity-regularization trade-off. Further, we instantiate the ProML for multi-label feature selection task by integrating global-local label priors and row-sparse regularization. Extensive experiments and controlled optimization analyses on 12 benchmark datasets empirically corroborate the theoretical analysis and demonstrate the robust and competitive performance of the proposed method.
[LG-216] An Attention-Guided Graph Neural Network Surrogate for Boundary Integral Computation in Physical Field Reconstruction
链接: https://arxiv.org/abs/2610.15590
作者: RongZe Gao,YuWei Wu,XinRan Ji,XingYue Ren
类目: Machine Learning (cs.LG)
*备注:
Abstract:In practical engineering and scientific computing, a real physical system can be reduced to a boundary value problem, and the adequate boundary information to determine the interior spatial distribution and evolutionary state of the physical field can be solved with linear equations, alternatively, the the real physical problem can be treated with boundary problems with in adequate information problems in field of machine learning. This paper proposes a deep learning framework for the recovery of missing boundary information using an attention-guided graph neural network (GNN) surrogate. The framework extracts features from the physical information on the discretized boundary into a latent space, and the missing boundary quantities are then predicted by the network. Thus, the nonlinear mapping from known boundary quantities to missing boundary quantities is learned using a variational autoencoder (VAE) architecture, while an attention mechanism is incorporated to better model the correlations among different types of boundary information. Furthermore, the interior physical field can be reconstructed using the conventional boundary element method (BEM) integral formulation. The effectiveness of the GNN-BEM framework is evaluated using numerical examples of the two-dimensional Laplace and Poisson equations, which demonstrate good performance in reconstruction accuracy, local detail preservation, and model stability. The results show that GNN-BEM provides a feasible computational framework for boundary information recovery and fast interior physical field reconstruction for partial differential equations under complex boundary conditions.
[LG-217] UNISON: A Unified Framework for Learning from Scored Bags
链接: https://arxiv.org/abs/2610.15564
作者: Ofek Kirshenboim,Or Shkuri,Shani Finkelshtein,Yoram Louzoun
类目: Machine Learning (cs.LG)
*备注:
Abstract:We introduce UNISON, a unified framework that bridges multiple instance learning and recommendation systems by treating both as instances of learning from scored bag collections of (item, score) pairs. Unlike traditional approaches that address bag-level classification and personalized recommendation as separate problems, UNISON decomposes learning into three interconnected stages: (1) learning universal item representations, (2) modeling bag-specific preference patterns through lightweight personalized parameters, and (3) inferring bag-level characteristics from learned preferences. We justify the architectural choice of linear bag-specific scoring through a first-order analysis: when item embeddings are sufficiently smooth in the score, a linear function approximates the score up to a Taylor remainder controlled by the curvature of the underlying preference function. For cold-start scenarios, we show that the cold-bag fitting problem reduces to ridge-less linear regression with a closed-form solution, and provide convergence guarantees for the iterative variant. Empirically, UNISON matches or outperforms state-of-the-art approaches across diverse domains: entertainment (books, movies), immunology (T cell receptor (TCR) based classification), and microbiome, offering superior computational efficiency and interpretability. A controlled synthetic experiment further confirms that the bag-specific parameters ( w_j, b_j ) recover the latent factors that drive item scores even when the projection \phi has no access to the original scoring mechanism (AUC 0.998). UNISON is available at this https URL.
[LG-218] Vision-Predictor-Action: Towards Efficient Embodied AI via Neuro-Symbolic JEPA Predictors
链接: https://arxiv.org/abs/2610.15556
作者: Yeonseok Lee
类目: Logic in Computer Science (cs.LO); Machine Learning (cs.LG); Robotics (cs.RO)
*备注: 26 pages, 5 figures, 4 tables. Code: this https URL | Checkpoints: this https URL | Zenodo: this https URL
Abstract:Vision-Language-Action (VLA) models commonly generate robot actions token by token through Large Language Model (LLM) decoders, causing inference latency to grow with action sequence length and limiting closed-loop control frequency. We present Goal-Conditioned Vision-Predictor-Action (VPA), which replaces autoregressive decoders with a lightweight latent control loop. A shared vision encoder and a frozen text encoder embed observations, goal images, and instructions; a neuro-symbolic selector chooses a discrete skill; a Joint-Embedding Predictive Architecture (JEPA) predictor forecasts the next latent state; and a flow-matching solver generates an entire action chunk in a few integration steps, so that sequential network evaluations per decision step do not depend on the action horizon. We analyze inference complexity, regularization against representation collapse, and conditions under which latent sub-goal thresholding limits error accumulation across task phases. As an initial proof of concept, we evaluated VPA on the LIBERO-10 suite. The trainable model has under 20M parameters (checkpoint under 75 MB, excluding a frozen 63.4M-parameter CLIP text encoder); a decision step of a single-camera reference configuration takes 10-14 ms on an Apple M4; and trained only on LIBERO-10 demonstrations in under three hours on a single GPU, a two-camera policy succeeded in 40.0% of 200 evaluation episodes. These preliminary results from a single training run suggest the approach is workable. Code and weights are available at this https URL and this https URL, and archived at this https URL.
[LG-219] Sensitivity Fidelity Reveals Physical Dependencies Beyond Forecast Accuracy in Deep-Learning Earth-System Models
链接: https://arxiv.org/abs/2610.15540
作者: Qi Chen,Peng Zhan,Hongxing Cui,Wenqing Cui,Ibrahim Hoteit
类目: Machine Learning (cs.LG); Atmospheric and Oceanic Physics (physics.ao-ph); Geophysics (physics.geo-ph)
*备注:
Abstract:Deep-learning models increasingly rival numerical simulation in Earth-system forecasting, yet forecast error alone cannot determine whether their learned input-output dependencies are physically consistent. Here we introduce sensitivity fidelity as an additional criterion: alongside prediction error, model responses to input perturbations should be examined against a trusted reference or established physical knowledge. In an idealized Rayleigh-Bénard convection benchmark, four surrogate architectures achieved low forecast errors yet encoded markedly different sensitivities. Quantitative comparison with a matched adjoint distinguished the models far more sharply than forecast accuracy and showed that predictive skill and sensitivity fidelity need not coincide, while finite-perturbation experiments further linked these differences to the propagation of initial-condition uncertainty. We then applied this strategy to El Niño-Southern Oscillation prediction, where no adjoint reference exists and learned sensitivities showed event-wise robustness, validity under finite perturbations, and consistency with established precursors. These results identify gradient-based evaluation as a discriminating complement to forecast-error metrics, providing a practical test of whether data-driven Earth-system models learn physically credible dependencies.
[LG-220] Poisoned into Action: Context Attacks on Typed Decision Workflows
链接: https://arxiv.org/abs/2610.15537
作者: Chenkai Wu,Xing Chang,Jun Yan
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注: 13 pages, 3 figures. Code available at this https URL
Abstract:Typed decision interfaces restrict models to predefined actions, yet schema-valid choices can still trigger policy-violating state changes. Evaluating context poisoning thus requires connecting model decisions to execution outcomes while accounting for legitimate automation retained by defenses. We introduce an execution-aware protocol tracing poisoned evidence through retrieval, model choice, automated action, and persisted state. We evaluate it with 1,659 fresh calls to Jev-1.13.0 across 20 development tasks in two synthetic workflows, using three repetitions per condition. Under natural retrieval, a standalone payload reaches all 60 trials and produces five security-violating executions, while a carrier-payload pair exposes the payload in 57 trials and produces one violation. An initial source-exclusion gate blocks the pair’s observed violation but also withholds ten otherwise correct attacked decisions and sixteen correct clean decisions. Correcting its empty-scope behavior restores twelve clean decisions without additional model calls. Factorial controls show that adding correct evidence reduces wrong executions from 119/120 to 55/120 in the semantic arm and from 60/120 to zero in the neutral arm, suggesting that much of the improvement reflects general evidence value rather than attack-specific protection. Three prespecified contrasts are not statistically supported after Holm correction, so the findings remain exploratory. The protocol provides an auditable framework for studying which poisoning-induced errors reach execution and how much correct automation defenses preserve. Code is available at this https URL.
[LG-221] Revisited Support Vector Machines: A Unified Joint Optimization Framework with Nyström Approximation for Scalable Adaptive Learning
链接: https://arxiv.org/abs/2610.15498
作者: Nassou Gongon Benjamin,Noulapeu Ngaffo,Patrick Njionou Sadjang
类目: Machine Learning (cs.LG)
*备注:
Abstract:The theoretical elegance of Support Vector Machines (SVMs), grounded in statistical learning theory, establishes them as a reference for robust classification. However, in the era of Big Data, classical SVMs face two major bottlenecks: the prohibitive cubic complexity \mathcalO(N^3) of training and the rigidity of static kernels. Traditional methodologies treat kernel selection as a discrete, “brute-force” GridSearch problem, effectively decoupling the learning of data representation from the optimization of the decision boundary. This paper proposes a paradigm shift with a so-called “Revisited SVM”, a unified framework that recasts the learning process as a continuous %bi-level joint optimization problem. We derive a novel objective function J(\alpha, \theta) that allows for the simultaneous optimization of dual Lagrange multipliers and kernel hyper-parameters. To solve this challenging non-convex problem efficiently, we introduce a hybrid Block Coordinate Ascent (BCA) algorithm, tightly integrated with Nyström-based low-rank approximation, which drastically reduces computational complexity to \mathcalO(Nm) . Extensive experiments conducted across four heterogeneous domains (Bioinformatics, Finance, Cybersecurity, and Computer Vision) demonstrate that our approach not only matches but often exceeds exhaustive GridSearch accuracy, while delivering training speedups of up to 100x. Notably, on the complex NSL-KDD intrusion detection dataset, our adaptive kernel approach achieves 96.25% accuracy compared to a mere 50% for standard baselines, proving its unique ability to model complex non-linear manifolds in high-dimensional spaces. We validate our results using rigorous statistical tests and provide a theoretical convergence analysis of the proposed algorithm.
[LG-222] Split Conformal Interval Length under Huber Contamination
链接: https://arxiv.org/abs/2610.15458
作者: Xiao Zhang,Bowen Zhang,Lie He
类目: Machine Learning (cs.LG)
*备注:
Abstract:Existing analyses of interval length in split conformal prediction largely focus on clean data. However, how contamination at the training and calibration stages alters interval length relative to the corresponding clean population oracle remains understudied. In this paper, we consider Huber contamination and derive a finite-sample upper bound that holds uniformly over contaminating distributions, with leading terms scaling as L(\delta_1+n_1^-1/2) +(n_2^-1/2+\delta_2)/((1-\delta_2)r_0) , in addition to a clean-sample estimation term. We further establish a minimax lower bound matching the upper-bound rate up to constant factors on a regular asymmetric location submodel. For calibration contamination, we obtain an exact worst-case characterization of the population interval-length deviation. We also extend the upper-bound results to conformalized quantile regression and adaptive residual scores. Experiments on synthetic and real-world datasets provide empirical support for our theoretical findings on training and calibration contamination.
[LG-223] Riemannian Graph Tokenizer for Structural Knowledge Transfer WWW26
链接: https://arxiv.org/abs/2610.15427
作者: Qimin Zhou,Haibo Liu,Yujie Wang,Li Sun,Chuan Shi
类目: Machine Learning (cs.LG)
*备注: Accepted at WWW26
Abstract:Foundation models are at the forefront of artificial intelligence. A tokenizer, converting the raw input into discrete representations that the model can understand, plays an important role to the success of foundation models. Unlike the text tokenizer that is well studied in large language models, graph tokenizer is still at its early stage, facing the challenges of tackling the non-Euclidean structures and capturing the structural semantics. How to design a graph tokenizer for structural knowledge transfer? To this end, we propose a Riemannian Graph Tokenizer (RGT) that bridges the structural knowledge and quantized representations to support cross-domain structural knowledge transfer. The connection is established by Riemannian geometry. Specifically, we first define the geometric vocabulary (trees, cycles and sequences), which captures fundamental structural patterns and reflects the intrinsic geometry of graph. Second, we construct a Riemannian quantizer with Riemannian Straight-Through Estimator to tokenise graph structures across multiple domains into discrete tokens. To ensure consistency and transferability across diverse geometric spaces, RGT further incorporates a geometry-aligned decoder that projects manifold-specific tokens into a unified tangent space. The theoretical analysis and geometric interpretations are provided to support the effectiveness of our proposed method. Extensive experiments across diverse datasets demonstrate that RGT significantly enhances structural knowledge transferability across graph domains.
[LG-224] Harnessing VLM in Embodied RL: From Reward Regressor to Success Detector via Exploration-Augmented Key-Observation Gating
链接: https://arxiv.org/abs/2610.15401
作者: Xin Liu,Yutong Jiang,Jian Su,Ziheng Wang,Chuang Wang,Haoran Li,Dongbin Zhao
类目: Machine Learning (cs.LG); Robotics (cs.RO)
*备注:
Abstract:VLMs provide promising solutions to embodied reward design. Most current methods that train VLMs as reward regressors not only require extensive annotations but are also primarily designed for finetuning pretrained policies, and perform less well in the more challenging from-scratch RL setting. This issue stems from a fundamental mismatch: regressing precise rewards is misaligned with what VLMs naturally excel at. We posit that it is more aligned with VLMs’ inherent multi-modal VQA capabilities for the models to function as success detectors, and propose VLM success reward via Exploration-Augmented Key-Observation Gating (VLMeg) to well harness VLM for embodied reward design. As the core of VLMeg, the proposed exploration-augmented key-observation gating first determines whether the current RL episode is judgment-worthy based on its similarity to a generated success trajectory. Then, the key observation with the lowest confusion from the judgment-worthy episode will be selected for VLM evaluation as episodic reward. This gating mechanism not only avoids redundant calls but also enables higher detection accuracy of VLM. Considering the sparsity of judgment-worthy episodes, the advanced exploration incentives based on the generated trajectory are introduced, enabling more efficient exploration toward judgment-worthy trajectories for more VLM reward triggers and thus gaining higher RL efficiency. Extensive experiments in MetaWorld demonstrate VLMeg achieves state-of-the-art from-scratch RL, outperforming advanced VLM-based baselines on 12 challenging reward-free tasks. Furthermore, VLMeg also enables effective RL without any ground-truth reward in 3 challenging real-world experiments. Code will be open after acceptance at this https URL.
[LG-225] GA-SVD: Generation-aware Singular Value Decomposition for LLM Compression
链接: https://arxiv.org/abs/2610.15398
作者: Shengkun Tang,Cong Zeng,Zongfang Liu,Dan Alistarh,Zhiqiang Shen
类目: Machine Learning (cs.LG)
*备注:
Abstract:Large language models (LLMs) demonstrate strong generalization across a broad range of reasoning and generation tasks, but their large parameter counts impose substantial memory and computation costs, hindering deployment in practical resource-constrained settings. One-shot low-rank compression is a promising solution since it reduces model size and inference cost effectively. However, existing SVD-based methods primarily target multiple-choice benchmarks and often fail to preserve performance on generation tasks. Moreover, their rank allocation strategies are typically heuristic. We propose GA-SVD, a generation-aware low-rank compression framework for LLMs. First, we introduce Generation-Aware Weight Whitening (GAWW), which leverages gradients from on-policy knowledge distillation to construct whitening transformations aligned with generative behavior. Second, we develop an evolutionary global rank allocation method that searches for effective layer-wise rank allocation under a given global compression ratio. Across extensive experiments on Mistral-7B, Qwen3 series, and Llama-3.3 models, evaluated on both multiple-choice and generative benchmarks, GA-SVD consistently outperforms existing SVD-based baselines and achieves state-of-the-art (SoTA) results. We further demonstrate scalability by applying our method to models with up to 70B parameters.
[LG-226] WUSPruner: Second-Moment-Optimal Transforms for Structured Pruning
链接: https://arxiv.org/abs/2610.15394
作者: Shengkun Tang,Dan Alistarh,Zhiqiang Shen
类目: Machine Learning (cs.LG)
*备注:
Abstract:Structured pruning is typically formulated as identifying which model dimensions to remove, often within the model’s original representation. However, the effect of coordinate deletion fundamentally depends on the basis in which pruning is performed. This raises a more fundamental question: how should the basis be chosen specifically for coordinate deletion? We introduce WUSPruner, a transform-based framework for structured pruning of large language models. We formulate basis selection as a local linear reconstruction problem that directly minimizes the reconstruction loss induced by pruning. Inspired by the second-moment transformation in WUSH, we consider a WUS transform constructed jointly from activation and weight statistics. We show that under positive-definite second moments, WUS transform minimizes the local reconstruction loss over all invertible changes of basis. In the resulting basis, the activation and weight second moments are simultaneously diagonalized and balanced, causing the deletion error to decompose into independent spectral contributions and yielding a principled criterion for coordinate selection. Guided by this result, we develop a scalable pruning method based on blockwise WUS transformations and compact input projections. Experiments on Llama and Qwen models from 8B to 72B parameters show consistent improvements over structured pruning baselines across language modeling, knowledge, commonsense reasoning, mathematics, and code generation benchmarks, while also delivering practical reductions in inference latency and memory usage.
[LG-227] DC-Flow: Target-Conditioned Cyclic Peptide Design via Delayed-Coupling Flow Matching
链接: https://arxiv.org/abs/2610.15391
作者: Yitian Wang,Minjie Cheng,Yaning Cui,Maohua Yang,Dongdong Wang,Hongteng Xu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Target-conditioned cyclic peptide design requires coordinating interfacial side chains with macrocyclic closure. Existing end-to-end generators concurrently update geometric and chemical variables from noise, assigning residues to disordered coordinates. Meanwhile, predictor-assisted workflows often get stuck in local traps because they lack feedback in the propose-and-filter stage and ignore 3D structural proposals during sequence search. To overcome the drawbacks of these two paradigms, we propose DC-FLOW, a staged generative framework built around Delayed-Coupling Flow Matching. In principle, DC-FLOW postpones sequence and side-chain activation until a target-conditioned backbone takes shape, resolving structural ambiguity before chemical assignment. Then, it feeds delayed activations back into the backbone, achieving co-evolution of backbone geometry and emerging residue identities by coupling their flows. This mechanism enables more reliable macrocyclic closure and favorable interface packing. For predictor-assisted workflows, DC-FLOW supplies 3D proposals with in-trajectory guidance, enabling cascaded test-time scaling to prune redundant backbones and terminate unpromising refinement early. Extensive experiments show that DC-FLOW improves cyclization fidelity, diversity, and interface energetics, substantially increasing the yield of confident cyclic peptide binders under a fixed final-candidate budget. Our code is available at this https URL.
[LG-228] Approximation in Contrastive Representation Learning NEURIPS2026
链接: https://arxiv.org/abs/2610.15388
作者: Yuanfan Li,Zihan Zhang,Yiming Ying,Ding-Xuan Zhou
类目: Machine Learning (cs.LG)
*备注: 54 pages, accepted by NeurIPS 2026
Abstract:Contrastive representation learning (CRL) is a fundamental paradigm for learning transferable representations. Recent work identifies its population target as the pointwise mutual information (PMI) \log\fracp(x,y)p_X(x)p_Y(y) between two modalities X,Y , where p is the joint density function and p_X,p_Y are the marginals. However, it remains unclear why standard inner-product scores f(x)^\top g(y) can effectively approximate a generally nonseparable population target PMI. In this paper, we address this question from the aspect of approximation for CRL. By linking PMI to a compact operator, we show that D -dimensional inner-product models perform rank- D spectral approximation, with error controlled by the spectral tail. We then study neural network realizations: in general settings, we show that deep neural network scoring functions are universally consistent, and the approximation error rates depend on the smoothness of the PMI function; under structured low-rank data models we obtain improved rates attained by moderate-width encoders. Finally, we establish a fast calibration rate from contrastive excess risk to downstream retrieval error under a density-ratio Tsybakov noise condition. Comments: 54 pages, accepted by NeurIPS 2026 Subjects: Machine Learning (cs.LG) Cite as: arXiv:2610.15388 [cs.LG] (or arXiv:2610.15388v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.15388 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Yuanfan Li [view email] [v1] Sun, 11 Oct 2026 06:28:12 UTC (225 KB) Full-text links: Access Paper: View a PDF of the paper titled Approximation in Contrastive Representation Learning, by Yuanfan Li and 2 other authorsView PDFHTML (experimental)TeX Source view license Additional Features Audio Summary Current browse context: cs.LG prev | next new | recent | 2026-10 Change to browse by: cs References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) IArxiv recommender toggle IArxiv Recommender (What is IArxiv?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[LG-229] ANTE: Anticipate Then Execute – Learning Internal World Models for Coding Agents from Verified Beliefs
链接: https://arxiv.org/abs/2610.15368
作者: Yi Ge,Kangrui Ruan,Jifan Chen,Manish Prajapat,Christian Bock,Xiaoyang Wang,Rein Houthooft,Aritra Sengupta,Varun Kumar
类目: Machine Learning (cs.LG)
*备注: 45 pages (10 main text), 19 figures, 12 tables
Abstract:Rapid advances in large language models have made software engineering agents capable of tackling complex coding tasks. Within a containerized environment, a coding agent receives tool responses or execution results, and determines its later actions by those earlier results. However, most existing coding agents never look ahead: their policies do not anticipate the consequences of actions, and the terminal reward barely attributes a failure to certain earlier steps, so an ill-judged edit can corrupt the repository state irreversibly before any signal arrives. Recent code world models aim to address these challenges, but they are either trained and then frozen before reinforcement learning (RL), or reduced to an auxiliary next-token loss on the environment’s outputs. We present ANTE (Anticipate, Then Execute), an agentic RL framework that verifies the policy’s pre-action predictions by real execution and turns them into per-step credit for its actions and trust-or-abstain decisions. With an internal belief stated before each action, Belief-Anchored Policy Optimization (BAPO) turns the settled predictions into per-step advantages by anchoring credit on the policy’s own forecasts. To turn the trained forecasts into better actions at inference, thereby facilitating effective test-time scaling, ANTE-TTO (test-time optimization) selects among candidate actions without an external scorer. ANTE raises Qwen3-32B from 25.0% to 57.2% pass@1 on SWE-bench Verified, 6.2% above a matched GRPO baseline trained without the belief state; its world model’s accuracy climbs from 44% to 69% during RL, and the learned foresight capability transfers across harnesses and out-of-distribution domains. In summary, we aim to build a code world model that knows what has happened, what holds now, and what the next action will achieve.
[LG-230] JSIR: An Audio-Aligned Multi-Task Dataset of Jazz Standard Recordings ICASSP2027
链接: https://arxiv.org/abs/2610.15356
作者: Sihun Lee,Dasaem Jeong
类目: ound (cs.SD); Digital Libraries (cs.DL); Machine Learning (cs.LG); Multimedia (cs.MM); Audio and Speech Processing (eess.AS)
*备注: Submitted to IEEE ICASSP 2027; 5 pages, 1 figure
Abstract:Beat tracking, chord estimation, and structure analysis are staple tasks of music information retrieval. Their datasets are drawn largely from Western popular music, and jazz remains a repertoire on which state-of-the-art systems underperform. We present JSIR, the Jazz Standards dataset for Information Retrieval. Its 267 recordings of 44 jazz standards span 25.6 hours, 3.7 times the audio of JAAH, the closest existing resource. For each recording, we provide annotations for beat and downbeat positions, chords, and sections, all produced by a jazz musician and revised in several passes. In baseline experiments, models trained with JSIR also improve on other jazz datasets: the downbeat F-measure of a state-of-the-art beat tracker rises from 87.7% to 89.2%, and the seventh-chord accuracy of a pop-trained chord estimator on JAAH from 34% to 41%. We publicly release the annotations, the annotation tool used to create them, and a web application that displays them in sync with the audio.
[LG-231] Full Waveform Inversion with a Learned Preconditioner
链接: https://arxiv.org/abs/2610.15331
作者: Jeongjin Park
类目: Machine Learning (cs.LG); Geophysics (physics.geo-ph)
*备注:
Abstract:Full-waveform inversion (FWI) is a powerful approach for reconstructing subsurface velocity models, but its practical performance is limited by ill-conditioning, sensitivity to the update direction, and the high cost of curvature-based methods. In this project, we investigate whether a neural operator trained on reverse-time-migration (RTM) images can provide a useful learned preconditioner for FWI. We compare projected FWI against neural-preconditioned FWI using either a standard Fourier neural operator (FNO) or a Jacobian-informed FNO (JI-FNO). The neural preconditioner applies the normal operator of the learned RTM map to modify the true Devito FWI gradient. Our experiments show that the JI-FNO preconditioner achieves the lowest model error while maintaining data-residual decay comparable to baseline projected FWI. In contrast, the standard FNO preconditioner is unstable and gives worse model recovery, despite using the same neural-preconditioning framework. We also identify a failure mode of a nonlinear neural Gauss Newton-type inner solve: directly matching the neural RTM output to the FWI gradient creates an objective-space mismatch and does not yield a reliable descent direction. These results suggest that forward RTM-image prediction alone is insufficient for learned preconditioning; accurate learned derivatives and objective-consistent use of the neural operator are critical for producing useful inversion updates.
[LG-232] FENAS: A Few-Shot Meta-Ensembling Framework for Efficient Neural Architecture Search with Synergized Zero-Cost Proxies ICONIP2026
链接: https://arxiv.org/abs/2610.15307
作者: Yifeng Mao,Yuze Feng,Yutong Shen,Liyang Gao,Zhixiang He
类目: Machine Learning (cs.LG); Logic in Computer Science (cs.LO); Neural and Evolutionary Computing (cs.NE)
*备注: 15 pages, 2 figures. Accepted at ICONIP 2026
Abstract:Efficient neural architecture search (NAS) requires accurate performance estimation under limited evaluation budgets. Existing zero-cost proxies (ZCPs) provide fast training-free signals, but each proxy usually captures only a partial property of an architecture, leading to unstable rankings across datasets and search spaces. In contrast, GNN-based predictors can improve accuracy, but often depend on labeled architectures, graph encoders, and costly representation learning, which makes them less suitable for extreme few-shot settings. We propose FENAS, a topology-agnostic few-shot prediction framework that turns complementary initialization priors into robust ranking signals. Its core component, CAST, is a feature pipeline that combines copula alignment and synergy transformation. CAST maps heterogeneous ZCPs into a compact feature manifold, enabling lightweight meta-learners to produce reliable rankings from scarce supervision. On NAS-Bench-201, with only 0.5% labeled samples, FENAS achieves a Kendall’s Tau ( \tau ) of 0.70, outperforming representative topology-based few-shot predictors in this low-label regime. In closed-domain search, FENAS evaluates the NAS-Bench-201 space in approximately 1138 seconds and achieves competitive test accuracy. By providing a fast, reliable, and structurally simple evaluation mechanism, FENAS supports the design of automated neural architectures under limited evaluation budgets.
[LG-233] Frag Flex: Frag ment-Based Flexible Molecular Generation for Exploring Vast Chemical Space NEURIPS2026
链接: https://arxiv.org/abs/2610.15284
作者: Sihyun Park,David Ryan Koes
类目: Machine Learning (cs.LG)
*备注: 15 pages, 5 figures, 5 tables. Accepted at the NeurIPS 2026 Workshop on AI for Drug Discovery: Bridging the Translation Gap
Abstract:Drug discovery requires efficient exploration of an astronomically large chemical space. Molecular graph diffusion models offer a promising approach to this task, but many require graph size to be fixed before generation. Although atom-level insertion and deletion can enable flexible-size generation, jointly modeling molecular size, atom types, and bonds while maintaining chemical validity remains challenging. We introduce FragFlex, a flexible-length discrete diffusion model that performs node insertion and deletion at the fragment level. During denoising, FragFlex jointly learns the size, fragment identities, and inter-fragment connectivity of the fragment graph. A neural attachment-site predictor and valence-aware assembler then convert the generated fragment graph into an atom-level molecule. In large-scale experiments generating 10 million molecules, FragFlex achieves near-perfect chemical validity while maintaining high uniqueness, demonstrating a strong balance between the two.
[LG-234] Imperfect Verifiers Exact Targets: Inference-Time Scaling for Continuous Generation
链接: https://arxiv.org/abs/2610.15281
作者: Kijung Jeon,Molei Tao
类目: Machine Learning (cs.LG); Computation (stat.CO); Machine Learning (stat.ML)
*备注: 74 pages
Abstract:Verifier-guided inference-time scaling uses additional computation to score intermediate states and steer generation toward better outputs. However, imperfect intermediate verifiers can distort sampling, and repeated refinement alone does not ensure recovery of the desired reward-tilted distribution. We introduce Continuous-State Verifier-Guided Backtracking (CS-VGB), an inference-time scaling method for sampling from reward-tilted distributions with continuous generative models through repeated denoising and renoising. With correctly paired denoising and renoising transitions, the stationary distribution of CS-VGB at the clean checkpoint equals the desired reward-tilted target, even when intermediate verifiers are inaccurate. Under stated regularity conditions, CS-VGB converges exponentially to the target despite verifier error, and the expected number of updates needed to reach a clean sample grows only linearly with the number of checkpoints. In analytically tractable settings, we empirically demonstrate that CS-VGB recovers the desired reward-tilted distribution despite verifier error, whereas the compared verifier-based sampling methods fail to recover the target under the same verifier errors. We also instantiate the framework with diffusion models and one-step stochastic posterior maps. Across prompt-guided ImageNet generation, facial-attribute generation on CelebA, and multi-robot planning, recurrent CS-VGB search improves reward and task success as inference budgets grow, with image-generation rewards continuing beyond the observed plateaus of denoising-only baselines.
[LG-235] Correlation-Aware Reconstruction for Multivariate Time Series Anomaly Detection
链接: https://arxiv.org/abs/2610.15250
作者: Md. Johir Raihan,Zahra Zamanzadeh Darban,Yueyang Liu,Flora D. Salim,Levin Kuhlmann,Mahsa Salehi
类目: Machine Learning (cs.LG)
*备注:
Abstract:Reconstruction-based detectors for multivariate time series anomaly detection typically learn a single entangled representation of normal behavior and flag anomalies based on reconstruction error. However, real-world anomalies are not only isolated spikes in a single channel but also disruptions in the relationships between channels, where each sensor individually appears normal, but their joint behavior deviates from the norm. Because these models compress both temporal evolution and inter-channel structure into this single space, they lack an explicit mechanism to isolate such inter-channel dependencies from normal temporal dependencies, leading to lower detection rates for cross-channel anomalies. We propose CARMA (Correlation-Aware Reconstruction for Multivariate Time Series Anomaly Detection), which is a reconstruction-based variational autoencoder with two complementary latent spaces: a temporal latent that captures sequential dynamics, and a correlation latent that is trained to reconstruct the window’s pairwise Pearson correlations. Both representations are fused in the signal decoder, and anomalies are detected from the reconstruction error. We evaluate CARMA on the TSB-AD-M benchmark suite (17 datasets, 180 series) against 29 baseline methods, and show that it significantly outperforms all of them across multiple threshold-free and range-aware metrics, achieving a VUS-PR of 0.4929. Ablations show that removing the correlation branch degrades performance, that adding a second latent without supervision recovers only part of the gain, and that supervising the correlation latent with an explicit correlation target accounts for the rest and produces a sharper error response at anomalies.
[LG-236] ObsSpec: Accelerating Agent ic Reinforcement Learning with Observation Speculation
链接: https://arxiv.org/abs/2610.15246
作者: Kyle Montgomery,Wang Zhang,Chenguang Wang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Tool-using agents alternate between model generation and tool execution, waiting for each tool call to complete before starting the next generation. This sequential execution is costly during agentic reinforcement learning (RL), which requires large batches of multi-turn rollouts. We introduce ObsSpec, an observation speculation pipeline for agentic RL that uses a small world model to predict tool outputs while the environment executes, allowing the policy to begin its next generation early. ObsSpec retains speculative work only when the predicted and actual observations match exactly, making rollout collection lossless, while co-training keeps the world model up to date as the policy learns. On terminal-use tasks, predictions from a 0.6B world model exactly match actual observations in 66.4% of attempts and arrive roughly 16 times faster. Using these predictions, ObsSpec reduces agent latency by 9.3% and 14.2% for 8B and 32B policies. Across controlled experiments, we characterize the regimes in which ObsSpec reduces agent latency under shared policy resources. Finally, because ObsSpec overlaps policy generation with tool execution, it is architecturally compatible with existing techniques for accelerating RL. Code is available at this https URL.
[LG-237] Zeroth-Order Fine-Tuning of LLM s in Learnable Low-Rank Subspaces
链接: https://arxiv.org/abs/2610.15225
作者: Chuang Tong,Ling Chen
类目: Machine Learning (cs.LG)
*备注: 7 pages, 4 figures
Abstract:Zeroth-order (ZO) optimization enables memory-efficient fine-tuning of large language models (LLMs). However, ZO optimization suffers from high variance in high-dimensional parameter spaces. While recent methods restrict perturbations to random low-rank subspaces, these subspaces often misalign with task-specific descent directions. We propose AdaSub-ZO, a novel forward-only fine-tuning framework in learnable low-rank subspaces. Instead of relying on fixed random matrices, AdaSub-ZO iteratively learns layer-wise projection matrices directly from finite-difference feedback to track task-aware descent directions. Furthermore, AdaSub-ZO incorporates a subspace momentum acceleration mechanism to smooth noisy gradient estimates across evolving subspace bases. Extensive experiments on natural language understanding tasks demonstrate that AdaSub-ZO outperforms existing ZO baselines, substantially reducing gradient estimation variance and accelerating convergence.
[LG-238] DeltaNet-PC: Test-Time Memory Without Backpropagation Through Time
链接: https://arxiv.org/abs/2610.15185
作者: Todd Morrill,Richard Zemel
类目: Machine Learning (cs.LG)
*备注:
Abstract:Test-time memorization of sequential information is a foundational capability for any intelligent system that needs to rapidly store and retrieve data, as demonstrated by the attention mechanism in large language models. However, Transformers and linear attention variants are trained via backpropagation through time (BPTT), which has drawbacks such as its enormous memory requirements. Predictive coding (PC), an algorithm that optimizes a prediction-error energy function with local updates, offers a compelling solution but still lacks test-time memory for sequential data. We develop a solution to this problem by treating the per-step memory update \Delta S_t as a latent variable, as opposed to a slow weight to be learned. By taking the log probability of a standard PC-based probabilistic model, we derive a PC-native solution for verbatim sequence memory, and relate our update rule to common choices in the linear attention literature (e.g., DeltaNet). We present an extensive set of empirical tests to probe our system’s properties, including its scaling capacity, memory footprint, and ability to infer a context from evidence (e.g., bank (river) vs. bank (money)). We show how DeltaNet-PC can be used for storing single sequences of data, multiple sequences of data, and key-value associative memory pairs with perfect recall accuracy.
[LG-239] Game-Theoretic Reinforcement Learning based Topology-Agnostic Defense for Virtual Power Plants
链接: https://arxiv.org/abs/2610.15183
作者: Pooja Algikar,Stefano Fenu,Kumarsinh Jhala,Siby Plathottam
类目: Machine Learning (cs.LG); Systems and Control (eess.SY)
*备注:
Abstract:The rapid proliferation of distributed energy resources (DERs) aggregated by virtual power plants (VPPs) elevates cybersecurity risk by expanding the attack surface. A coordinated cyberattack has the potential to disrupt distribution-system, bulk-system, and market operations. Existing cyberphysical defense strategies largely assume access to grid network models and operating conditions, which may not be available to third-party aggregators. This paper proposes a novel hybrid, perturbation-based cyber-physical defense for VPP-operated DERs that blends game-theoretic planning with model-free reinforcement learning (RL) under a Bayesian belief-update framework. Topology agnosticism is defined from the VPP information and implementation perspective: the VPP does not require access to the feeder topology, line parameters, switching configuration, or full system state. This work advances the state of the art through three contributions: (1) a multi-output Gaussian process with Hamiltonian Monte Carlo for discrete attacker-type inference, probability of detection, and confidence of detection for modeling the belief distribution; (2) moment-matched Gaussian propagation of posterior beliefs across episodes; and (3) an RL policy that adaptively integrates Stackelberg leader (VPP) actions with exploration. Furthermore, we establish equilibrium existence and derive an episode-level closed-loop utility bound connecting attacker-belief uncertainty with GT-RL action alignment. Evaluations on IEEE 13- and 123-Bus feeders with DER portfolios demonstrate effective detection under attacker-model uncertainty and evaluated feeder re-configuration conditions, while requiring no explicit feeder-topology knowledge. Overall, the results indicate that combining bi-level game-theoretic planning, nonparametric Bayesian inference, and adaptive RL is a promising direction for VPP cybersecurity.
[LG-240] Only What Youve Signed up For: Secure Workflows with Constrained Diffusion Agents
链接: https://arxiv.org/abs/2610.15177
作者: Michael Cardei,Anjun Hu,Chenrui Ma,Francesco Pinto,Chawin Sitawarin,Chulin Xie,Ferdinando Fioretto
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
*备注:
Abstract:As agentic AI systems take on increasingly complex and consequential tasks, securing them against prompt injection becomes critical. Existing defenses often rely on detecting or reasoning about injected malicious instructions and can therefore remain vulnerable to adaptive attacks that optimize the injection itself. Diffusion language models (dLLMs), an emerging framework for language generation, offer a new opportunity for security enforcement: by iteratively refining their sequence wide predictions, they enable broader-scope constraint enforcement throughout generation. This paper introduces Policy-Constrained Diffusion Agents (CoDA), a framework that enforces task dependent security policies directly within the denoising process of diffusion language models. At each denoising step, CoDA projects token distributions onto a policy-compliant set, maintaining constraint satisfaction throughout generation. We evaluate CoDA across state-of-the-art diffusion language model families under both static and adaptive prompt-injection attacks. Results demonstrate substantial reductions in attack success rate with competitive task utility and practical inference overhead.
[LG-241] LexiCan: Lexicographic Canonicalization of Point Clouds for Machine Learning Applications
链接: https://arxiv.org/abs/2610.15171
作者: Daniel Sanchez,Djordje Mihajlovic,Federica Valeau,Luka Benić,Raghuwansh Raj,Soutrik Sarangi,Snir Hordan
类目: Machine Learning (cs.LG)
*备注:
Abstract:Canonicalization resolves the rotational and permutation ambiguity of 3D point clouds by selecting a unique orbit representative, maintaining their geometric information while allowing standard architectures, such as Transformers, to operate on them. Existing canonicalizations, however, exhibit discontinuity jumps on molecular configurations and non-trivial invariance errors in popular molecular datasets, which lead to ambiguity in their representation. To bridge this gap, we introduce LexiCan, a canonicalization that is provably invariant under either O(3) or SO(3) , empirically invariant to rotations to within \sim 10^-6 angstroms on the QM9 and QM-sym molecular datasets, and mitigates discontinuity jumps on QM-sym, while running orders of magnitude faster than an existing canonicalization. Finally, non-invariant models perform better on QM9 property prediction and ModelNet40 shape classification with LexiCan than with that canonicalization.
[LG-242] Beyond Adjacent Layers: Graph-Guided Layer Fusion for Compressing Large Language Models NEURIPS2026
链接: https://arxiv.org/abs/2610.15159
作者: Muhammad Usama,Yeoreum Jung
类目: Machine Learning (cs.LG)
*备注: Accepted at NeurIPS 2026. Code: this https URL
Abstract:Existing depth-compression methods for large language models (LLMs) score layers individually or group only adjacent or contiguous layers, implicitly assuming that functional redundancy is local. We test this assumption by computing full pairwise Centered Kernel Alignment (CKA) similarity matrices for thirteen LLMs from six architecture families (1.6B-14B parameters). Non-adjacent layers have 3.6-15.9x more high-similarity pairs than adjacent ones, a structure that prior similarity-based layer merging has not systematically exploited. We propose Fusion via Community Mining (FCM), which builds the full L x L similarity graph, uses spectral community detection to find clusters of functionally equivalent layers, keeps the highest-Block-Influence layer in each cluster, and applies LoRA recovery; a pre-specified rule computed only on calibration data chooses between cosine and CKA similarity. At 2x compression, FCM lowers mean perplexity below ShortGPT on twelve of thirteen models, significantly on ten (Welch’s t-test with Holm-Bonferroni correction; up to 50% on Qwen3-4B), with Mistral-7B the only loss (+13.4%). Against FlattenGPT, SWM, and DeltaLLM under matched calibration and recovery, FCM gives the best perplexity on ten of thirteen models (mean rank 1.15), and it also outperforms MKA on three anchor models. FCM matches or exceeds ShortGPT on 15 of 20 zero-shot reasoning cells and also improves MMLU, LAMBADA, HumanEval pass@1, and free-form generation quality. The 2x-compressed Qwen2.5-7B has 46% fewer parameters, 42% lower peak VRAM, and 84% higher decode throughput on a single A100. Spectral grouping outperforms Fisher contiguous segmentation by 41% and uniform partitioning by 51%, and FCM’s advantage grows beyond a 2x crossover up to 3x and persists at the 14B scale. Code: this https URL
[LG-243] Sensor-Aware Joint Data Fusion and State Estimation under Sparse Multi-rate Distribution Grid Sensing
链接: https://arxiv.org/abs/2610.15156
作者: Pooja Algikar,Stefano Fenu,Kumarsinh Jhala,Siby Plathottam
类目: Machine Learning (cs.LG)
*备注:
Abstract:Distribution-system state estimation is a growing necessity for the modern power grid as the addition of local storage and generation add system complexity that an operator or aggregator is not able to directly monitor. Classical methods work well when all system elements are time-synchronized and at least partly observable, but multi-rate monitoring of sparse heterogeneous sensors typically relies on machine learning approaches. It is a common practice in the literature to structure this task as a two-stage pipeline, first fusing data from disparate sensors and then feeding the fused pseudo-measurements to a separately trained state estimator. We propose a heterogeneous latent graph ODE in which measurement functions and sampling processes are typed and localized on specific buses, paired with a graph- transformer state estimator solving DSSE specifically for multi-rate heterogeneous sensor placements, and demonstrate its effectiveness across multiple IEEE standard topologies as observability degrades.
[LG-244] When Dynamic Scaling Fails: A Scale-State Feedback Governs Recurrent-State Quantization in Linear-Attention and SSM LLM s NEURIPS2026
链接: https://arxiv.org/abs/2610.15139
作者: Yuhang Jiang
类目: Machine Learning (cs.LG)
*备注: 14 pages, 6 figures, 7 tables. NeurIPS 2026 Workshop on Long-Context Foundation Models (non-archival). Extended version: adds a cross-architecture containment comparison on Mamba-2 and corrects two claims in the workshop version. Code and project page: this https URL
Abstract:Linear-attention and state-space LLMs carry long context in a fixed-size recurrent state, and quantizing that state is the natural way to cut its cost during long decode. Unlike a weight tensor, the state is rewritten at every step, so a quantizer applied at each step sits inside the recurrence and its errors compound. INT4 that leaves synthetic recall perfectly intact at the start destroys it a thousand steps later. The mechanism is a feedback between the quantizer scale and the state it is derived from. Yoking that scale to a full-precision reference removes the resulting norm explosion entirely, yet recovers only 1.8 of the 5.4 nats lost, so numerical instability and functional degradation are separable failures. The remainder is a resolution problem that locating outliers does not solve, since every outlier-aware scheme here loses to plain persistent scaling. Allocating bits across layers does help, but under article-disjoint splits a learned allocation collapses onto the trivial rule of protecting the first few layers. Run out to 32K, group-wise INT4 is worth no more than discarding the state. Contain the scale rather than locate outliers, and report the decode horizon at which a bit-width was certified.
[LG-245] RELAGRAM: Recursive Latent Graph Model for the Capacitated Vehicle Routing Problem NEURIPS2026
链接: https://arxiv.org/abs/2610.15106
作者: Vladimir Kirill Bickov,Neil Yorke-Smith
类目: Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE); Optimization and Control (math.OC)
*备注: Accepted to the NeurIPS 2026 MLxOR workshop. Paper contains 6 pages (4 main text, 1 references, 1 appendix); 1 figure and 3 tables
Abstract:Latent recursion repeatedly processes internal representations through the same neural network. Recent work has shown this approach to be effective on structured reasoning tasks such as Sudoku and maze solving. However, whether latent recursion can support neural routing remains unclear. We introduce RELAGRAM (Recursive Latent Graph Model), a novel approach that brings latent recursion to graph combinatorial optimization. RELAGRAM encodes a capacitated vehicle routing problem (CVRP) instance as edge tokens; a compact transformer recurrently updates the latent representations of these tokens. An output head maps the answer state to an edge heatmap, from which an autoregressive decoder constructs a feasible solution. On CVRP-100 instances, the 0.53M-parameter RELAGRAM achieves gaps to a PyVRP reference of 13.4%, 9.8%, and 8.3% under greedy decoding, eightfold symmetry augmentation, and 256-candidate search, respectively. Trained only on CVRP-100, RELAGRAM generalizes zero-shot to CVRP-500 and CVRP-1000, achieving gaps of 15.6% and 21.7% under 256-candidate search. Together, these results demonstrate that latent recursion is a promising approach to neural routing.
[LG-246] Random Feature Pruning for Multi-Class Classification
链接: https://arxiv.org/abs/2610.15101
作者: Mohammad Amin Omidi,Babak Seyfe,Shahrokh Valaee
类目: Machine Learning (cs.LG)
*备注: 43 Pages, 6 Figures
Abstract:Weighted sums of random kitchen sinks (W-RKS) is a commonly used feature extraction technique for supervised learning tasks. A major drawback of this method is the redundancy and ineffectiveness of many extracted random features. This paper investigates efficient methods for reducing such redundancy in multi-class classification problems. We specifically formulate feature selection as a variant of the sequentially nested model selection problem and introduce the Successive Empirical Excess Risk Difference Maximization (SEED-MAX) algorithm. It recursively computes empirical risks to identify the most parsimonious yet informative set of features. We derive an analytical lower bound for the probability of correct feature selection by the SEED-MAX algorithm. This lower bound on the probability of correct feature selection can be made arbitrarily close to one by increasing the training dataset size. To further enhance the rate of pruning, we evaluate the features retained by SEED-MAX using the Fisher test (F-Test). Empirical evaluations on UCR time-series classification datasets show that the sequential application of SEED-MAX with the Fisher test achieves approximately 92 percent rate of pruning with negligible degradation in classification accuracy. Implementations of the SEED-MAX and SEED-MAX-F-Test methods are available at this https URL.
[LG-247] Anchors Enhance Controllability and Reasoning of Diffusion Language Models
链接: https://arxiv.org/abs/2610.15097
作者: Feiyu Qu,Zheng Huang,Andrea Zanette
类目: Machine Learning (cs.LG)
*备注:
Abstract:Diffusion language models (DLMs) offer new opportunities for controllable text generation. However, their controllability, defined as the ability to satisfy task constraints while preserving their original capabilities, remains underexplored. DLMs’ bidirectional attention allows control tokens to guide generation from arbitrary positions in the output sequence, but effective guidance depends on the compatibility between their content and placement. To this end, we propose AnchorDLM, a framework in which an anchor generator produces an anchor consisting of a short output fragment and its position for each query. The anchor is placed as control tokens at the specified position to initialize the output sequence of a frozen DLM and guide subsequent denoising. We train the anchor generator with reinforcement learning (RL) to jointly optimize anchor content and placement, using rewards computed from the DLM’s generated outputs based on answer correctness and constraint satisfaction. Moreover, through the lens of discrete flow matching (DFM), we interpret anchor placement as helping to reshape the initial velocity field that guides generation. Experiments on code generation and mathematical reasoning across five DLM backbones show that AnchorDLM improves controllability and further yields gains in reasoning performance across diverse benchmarks.
[LG-248] Swarm Telepathy: Prefill-Free KV-Cache Transfer Across Families
链接: https://arxiv.org/abs/2610.15095
作者: Abdulrahman Alabdulkareem,Irene Tenison,Yang Zhang,Yada Zhu,Boris Katz,Dan Gutfreund
类目: Machine Learning (cs.LG)
*备注:
Abstract:LLM multi-agent systems split work across agents that communicate through text. An agent summarizes its findings, another agent receives the text. This constrains communication to discrete tokens and costs redundant compute, limiting bandwidth and capabilities of multi-agent systems. Recent works demonstrate KV-cache transfer between LLMs but they either require the LLMs to be within the same family or cannot function across context, thus requiring the receiver to prefill text. We present SwaT (Swarm Telepathy), the first cross-family KV-cache transfer method that allows prefill-free communication in multi-family multi-agent systems. In an agent-to-agent communication setting, combining SwaT with recent advances in KV-cache compaction beats text-to-text communication in both accuracy and speed, with a mean of 40% higher accuracy and a mean of 4\times faster wall-clock time. Experiments show SwaT achieves high performance on 17 benchmarks across 12 LLMs and 6 families. Additionally, given a sender Vision-Language Model (VLM), SwaT is able to transfer the vision capabilities to a text-only receiver LLM exceeding text-to-text communication despite SwaT being trained on text-only data. SwaT keeps both sender and receiver models frozen without fine-tuning and only requires 50M tokens and 1 GPU-day to train.
[LG-249] On the Theoretical Limits of Learning Unconstrained Interatomic Potentials
链接: https://arxiv.org/abs/2610.15078
作者: Prajwal Ananth,Shuwen Yue
类目: Machine Learning (cs.LG); Chemical Physics (physics.chem-ph)
*备注:
Abstract:Machine-learned interatomic potentials (MLIPs) accelerate molecular dynamics (MD) simulations by learning the energies and forces that govern atomic motion. Recent work has shown that relaxing physical inductive biases can improve computational efficiency and predictive accuracy, raising the question of whether energy-and-force matching alone is sufficient to ensure accurate sampling. We show that, for unconstrained hypothesis classes, the relationship between prediction error and sampling accuracy is exponentially ill-conditioned in dimension. In particular, prediction errors that decay exponentially with dimension can coexist with nearly maximal total variation distance between the learned and target distributions. We identify two failure mechanisms. The first is kinetic trapping, where relaxation to an accurate learned equilibrium requires exponentially long times, even when simulations are initialized from training configurations at which energies and forces are matched exactly. The second is equilibrium mismatch, where the learned stationary distribution itself remains far from the target. Under explicit assumptions, we show that equilibrium mismatch can persist even in local, symmetry-preserving models, where small per-atom errors coexist with a distributional discrepancy that approaches its maximum as system size increases. Finally, we establish global energy-control conditions that guarantee equilibrium accuracy and identify additional conditions required for rapid relaxation.
[LG-250] oward a Multimodal Foundation Model for Plasma State Reconstruction NEURIPS2026
链接: https://arxiv.org/abs/2610.15076
作者: Bin Xia,Chuanfei Dong,Boting Li,Yilan Qin,Zikang Xie,Sung Hyun Son,Adam J. Stanier,Jongsoo Yoo,Hantao Ji,Ahmed Diallo,Han Ding,John Wise
类目: Machine Learning (cs.LG); Plasma Physics (physics.plasm-ph)
*备注: Accepted at the NeurIPS 2026 Workshop on Physical World AI (PhysWorldAI). 15 pages, 5 figures, 5 tables
Abstract:Plasma experiments observe coupled physical states through sparse, heterogeneous, and incomplete diagnostics. We present a proof of concept toward a multimodal foundation model for plasma state reconstruction during magnetic reconnection, motivated by the Facility for Laboratory Reconnection Experiments (FLARE). Using 250 synthetic VPIC simulations at fixed reference electron beta \beta_e0=0.2 , with 225 training runs and 25 disjoint validation runs, we train a 5.5M-parameter residual 3D U-Net to reconstruct three magnetic-field components and electron density from incomplete observations. The model combines spatial-only pooling, which preserves all 24 temporal frames, with global spatiotemporal self-attention and rotary positional encoding. Independently masking magnetic and density modalities across random, grid, block, and temporal patterns enables sparse sensing, cross-modal completion, density super-resolution, and temporal completion with a single set of weights and no task-specific retraining. With density fully hidden, just 0.1% magnetic spatial visibility reduces median density NRMSE from approximately 0.9 to 0.1. Conversely, observed density improves current-density reconstruction when magnetic data are absent or extremely sparse. With no magnetic observations, roughly 1% density visibility approaches the density reconstruction accuracy achieved with fully observed magnetic fields, although current structure remains more dependent on direct magnetic measurements. Full magnetic conditioning also stabilizes completion of missing future density frames. These results demonstrate bidirectional cross-modal predictive coupling within this simulation regime. Generalization across plasma regimes or to experimental measurements remains untested.
[LG-251] he Optimal Decoder is Tilted Temperature
链接: https://arxiv.org/abs/2610.15058
作者: Nikolaos Papagiannis,Ananth Grama,Wojciech Szpankowski
类目: Machine Learning (cs.LG)
*备注:
Abstract:Larger sample budgets benefit from higher sampling temperatures in LLMs, but what temperature tuning optimizes and why it works remain poorly understood. We seek to answer these questions: we pose decoding as pass@ k maximization subject to two constraints that encode critical decoding failure modes: an entropy floor preventing diversity collapse, and a cross-entropy ceiling preventing implausibility. We prove that this maximization takes the form of a single closed-form conditional sampling rule, the \emphtilted temperature q^\star \propto \bigl(p e^\eta V\bigr)^1/T . Here, the tilt comes from a verifier potential V that quantifies how likely a partial generation is to eventually pass. Ordinary temperature sampling, verifier-greedy decoding, and reward-tilted policies represent special cases. We further show that, under a natural rule for scaling the diversity budget with sample size k , the resulting optimal temperature increases with k , turning empirical choice into a theorem. The optimum still needs the unknown verifier potential, which we estimate using two design approaches. In the online setting, our J-EOS algorithm jointly tracks the two constraint multipliers and learns the verifier potential from binary pass/ fail feedback alone. In the offline setting, we show that when the model is calibrated, its log-probability tracks correctness up to an affine map. Under this condition, the tilt collapses into the exponent and the optimum reduces to ordinary temperature sampling. Calibrated Temperature Decoding (CTD) then fits one scalar offline and samples with zero additional inference-time overhead relative to temperature sampling. Building on these decision rules, we design a practical decoding algorithm for code generation and evaluate it alongside CTD on benchmarks including HumanEval and MBPP, showing improvements over standard and adaptive decoding baselines.
[LG-252] Information-Theoretic Foundations of Online Temperature Sampling
链接: https://arxiv.org/abs/2610.15043
作者: Jin Sima,Nikolaos Papagiannis,Ananth Grama,Wojciech Szpankowski
类目: Machine Learning (cs.LG)
*备注:
Abstract:Repetitive, degenerate generation is a persistent failure of LLM decoding, and the standard remedies (temperature, nucleus, and top- k sampling) are static heuristics. By Kac’s lemma and the AEP, a stationary source takes time exponential in its entropy to repeat a block. We therefore maximize the entropy of the sampling distributions under one cross-entropy budget that keeps the output plausible under the model. Offline, the solution is temperature sampling, with the inverse temperature as the budget’s Lagrange multiplier. \emphOur contribution is the online problem: the conditionals p_1, p_2, \dots are revealed one at a time, and the sampling distribution q_t must be chosen causally from p_1, \dots, p_t while the budget couples the entire horizon. The optimal temperature is thus a global shadow price that must be estimated from a prefix. We propose Entropic Online Sampling (EOS), which re-solves for the temperature on each prefix. We prove it loses O(R\log n) entropy against the offline optimum while overshooting the budget by O(R\log n) , with R an instance-dependent drift parameter. A budget-tightened variant is exactly feasible at the same rate. Conversely, every exactly feasible online algorithm must lose \Omega® , so EOS is optimal up to a \log n factor. When the conditionals are drawn from a finite mixture, R=\tildeO(\sqrtn) . Experiments confirm that EOS tracks the offline temperature and beats fixed-temperature sampling in entropy.
[LG-253] Sparsity Saves Compute Not Samples: Oracle Inequalities for Dense and Sparse Mixture of Experts
链接: https://arxiv.org/abs/2610.15039
作者: Sagar Ghosh,Huy Nguyen,Alessandro Rinaldo
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Mixture-of-Experts (MoE) models combine specialized experts through an input-dependent routing mechanism, and their sparse variants underpin many modern large-scale systems. Yet the statistical cost of adaptive routing remains only partially understood. We study the estimation error of least-squares MoE estimators when both the experts and the gating network are Lipschitz in their parameters. We represent each MoE as a gate–expert pair in a Gated Hilbert Space, whose norm |\cdot|\Delta separates the contributions of routing and expert discrepancies. Using its empirical representation, we derive an oracle inequality with a fast rate of order \tilde\mathcal O\big((Kd\mathrmpr,e+d_\mathrmpr,g)/n\big) for dense MoEs with K experts. For sparse Top- k and Top- p routing, the selection map is discontinuous in the gating parameters. We handle this by partitioning the gating class according to the selection patterns it induces on the sample. The resulting bounds match the dense rate up to an additive routing cost governed by the Natarajan dimension of the selector. We show that such a routing-complexity condition cannot be dropped: Lipschitz continuity in the parameters alone does not suffice. For linear softmax gating, the routing cost is of the same order as the gating term. Finally, for linear experts, we establish a matching minimax lower bound of order Kd_\mathrmpr,e/n , showing that the rates are optimal up to logarithmic factors.
[LG-254] PATTERN Can Be Solved from Memory: Generator-Aware Baselines for Synthetic Node Classification
链接: https://arxiv.org/abs/2610.15020
作者: Amin Najafgholizadeh,Mohsen Afsharchi
类目: Machine Learning (cs.LG)
*备注: 30 pages, 4 figures
Abstract:PATTERN, a synthetic node-classification benchmark still used to rank graph networks, plants a random subgraph in a random host graph and asks which nodes belong to it. Its generator draws 100 patterns once and plants the same ones in every split, so each test graph carries a pattern that also appears in 100 training graphs. A matcher that reads these patterns off the training labels and looks them up in each test graph reaches 99.2 class-weighted accuracy with no trained parameters, against 89.8 for the best published model. Inference under the generator without the patterns reaches 86.8. Two published models, CKGConv and GRIT, score lower on patterns they have not seen: on new graphs with fresh patterns both fall below this pattern-blind predictor, and withholding patterns from CKGConv’s training lowers its accuracy on the test graphs that carry them. Training CKGConv with a new pattern in every graph raises its accuracy on fresh patterns by about 2 points, to the level of the pattern-blind predictor, at some cost on the reused patterns. The same kind of inference under the generator matches the best published result on CLUSTER and, without any training, scores above every published result we found on Minesweeper’s released board. We recommend that synthetic benchmarks report scores on fresh templates beside reused ones, publish baselines built from their generators, and state how their metric is weighted and pooled.
[LG-255] Algorithmic Fairness Over Time: Advances Prospects
链接: https://arxiv.org/abs/2610.15015
作者: Miriam Rateike
类目: Machine Learning (cs.LG)
*备注: Accepted at EWAF 2024, unpublished
Abstract:In the presence of feedback loops, algorithmic decisions at one time step can influence observed data in subsequent steps, potentially introducing selection bias or shifts in data distribution. Research has demonstrated that ignoring these effects can lead to unfairness, which may worsen over time as the model remains deployed within society. This raises important questions of how to measure and ensure fairness of algorithmic decision-making models over time. In this short paper, we present a taxonomy that distinguishes between different types of static and temporal fairness notions in recent advancements. We then suggest prospects for future research by identifying common implicit temporal assumptions, open questions to enhance our temporal understanding of fairness, and other open challenges.
[LG-256] Conformal Sequential Decision Making With Cost and Sublinear Policy Regret Guarantees
链接: https://arxiv.org/abs/2610.15011
作者: Jayadev Naram,Giuseppe Durisi,Matteo Zecchin,Osvaldo Simeone
类目: Machine Learning (cs.LG)
*备注:
Abstract:Automated decision makers, such as AI agents that control robots or network devices, must often satisfy hard cost constraints while maintaining competitive performance. In such settings, a cost measure, such as a safety violation rate, energy consumption, or a distortion level, must stay within a given budget in every episode of operation. Furthermore, a cumulative reward metric, accounting for per-step performance, must stay close to the reward of a meaningful reference policy. We formulate this setup as an episodic sequential decision problem in which the state sequence is arbitrary, and the reward and cost may depend on past actions within the same episode. We propose the online conformal episodic agent (OCEA), a policy that selects one candidate action per episode using a regret-controlled algorithm, and switches to a zero-cost action when the remaining budget cannot absorb the maximum cost of one more round. We prove that OCEA satisfies per-episode cost constraints for every state sequence and for every realization of its internal randomness, while achieving a sublinear expected policy regret with respect to the best action sequence in hindsight from a comparator class satisfying the same per-episode cost constraints. We apply OCEA to risk-averse control and online compression with distortion guarantees.
[LG-257] Characterizing the Separability Gap in Prior-Data Fitted Networks
链接: https://arxiv.org/abs/2610.14989
作者: Julius Kobialka,Manuel Hülskamp,Thomas Nagler,David Rügamer
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Prior-data fitted networks (PFNs) are transformers pretrained on simulated datasets to approximate Bayesian posterior predictive distributions. Standard PFN theory assumes a separable task prior, under which the feature distribution and the mechanism generating labels from features are independent. We show that common PFN priors are non-separable, even in simple settings. Consequently, standard pretraining targets a predictive distribution based on a task posterior updated using both the labeled context and the query features. We characterize its difference from the predictive based on the context posterior alone and show that this difference vanishes under posterior concentration. Under misspecification (i.e., when the true joint distribution lies outside the prior’s support), however, prior non-separability can favor better feature fit at the expense of conditional label fit. We illustrate the consequences for feature attribution, Bayesian optimization, and martingale checks of Bayesian behavior. The same dependence also enables PFNs to learn from additional unlabeled observations. We demonstrate this with SemiTabICL, a PFN explicitly pretrained to use these observations, which improves accuracy over a matched supervised baseline. Our findings highlight the need for explicit rules on unlabeled data use in PFN evaluation and establish the dependence between feature and label distributions as a consequential prior design choice.
[LG-258] SEEK: Agent ic Knowledge Graph Retrieval for Secure Code Generation
链接: https://arxiv.org/abs/2610.14968
作者: Varun Gadey,Markus Tohn,Lichao Wu,Ahmad-Reza Sadeghi,Alexandra Dmitrienko
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注: 21 Pages, 3 figures, Under revision
Abstract:Large Language Models (LLMs) can generate functionally correct code while introducing exploitable security weaknesses, particularly when developers specify functionality without explicit security requirements. Recent state-of-the-art approaches mitigate this problem using external security knowledge, but largely rely on flat similarity-based retrieval over independent artifacts, making it difficult to infer the relevant vulnerability space and retrieve targeted knowledge when direct matches are unavailable. We propose SEEK, an agentic Graph RAG framework for secure code generation that constructs a security knowledge graph using the Common Weakness Enumeration (CWE) taxonomy as its structural backbone and injects it with curated real-world CVE vulnerability-fix instances. At inference time, a Planner Agent predicts potential CWE weaknesses associated with the developer query, which guide Graph Scoping to restrict retrieval to relevant regions of the CWE taxonomy before deterministic multi-faceted candidate ranking. A Selection Agent subsequently validates the predicted weaknesses and retrieved candidates prior to code generation. Across five open-source LLMs, we evaluate SEEK in an out-of-distribution setting on CodeGuard+ and CyberSecEval, spanning C, C++, and Python. SEEK achieves the highest SecurePass@1 of 77.45% on CodeGuard+ and the highest aggregate ICD pass rate of 79.45% on CyberSecEval, surpassing prior approaches. Moreover, SEEK achieves 85.2% SecurePass@1 across 13 unseen CWE types with no directly linked vulnerability-fix instances, demonstrating strong generalizability, while maintaining acceptable inference latency costs.
[LG-259] Cross-Material Magnetic Core Loss Prediction via Operating-Condition Matching and Partial Optimal Transport
链接: https://arxiv.org/abs/2610.14964
作者: Cong Yao,Yu Zheng,Zehao Hou,Rui Xu
类目: Machine Learning (cs.LG)
*备注: 11 pages, 9 figures, 6 tables
Abstract:Cross-material magnetic core loss prediction requires adapting to new material responses with limited measurements. Global feature alignment overlooks differences in source and target excitation coverage. This motivates selecting comparable samples and controlling each source measurement’s alignment contribution. This paper proposes T-PCOT, a Transformer transfer method based on condition-supported partial optimal transport. Frequency, flux excursion, temperature, and waveform descriptors define a sparse graph of comparable operating conditions. A capacity-constrained transport plan distributes alignment across source measurements and leaves costly target mass unmatched. The plan is constructed from measured inputs before training and guides local waveform-feature consistency. An importance-weighted sampling scheme preserves both uniform source regression and the transport objective during minibatch optimization. Experiments on three material pairs and three target label budgets show that T-PCOT reduces mean relative error (MRE) and 95th-percentile error (P95) relative to Transformer-based multiple-kernel maximum mean discrepancy (T-MKMMD) in all nine comparisons. Macro MRE decreases from 12.60% to 9.65%, and macro P95 from 30.94% to 25.06%, corresponding to relative reductions of 23.41% and 19.01%. Controlled comparisons and ablations examine the roles of input processing, condition correspondence, and source capacity. The results demonstrate the predictive advantage of organizing cross-material alignment through operating-condition support over the compared global alignment method. The proposed method achieves state-of-the-art predictive performance in magnetic core loss transfer learning.
[LG-260] Exact Phase-Type Point Processes for Certified Measurement and Prediction of Event Streams
链接: https://arxiv.org/abs/2610.14948
作者: Sayak Dutta
类目: Machine Learning (cs.LG); Statistical Finance (q-fin.ST); Trading and Market Microstructure (q-fin.TR)
*备注: 27 pages, 5 figures, 11 tables. Code: this https URL
Abstract:Temporal point processes are used both to measure market mechanisms, for example how fast prices absorb news and how much of the reaction is the market responding to itself, and to predict the next event. We show that one algebraic object, the phase-type (Erlang-chain) state space, serves both purposes with exact likelihoods. For measurement, we build a multiscale phase-type Hawkes process whose parameters all enter linearly. Its log-likelihood is concave, and we compute the global maximum together with a Fenchel duality-gap certificate. A Markov embedding gives the response to an exogenous shock, echoes included, in closed form, and this response is stable if and only if the spectral radius of the branching matrix is below one. Applied to six markets around 670 scheduled macroeconomic releases (2022-2026) with a built-in placebo test, the model shows that the direct reaction to CPI, payrolls, PPI and the FOMC statement is absorbed in about a second, while the market’s own echoes stretch half of the response to 16-137 s. For prediction, we introduce EPT-X, a neural point process that adds a history-dependent mixture-of-Erlang renewal channel to a linear Hawkes backbone. Its intensity and compensator remain in closed form, its states follow an affine recursion that trains with a parallel scan, and it can start from the certified fit of the measurement model. EPT-X has the best test likelihood on all five LOBSTER order-book streams and on Retweet and Taobao, and is level with the best published model on StackOverflow. We also find that the EasyTPP implementation of IntensityFree clamps short inter-event times before scoring, and that the gaps of the EasyTPP Amazon data are uniform on two fixed intervals.
[LG-261] Compositional Averag e Velocity Modeling for Efficient Diffusion Offline Reinforcement Learning IROS2026
链接: https://arxiv.org/abs/2610.14941
作者: Thanh Nguyen,Abu Hanif Muhammad Syarubany,Hwanhee Kim,Chang D. Yoo
类目: Machine Learning (cs.LG); Systems and Control (eess.SY)
*备注: 10 pages, IROS 2026
Abstract:Diffusion-based policies have recently emerged as a powerful paradigm for offline reinforcement learning, but their reliance on multi-step denoising introduces substantial computational overhead and requires backpropagation through time during training, limiting scalability and practical deployment. In this work, we propose \textbfInstant Flow Q-Learning (InsQL), a single-step generative policy learning framework that retains the expressive capacity of diffusion models while eliminating iterative sampling and BPTT. InsQL builds on the flow-matching perspective of diffusion policies, but replaces marginal velocity parameterization with \emphaverage-velocity modeling and introduces a principled compositional mechanism, yielding a theoretically grounded objective that learns average velocity consistently across scales. This formulation provides a direct noise-to-action mapping and enables single-step action generation. InsQL further unifies compositional average-velocity modeling with behavior-regularized actor–critic optimization in a single framework, requiring only a single policy network and avoiding distillation, multi-stage procedures, and Jacobian computations during training. At inference time, actions are generated in a single forward pass and can be refined via value-guided candidate selection to improve performance. Extensive experiments on standard offline RL benchmarks and real-world robotic manipulation tasks show that InsQL matches or surpasses state-of-the-art diffusion-based methods while substantially reducing both training and inference time. Overall, these results establish compositional average-velocity modeling as an efficient and scalable alternative to multi-step generative policies for offline RL.
[LG-262] CFG-Policy: Unified Diffusion Policy Learning Framework via Classifier-free Guidance
链接: https://arxiv.org/abs/2610.14936
作者: Shutong Ding,Zhaoyang Liu,Yimiao Zhou,Jingya Wang,Jiayuan Gu,Ye Shi
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注:
Abstract:Diffusion policies are increasingly used in robotic control and manipulation, vision-language-action models, and world-model-based agents. Their ability to represent multimodal action distributions offers rich opportunities for exploration in reinforcement learning (RL). However, optimizing diffusion policies with RL can reduce the behavioral diversity and further limit their exploration capability. In RL post-training, directly fine-tuning a pretrained policy may also disrupt its acquired capabilities, while action-space residuals may lack the expressive power for multimodal exploration. Together, these issues reveal a common challenge across online and offline-to-online diffusion RL: improving task performance while retaining the behavioral diversity and capabilities needed for continued adaptation. We introduce CFG-Policy, a unified, plug-and-play framework that separates exploration and exploitation into two generative policies and combines their score or velocity fields during sampling through classifier-free guidance. For online RL, a frozen exploration branch sustains behavioral diversity while an exploitation branch learns from task rewards; for offline-to-online RL, a pretrained exploitation branch remains frozen while a trainable branch steers the composed behavior. CFG-Policy preserves the underlying RL objectives and improves performance with QSM and QVPO on 5 MuJoCo tasks and with Flow-SDE+PPO on 4 RoboTwin manipulation tasks. Project page is at this https URL.
[LG-263] BSIPipe: Bidirectional Sequence-Level Interleaved Pipeline Parallelism for Long-Sequence Large Language Model Training
链接: https://arxiv.org/abs/2610.14935
作者: Yuzhou Ni,Ling Chen
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
*备注: 7 pages, 4 figures
Abstract:Pipeline parallelism has become a critical technique for training large language models. To address the throughput limitation of existing sequence-level methods that schedule subsequences unidirectionally, we propose bidirectional sequence-level interleaved pipeline parallelism (BSIPipe), which co-schedules two opposing V-shaped interleaved pipelines at the subsequence level. To bound the activation accumulation from subsequence scaling, BSIPipe introduces a basic-unit concatenation mechanism that keeps activation memory invariant to the subsequence count. To eliminate communication deadlock risk from bidirectional scheduling, BSIPipe employs a communication optimization strategy that combines topology-aware device mapping with communication-computation overlap. In addition, BSIPipe composes with tensor parallelism to overcome the memory limitation of single-dimensional parallelism, neutralizing its bidirectional memory overhead while maintaining throughput. Experiments on GPT and LLaMA3 demonstrate that BSIPipe achieves average speedups of 1.11 \times to 1.13 \times in long-sequence training over SOTA sequence-level baselines.
[LG-264] LA-Verus: Generating and Solving Temporal Proof Obligations with LLM Agents
链接: https://arxiv.org/abs/2610.14923
作者: Kseniia Strelbytska,Balázs Szilágyi,Anna Mészáros,Ferenc Huszár
类目: Machine Learning (cs.LG)
*备注:
Abstract:The TLA+ (temporal logic of actions) formal specification toolkit has seen widespread adoption in modeling modern distributed software systems including databases, storage systems, ledgers and coordination protocols. This popularity is despite its two significant limitations for software verification: Firstly, TLA+ offers no way to detect drift between software implementation and mathematical specification; secondly, TLC and TLAPS cannot provide satisfactory proofs of arbitrary temporal logic properties. We take a step towards addressing these shortcomings by translating TLA+ specifications into Verus, an auto-active verifier built for Rust, via both a deterministic transpiler and an agentic translator. We demonstrate that our LLM-based pipeline is able to generate proof obligations from TLA+ specifications at scale, and to produce fully machine checkable proofs of both safety and liveness properties. Since temporal logic reasoning in Verus is a largely new task for LLMs, our work yields a valuable evaluation dataset, which we use to map out current generation model capabilities in Verus theorem proving.
[LG-265] Representation Drift in Continual Learning: LoRA versus N-gram Memory
链接: https://arxiv.org/abs/2610.14921
作者: Hao Wu
类目: Machine Learning (cs.LG)
*备注: 28 pages, 7 figures
Abstract:A language model that keeps learning online must not slowly erode its own base capabilities. We ask which part of a modern LLM should absorb continual updates, by measuring how each choice changes the model’s internal representations. On a 48-layer mixture-of-experts model with a native pretrained N-gram memory, we train a seven-task chain under three update mechanisms with identical data and budgets: updating only the N-gram memory table, continually training one shared LoRA adapter, and training a fresh LoRA per task that is merged into the backbone. We capture fixed-context representations at every layer and task boundary. Updating only the N-gram memory changes the base model’s representations of general math and code inputs by 26%/15% (relative RMS), against 64%/35% for merged and 97%/39% for shared LoRA. This drift does not grow with the number of tasks: it reaches 25% after the first task and stays there for all seven, while both LoRA variants keep accumulating. Its generic behaviour is essentially unchanged (no code problem lost, math NLL -0.01), and every change is confined to inspectable, reversible table rows. The cost appears elsewhere: N-gram memory acquires new tasks more weakly and its stored tasks interfere with one another, losing as much old-task NLL as shared LoRA. We show that this forgetting is governed by drift on the learned tasks’ own inputs, measured from the checkpoint at which each was learned, not by generic drift (Spearman rho=0.85; rho=0.71 on Llama-3-8B), and we derive the exact relations behind it: with a frozen head, every pairwise logit margin changes by the projection of drift onto a head-difference direction. These results support a design principle for long-running online learning: write new knowledge only into the memory table and keep the backbone fixed; the remaining problem, interference between stored tasks, then lies inside the memory.
[LG-266] Prediction with Expert Advice under Heavy-Tailed Losses
链接: https://arxiv.org/abs/2610.14904
作者: Yuki Shibukawa,Shinji Ito
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 49 pages
Abstract:We study prediction with expert advice over T rounds with K experts under heavy-tailed losses whose (1+\epsilon) -moment is bounded by \theta0 for some \epsilon\in(0,1] . In this setting, existing algorithms designed for the case \epsilon=1 can be suboptimal when \epsilon1 , suffering an additional polynomial dependence on K . We characterize the minimax pseudo-regret rate as \Theta((\theta T)^\frac11+\epsilon(\log K)^\frac\epsilon1+\epsilon) and establish gap-dependent lower bounds in a self-bounded environment. To attain this minimax rate without prior knowledge of either \epsilon or \theta , we propose Adaptive Heavy-Tailed FTRL, which uses robust data-dependent scale adaptation together with truncation. We further develop Adaptive Heavy-Tailed OMD, which likewise requires neither \epsilon nor \theta and achieves near-optimal regret guarantees in both adversarial and self-bounded environments. Finally, when \epsilon and \theta are known, an FTRL algorithm with Shannon-entropy regularization achieves optimal regret guarantees in both environments.
[LG-267] Pro-t-VAE: Closed-Form Students t Mixtures for Robust Prototype Discovery NEURIPS2026
链接: https://arxiv.org/abs/2610.14894
作者: Zhikang Xu,Jiarui Xing,Jian Wang
类目: Machine Learning (cs.LG)
*备注: Accepted at NeurIPS 2026. Code: this https URL
Abstract:We present a robust deep generative framework for unsupervised image prototype discovery, built on a heavy-tailed Student’s t mixture prior, that bridges statistical robustness and mathematical tractability. Existing generative approaches typically rely on noise-prone Gaussian priors or computationally intensive diffusion models; neither simultaneously provides explicit latent structure nor exact analytic optimization. While diffusion models can yield strong priors, iterative sampling and intractable likelihoods limit scalability and interpretability for prototype discovery. We address these issues by constructing a Student’s t latent space via a Bayesian scale-mixture formulation. This yields a fully closed-form augmented ELBO for unsupervised Student’s t mixture variational autoencoders, which, to our knowledge, is the first such objective applied to robust image prototype discovery, avoiding biased variational approximations and high computational costs in prior heavy-tailed or diffusion-based frameworks. Treating precision as a latent variable automatically modulates each data point’s influence, downweighting outliers to learn a data-dependent measure of trust. The result is improved robustness, accuracy, and efficiency, with well-separated latent representations and more stable manifold learning than state-of-the-art methods. We validate our model on standard vision benchmarks and challenging neuroimaging tasks. In brain MRI-driven neurodegenerative disease analysis, our model recovers clean, anatomically coherent prototypes where clinical noise often collapses rare subtypes. These results highlight the ability of our framework to preserve anatomy-associated patterns for precise discovery in clinical settings. Our code is available at: this https URL.
[LG-268] Contrastive Linearized Laplace: Loss Geometry Valid Repairs and Reliable Cores
链接: https://arxiv.org/abs/2610.14875
作者: Francisco J. Sáez-Maldonado,Luis A. Ortega,Andrés R. Masegosa,Thomas D. Nielsen
类目: Machine Learning (cs.LG)
*备注:
Abstract:While contrastive learning is widely used to learn deterministic representations, it does not inherently provide estimates of uncertainty. Post-hoc uncertainty estimation methods, such as the Linearized Laplace approximation (LLA), offer a way to quantify uncertainty without retraining the encoder. However, we show that directly using the LLA induced distribution over the contrastive embedding does not yield reliable pointwise uncertainty estimates for downstream tasks: embedding variance alone does not reflect whether decisions such as nearest-neighbor retrieval or cluster assignment are stable. To address this limitation, we develop a post-hoc framework based on LLA that propagates encoder uncertainty through the downstream task. We characterize the curvature of contrastive objectives and introduce positive-semidefinite repairs where needed to obtain valid local Gaussian approximations without retraining the encoder. We test the proposed approach in metagenomic binning and image classification tasks. For metagenomic binning, posterior coassignment probabilities support consensus partitions and the selection of reliable cores, leading to improvements over existing baselines. Additional image classification experiments show calibration improvements on standard benchmarks and real histopathology datasets.
[LG-269] Blending experimental electron density information to protein-ligand interaction
链接: https://arxiv.org/abs/2610.14862
作者: Seonghyun Park,Shuan Chen,Jun Jin Choong,Noboru Kuno,Gyubok Lee,Kaushalya Madhawa,Keisuke Ozawa,Hikaru Shindo,Sungsoo Ahn
类目: Machine Learning (cs.LG)
*备注:
Abstract:Machine learning models for protein-ligand interactions commonly rely on atomic coordinates, which are constructed by fitting molecular models to experimental electron-density maps. However, most models represent each complex as a single fitted structure, which can omit information about conformational heterogeneity, positional uncertainty, and local density support for individual atoms at binding interfaces. To fill this gap, we present CoDE, a Coordinate and Density Encoder, pre-trained on experimental density maps. We represent each binding site on a shared three-dimensional voxel grid combining atomic occupancy, experimental electron density, and its gradient magnitude. Using approximately 100K curated complexes, we pre-train a channel-wise vision transformer to jointly reconstruct masked atomic and density channels, learning a representation that integrates structural and experimental information. We evaluate the frozen encoder on two downstream tasks: (i) binding-affinity regression under leakage-controlled splits and (ii) pocket-conditioned ligand generation. Across these settings, CoDE improves over an identically pre-trained coordinate-only encoder and remains competitive with task-specific baselines, showing that density complements fitted atomic coordinates for protein-ligand interaction modeling and that the density-blended representation is useful for structure-based ligand generation.
[LG-270] World2TS: Real-World Event Context for Time-Series Forecasting
链接: https://arxiv.org/abs/2610.14861
作者: Zhiding Liu,Jiqian Yang,Mingyue Cheng,Zhi Li,Enhong Chen
类目: Machine Learning (cs.LG)
*备注:
Abstract:Time-series forecasting plays a vital role in a wide range of real-world applications. Recent approaches have achieved strong performance by modeling temporal correlations and cross-channel dependencies, yet most forecasting benchmarks still represent time series primarily as numerical histories. In practice, time series are observations of complex real-world processes, and their trajectories may reflect events, interventions, disruptions, and changes in measurement or reporting that cannot always be inferred from past values alone. A central barrier to studying context-aware forecasting is therefore the lack of large-scale datasets that align real-world events with the individual series they affect. We present World2TS, a scalable construction framework and large-scale dataset for time-series forecasting with real-world event context. We first develop world2ts-core, a reusable tool that combines agent-led semantic reasoning and Web search, command-line execution and validation, and human review within a feedback-driven discovery workflow. Given heterogeneous time series with semantic descriptions, the workflow discovers candidate events, consolidates retrieved evidence into verified events, and links them to relevant series under explicit temporal and semantic scopes. Using this framework, we construct world2ts-data, which contains 31 datasets across six domains, comprising 4,179 series, 97.1 million observations, 1,159 events, and 56,846 event-series links. We further conduct extensive data analyses and controlled comparative experiments to characterize the dataset and demonstrate the utility of the collected context for improving forecasting accuracy. World2TS provides a reusable data foundation for developing and evaluating context-aware forecasting models and agents. The project homepage is available at this https URL.
[LG-271] rrarium: Growing Personal Assistants via Scalable Synthesis of App Worlds
链接: https://arxiv.org/abs/2610.14860
作者: Mengzhou Wu,Yuzhe Guo,Jiawen Tao,Xiaokun Yuan,Yuan Cao,Gang Cao,Yuetang Deng,Zhi Gong,Dezhi Ran,Linyi Li,Wei Yang,Tao Xie
类目: Machine Learning (cs.LG)
*备注:
Abstract:Advances in large language models (LLMs) have brought the vision of a general-purpose personal assistant closer to reality. Such assistants could provide a unified natural-language interface to a broad ecosystem of applications. However, real-world applications do not readily support training at scale because access is often restricted and their state is difficult to reset. To address this challenge, we introduce Terrarium, a framework that synthesizes executable App Worlds from real-world application information, generates training task packages with evaluator-side rubrics from these worlds, and iteratively validates and refines both through agent rollouts. For App World synthesis, Terrarium organizes public Web information and in-application observations into structured app specifications that guide the construction of self-contained runtimes with callable operations and mutable, resettable state. For task synthesis, Terrarium constructs tasks grounded in App World state together with evaluator-side outcome rubrics that assess success without prescribing reference trajectories. We construct a training suite of 275 App Worlds and 13,344 tasks for outcome-based reinforcement learning across three Qwen3 model scales (4B, 8B, and 14B parameters). Across these scales, models trained with Terrarium outperform the corresponding Qwen3 models by an average of 18.34 percentage points on the \tau^2 -bench and 10.94 percentage points on the AppWorld bench. These results suggest that app-grounded world and task synthesis offers a scalable path toward general-purpose assistants capable of operating across broad application ecosystems.
[LG-272] Disentangling Marginal Capacity from Geometry in Optimal-Transport Label Transfer
链接: https://arxiv.org/abs/2610.14853
作者: Ali Khodabandeh Yalabadi,Azim Dehghani Amirabad,Artem Moskalev,Yun-Ching Chen,Mehdi Pirooznia,Scott Oloff,Rui Liao
类目: Machine Learning (cs.LG)
*备注:
Abstract:Optimal transport (OT) is widely used to align distributions and transfer labels between unpaired source and target data, but source-target composition shifts can cause rare classes to be systematically under-predicted even with high transport confidence. We identify a source of this failure in the transport marginal: under the standard uniform weighting of source points, each class has transport capacity proportional to its source frequency, which can be insufficient to meet its target demand even when separated well in the embedding space. We show that displaced mass is preferentially absorbed by high-capacity classes rather than geometrically similar classes, producing confident errors that are difficult to detect from the transport plan. Downstream read-outs, entropic smoothing, and marginal relaxation provide limited recovery, whereas matching the transport marginal to the target composition substantially improves label transfer. On single-cell data, the oracle composition-matched marginal raises macro-F1 from 0.322 to 0.623, and the best deployable estimator closes 69% of the oracle gap. Correcting the marginal also separates a coverage-dependent component of error from residual error associated with class separability, showing that transport capacity and representation geometry are distinct sources of transfer error.
[LG-273] Sequence-Structured Decision Learning for Branch-and-Bound in Mixed-Integer Linear Programming
链接: https://arxiv.org/abs/2610.14835
作者: Tengfei Liu,Jinrun Liu,Bai Yan,Yuhui Shi,Qi Zhao
类目: Machine Learning (cs.LG)
*备注:
Abstract:Branch-and-bound (BnB) solves mixed-integer linear programs by constructing and exploring a search tree. Node selection and branching are two recurrent decisions that directly determine the tree. Learning-based approaches typically optimize either decision in isolation and condition on a mostly local view, leaving the coupled evolution of the search tree insufficiently modeled. This paper proposes Seq-BnB, a sequence-structured framework that learns branching and node selection from a shared global representation of the evolving BnB process. Seq-BnB serializes node creations, node updates, node-selection events, and branching events into typed tokens, producing a causal event sequence of the search trajectory. A self-attention encoder summarizes this global search history, a bipartite graph neural network encodes the current LP relaxation, and context-conditioned decision heads score branching variables and open nodes. To reduce distribution shift between expert demonstrations and learned policies, Seq-BnB is trained with DAgger-style data aggregation. On four MILP benchmark families, Seq-BnB consistently processes fewer BnB nodes than heuristic and learning-based baselines, with strong solution times on both standard and larger transfer instances. While this paper evaluates Seq-BnB instantiation for branching and node selection, it accommodates additional solver decisions, such as cut selection, by introducing new action tokens and decision heads.
[LG-274] Explore This! Beat and Downbeat Tracking From a Learned Tatum Grid
链接: https://arxiv.org/abs/2610.14817
作者: Robert Kihlborg,André Holzapfel,Jan Schlüter
类目: ound (cs.SD); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
*备注: Accepted at the 27th International Society for Music Information Retrieval Conference (ISMIR), 2026
Abstract:State-of-the-art beat tracking is based on neural networks that model beats and downbeats as individual events, often followed by a separate Dynamic Bayesian Network (DBN) enforcing musical consistency. We propose a novel architecture that combines deep learning with differentiable theory-based modules: Specifically, we design a system that first predicts a tatum grid, then groups tatum events to beats and downbeats, explicitly modelling multiple options for the current tempo and time signature in a bar-pointer-like fashion. Our approach improves continuity metrics compared to existing DBN-free systems and is more versatile than the de-facto standard DBN, but still falls behind the best DBN-based methods on average. Given the many design choices and hyperparameters involved, we assume we have not yet explored the full potential of this idea and publish our code and invite others to explore this further.
[LG-275] Machine Learning-Based Prediction of Dengue Infection Using Clinical and Hematological Data from Bangladesh: A Comparative Analysis of Benchmark and Extended Prediction Models
链接: https://arxiv.org/abs/2610.14813
作者: Neesha Naimeen,Victor Miranda Soberanis
类目: Machine Learning (cs.LG)
*备注:
Abstract:Dengue is a viral infection caused by the dengue virus (DENV) and a major public health concern in Bangladesh, where symptom overlap with other febrile illnesses complicates early diagnosis. This study compares Logistic Regression, Random Forest and XGBoost to predict dengue from clinical and hematological data using two Bangladeshi hospital datasets (2024, n=1,511; 2025, n=1,283) with SMOTE. Extended models outperformed benchmarks within 2024 (Random Forest ROC-AUC 0.640 to 0.661), with RDW-CV the most consistently influential added predictor. 2025 benchmark performance was higher (ROC-AUC 0.904), unexplained by class imbalance and possibly reflecting data quality. External validation is needed.
[LG-276] KedgeFlow: A Programmable Control Plane for Enterprise AI Agents
链接: https://arxiv.org/abs/2610.14810
作者: Jun He,Deying Yu
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注: 12 pages, 1 figure, 3 tables. Code and artifact available at this https URL
Abstract:Enterprise AI agents can execute altered versions of approved actions, concurrently reuse the same permission, or retry operations whose responses were lost. These failures arise when approval and resource execution lack a shared contract. KedgeFlow specifies language-independent contracts for trusted controllers to authorize state-changing operations and privileged gateways to enforce that authority independently of agent reasoning. Inspired by OpenFlow, it connects policy checks supported by evidence, signed permission for an exact action, a shared allowance that can be consumed only once, target-side conditions for applying the change, and retained outcome records. Under trusted issuance of exact-action authority, two conditional properties establish at most one successful claim per allowance and link every governed dispatch to a matching grant, claim, and subsequent dispatch authorization, even when the agent is compromised. A reference implementation exposes this execution path to agent systems through the Model Context Protocol (MCP). In the single-resource evaluation, 6,560 concurrent ledger claim attempts yielded exactly one successful claim for each of 82 allowances. Boundary tests rejected unauthorized and altered requests before consumption. A 32-call gateway race produced one committed mock effect, and a retry after a lost response produced no additional dispatch. Deliberately permissive policy nevertheless authorized an undesirable effect, exposing the limits of execution authorization. Evaluation uses three members of the etcd coordinated key-value store on one host, synthetic authentication and authorization inputs, and a mock target that rejects stale requests. Distributed reservations, enforcement of every production access route, native crash recovery, exactly-once effects at real targets, and interoperability between independent implementations remain untested.
[LG-277] Groups Dont Move as One: A Bregman Clustering View of Linearized Group Influence
链接: https://arxiv.org/abs/2610.14809
作者: Hyunho Choi,Jaeseung Heo,Youngbin Choi,MoonJeong Park,Dongwoo Kim
类目: Machine Learning (cs.LG)
*备注:
Abstract:Influence functions provide an efficient alternative to retraining for assessing the effects of removing or reweighting training data. For groups, however, the standard first-order estimate reduces individual contributions to a sum that can conceal substantial differences among members, including opposing effects that cancel. We establish a Bregman centroid formulation of the linearized group influence objective, interpreting this aggregate as a single representative and using Bregman information (BI) to quantify the variation it omits. Extending this formulation to multiple clusters reveals distinct contributions whose sum recovers the original estimate. We further propose an estimator that jointly optimizes the weights of these cluster contributions, rather than simply summing them. Experiments uncover cluster-specific influence patterns and a positive association between target-specific BI and approximation error. The proposed estimator achieves higher R^2 than standard group influence in predicting target changes under the nonlinear proximal Bregman response function. Simplified clustering and correction variants reduce computational cost. The centroid view thus explains what additive group influence leaves out and provides a basis for using that information to improve estimation.
[LG-278] Formalizing Forgetting: Theory of Forgetting for Online Continual Learning
链接: https://arxiv.org/abs/2610.14787
作者: Renjie Cao,Yifei Cheng,Li Shen
类目: Machine Learning (cs.LG)
*备注:
Abstract:Online continual learning (OCL) requires a learner to acquire a sequence of tasks in a single pass over the data, and catastrophic forgetting of earlier tasks remains its central challenge. Despite the many methods proposed to mitigate forgetting, the theoretical understanding of forgetting in OCL is still limited: existing analyses mostly focus on linear, overparameterized, or neural tangent kernel regimes, and it remains unclear which factors govern forgetting and how they affect it. To fill this gap, we analyze single-pass Shuffle SGD under strongly convex and general convex objectives and derive explicit upper bounds on the expected final forgetting. Further analysis of these bounds shows how the per-task data amount, task similarity, and task ordering enter them, and that their effects can differ across objective classes. Moreover, experiments with neural networks on Split-CIFAR-100 and PMNIST show that the trends suggested by our theory also appear in practice, indicating that these insights extend beyond the convex setting.
[LG-279] CrossMean: Model Merging by Pairing Merged and Expert Activations
链接: https://arxiv.org/abs/2610.14762
作者: Moshe Eliasof
类目: Machine Learning (cs.LG)
*备注:
Abstract:Model merging combines fine-tuned experts into one network without retraining, and closed-form methods compute each merged layer from statistics of its input activations. However, these statistics are measured on a single expert, while in the merged model, each layer receives activations from the merged layers that precede it. Recent work re-measures these statistics in the merged model but still forms them from a single network. We formulate closed-form methods as a linear system and show that they use the same matrix on both sides. We then introduce CrossMean, which replaces this matrix on one side with a bilinear statistic that pairs merged and expert activations. From the same statistics, we also derive a label-free criterion that indicates the task sets in which accounting for the merged activations improves the merge more. Empirically, task sets with a lower criterion value show a larger improvement. On the eight-task benchmark of each vision backbone that we tested and on BERT encoders, CrossMean improves on merging with statistics measured on the experts, and with all linear layers merged, it also improves on re-measuring these statistics in the merged model. These results suggest that merging benefits from pairing merged and expert activations.
[LG-280] Random Fourier Features in Near-Linear Time
链接: https://arxiv.org/abs/2610.14751
作者: Sepideh Mahabadi,Sandeep Silwal,Tal Wagner
类目: Machine Learning (cs.LG); Data Structures and Algorithms (cs.DS)
*备注:
Abstract:Random Fourier features approximate a shift-invariant kernel in \mathbbR^d , such as the Gaussian kernel \exp(-\tfrac12|x-y|_2^2) , using independent samples from its spectral measure to construct an unbiased estimator for the kernel value with subgaussian concentration. However, generating m such features requires O(md) time. A long line of work on structured random features has accelerated this computation for the Gaussian kernel and other radial kernels, albeit with weaker concentration guarantees. We close this gap by showing a random feature map that can be computed in near-linear time O((d+m)\log d) and yields near-subgaussian concentration. We obtain it through a new and tighter analysis of the Fastfood feature map of Le, Sarlós and Smola (ICML 2013), using Brascamp-Lieb inequalities and the Schoenberg representation of radial kernels.
[LG-281] More Domains No Gains? FLO: Fast-Slow Reciprocal Learning for Multi-Domain Graph Pre-training
链接: https://arxiv.org/abs/2610.14748
作者: Yupeng Tong,Lianze Shan,Jitao Zhao,Di Jin,Dongxiao He
类目: Machine Learning (cs.LG)
*备注: 28pages,7 figures
Abstract:Inspired by the great success of foundation models in the development of artificial general intelligence, Graph Foundation Models (GFMs) have recently attracted significant attention and have the potential to advance graph intelligence. Exhibiting such intelligence requires effectively learning knowledge from large-scale, multi-domain graph data during pre-training. Increasing domain diversity during pre-training is expected to yield richer and more generalizable knowledge. Most existing methods pursue this goal by designing domain alignment strategies that help a single graph model learn from multiple domains. However, we observe that existing methods struggle to benefit from increasing domain diversity. Through empirical and theoretical analysis, we reveal two key limitations. Firstly, pre-training in most GFMs is confined to a single message-passing paradigm. Since propagation flattens relations among neighboring nodes, the graph model can hardly perceive the knowledge beyond propagation, namely the attribute-predictable component of these relations. Secondly, domain alignment strategies often rely on randomly initialized parameters that are jointly updated with the graph model using immediate loss feedback, leaving early knowledge acquisition vulnerable to cross-domain interference. To address these issues, we propose FLO, a Fast–sLow reciprOcal approach for multi-domain graph pre-training. Specifically, a propagation-free attribute branch predicts the graph model’s relations without message passing and returns these predictions to refine the model. We further stabilize knowledge acquisition by pairing a fast process that adapts to the current domain with a slow process that integrates knowledge across domains. On 12 datasets, FLO benefits from increasing domain diversity and achieves the best average performance in cross-domain and cross-task transfer.
[LG-282] VitaBench: Memory Lifecycles Beyond Retrieval NEURIPS2026
链接: https://arxiv.org/abs/2610.14731
作者: Muquan Yu,Dihan Hu
类目: Machine Learning (cs.LG)
*备注: Accepted at the NeurIPS 2026 Workshop on Continual Learning in the Era of Foundation Models and Embodied Agents (CL4FMAgents)
Abstract:Long-term memory benchmarks for language agents primarily evaluate whether information from earlier interactions can be retrieved and used when relevant. Persistent memory systems face an additional problem before future use is known: deciding whether a memory should remain exact, be compressed, consolidated, or deleted. Existing benchmarks capture long-range retrieval, temporal reasoning, evidence reuse, and incremental interaction, but do not directly evaluate these earlier representation decisions. We formalize memory lifecycle evaluation and introduce VitaBench, a benchmark that makes lifecycle decisions measurable through repeated memory reviews and full-horizon counterfactual evaluation. VitaBench contains a controlled Diagnostic tier that isolates future-use patterns and a Naturalistic tier that composes them into generated customer-service-style interaction streams. We characterize lifecycle decision structure with three diagnostics: lifecycle headroom, measuring the advantage of adaptive choices over the best fixed strategy; horizon reversal, measuring when short-horizon decisions change under delayed consequences; and optimal-action diversity, measuring whether different memory states favor different lifecycle operations. Across both tiers, VitaBench exhibits substantial lifecycle structure: lifecycle headroom is 1.364 in Diagnostic and 1.405 in Naturalistic, and 60.2% and 56.5% of states, respectively, change their preferred action between short and full horizons. A learned delayed lifecycle policy further improves over the best fixed strategy on VitaBench-Naturalistic, reaching 101.5 versus 83.4 joint utility. These results show that retrieving past information and deciding how it should persist are distinct evaluation problems, and that lifecycle management requires evaluation of the delayed consequences of representation decisions.
[LG-283] Multi-Token Random-Walk Learning via Meeting-Triggered Information Exchange
链接: https://arxiv.org/abs/2610.14716
作者: Taichun Yeh,Xingran Chen
类目: Machine Learning (cs.LG)
*备注:
Abstract:Random walk (RW)-based algorithms are widely used in decentralized learning due to their low communication overhead and scalability. While most existing methods rely on a single token, multiple tokens can explore the network in parallel and exchange information upon meeting. However, the fundamental benefits of such meeting-triggered interactions remain unclear. In this work, we investigate multi-token RW learning under two information exchange policies: \textscAverage-Gradient and \textscAverage-Model, which average stochastic gradients and updated models, respectively, when tokens meet. We establish iteration complexity lower bounds for non-convex, convex, and strongly convex objectives on complete graphs, which apply to both exchange policies. Compared with single-token RW learning, the scaling factor in these bounds is reduced by a factor of m^2 for non-convex and convex objectives, and by an exponentially in m for strongly convex objectives, where m denotes the number of tokens. Numerical experiments reveal distinct behaviors of the two policies under two performance criteria: mean-token loss (the average loss across all tokens) and best-token loss (the minimum loss across all tokens). \textscAverage-Model achieves lower mean-token loss and is relatively insensitive to the network-to-token ratio N/m , where N is the number of nodes. In contrast, \textscAverage-Gradient achieves lower best-token loss but exhibits a trade-off between parallel exploration and gradient interference as N/m decreases.
[LG-284] DeKoR: Decomposed Knowledge Routing for Efficient GNN–LLM Fusion
链接: https://arxiv.org/abs/2610.14710
作者: Peiyuan Li,Yongqi Huang,Jitao Zhao,Di Jin,Dongxiao He
类目: Machine Learning (cs.LG)
*备注: 19 pages
Abstract:Large Language Models (LLMs) have shown strong potential for text-attributed graph learning by leveraging rich pre-trained knowledge and language understanding capabilities. Combining LLMs with Graph Neural Networks (GNNs) offers a promising paradigm for expanding the modeling capabilities of graph learning. However, LLMs and GNNs follow distinct learning paradigms, leading to a pronounced mismatch between their knowledge spaces. Such discrepancies make effective knowledge fusion nontrivial. Moreover, existing methods often incur substantial additional computation for only limited performance gains. To address these challenges, we propose DeKoR, an efficient framework for GNN and LLM knowledge fusion. To enable efficient knowledge transfer, DeKoR decomposes LLM knowledge into complementary local and global views and uses the LLM as an offline teacher to distill them into lightweight experts. To enable selective knowledge fusion, an adaptive router learns how much each node should rely on each expert. Extensive experiments on multiple benchmarks demonstrate that DeKoR outperforms existing baselines on both node and subgraph classification, achieving superior effectiveness and efficiency.
[LG-285] Linear-Time Non-Local Codec Placement for Machine Learning Model Compression
链接: https://arxiv.org/abs/2610.14703
作者: Matteo Spallanzani
类目: Machine Learning (cs.LG)
*备注:
Abstract:On-device inference imposes constraints on model storage and operand traffic and thus benefits from compressed operand representations, which mandate the use of encoding and decoding operators. However, na"ıve placement of codec operators can cancel the operand traffic gains sought from compression. In this work, I present an algorithm to place codec operators into a source computational graph to make it compatible with target hardware. The edges of the graph are colored so that each induced subgraph is weakly connected and has a maximal node that both postdominates all subgraph inputs and dominates all subgraph outputs; this unique node - dubbed the \textitabsolute dominator - is the natural location to observe the traffic flowing through the subgraph and learn parametric codec operators. The algorithm’s time complexity is linear in the source graph’s size. With respect to na"ıve local codec placement, the algorithm reduces modeled operand traffic by 24% - 44% across four models, including 24% on SigLIP and 34% on SmolLM2.
[LG-286] Understanding Hidden Spatial Preferences in Bayesian Optimization
链接: https://arxiv.org/abs/2610.14696
作者: Maria Bånkestad,Jens Sjölund
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Bayesian optimization uses posterior uncertainty to decide where to explore, but this uncertainty is not spatially neutral. On bounded domains, Gaussian-process posterior variance is typically larger near faces, edges, and corners, giving variance-based acquisition functions a systematic boundary bias. This spatial preference is usually inherited from the surrogate rather than chosen as part of the exploration strategy. We separate these roles by keeping the predictive model fixed and modifying the spatial preference of the exploration term. We control the preference with spatial weight fields and amortize their construction with neural surrogates. Controlled experiments show that the Gaussian process’ default boundary bias changes which parts of the domain are explored and which optima are reached. We compare three alternative exploration principles and find, across standard benchmark functions and stochastic robot-pushing tasks, that they can often lower regret, particularly early in the optimization. The spatial structure of posterior uncertainty is therefore not a neutral default: it shapes the exploration strategy and can materially affect optimization performance.
[LG-287] Accelerating Cross-Model LLM Inference via Attention Map Reuse
链接: https://arxiv.org/abs/2610.14688
作者: Zhan Jin,Yuze Luo,Bin Cui,Xupeng Miao
类目: Machine Learning (cs.LG)
*备注:
Abstract:LLM-augmented data workflows often involve multiple heterogeneous models operating over substantial shared context, such as schemas, tool specifications, interaction history, and intermediate results. Although prefix caching can reduce prefilling overhead, existing approaches are largely limited to reuse within the same model because KV states are tightly coupled to model-specific architectures. We present MapFlow, a system for efficient heterogeneous LLM serving via cross-model attention map reuse. Instead of reusing KV states, MapFlow reuses attention maps, which capture token dependency patterns and are more transferable across models. To enable efficient and accurate reuse, we combine partial recomputation with lightweight linear translation across layers and heads. To make attention reuse practical for long contexts, we further introduce block-level sparse compression and a fused retrieval kernel. Experiments show that MapFlow reduces prefilling latency and improves end-to-end serving efficiency while preserving generation quality across heterogeneous model pairs and multi-turn routing workloads. These results suggest that attention maps provide an effective reusable state for heterogeneous LLM serving, especially in prefix-heavy data workflows. Our code is available: this https URL.
[LG-288] One-Way Coupling via Patchwise Adaptive Normalization in Transformer Surrogates
链接: https://arxiv.org/abs/2610.14687
作者: Emmanuel Gottlob,Hugo Thimonier,Ludovico Buizza,August Martens,Alexandre Gavaudan,Thiago Guzella
类目: Machine Learning (cs.LG)
*备注:
Abstract:Many physical systems exhibit one-way coupling: exogenous fields such as medium density or obstacle geometry influence the dynamics without being affected by them. We investigate how explicitly representing these fields can improve the accuracy and computational efficiency of transformer-based PDE surrogates. Building on spatially adaptive normalization, we introduce Patchwise Adaptive Normalization (PAdaNorm), which partitions exogenous fields into patches aligned with the state tokenization. We evaluate PAdaNorm on datasets from three different domains, using a Walrus-inspired transformer architecture. Across the three PDE families, PAdaNorm is on par with or up to 16% better than parameter-matched baselines in mean long-horizon variance-scaled RMSE on 60-step autoregressive rollouts. We also find that PAdaNorm better maintains hard physical constraints over long rollouts, such as preventing flow into impenetrable solid regions in porous media. When the conditioning is static, PAdaNorm caches the conditioning embeddings once and reuses them across the entire rollout, reducing inference FLOPs per step by up to 36% compared with parameter-matched baselines. An ablation over eight conditioning variants shows that modulating a single attention axis (temporal or spatial) suffices and that a lightweight shared MLP across blocks remains competitive with per-block projections.
[LG-289] riage-PiEvo: Evidence Admission for Principle-Evolving Scientific Agents
链接: https://arxiv.org/abs/2610.14683
作者: Peize Li,Yingming Pu,Xingguo Xu,Yiming Cui,Yuansheng Gao,Tao Lin
类目: Machine Learning (cs.LG)
*备注:
Abstract:Large language model (LLM)-based AI Scientists advance scientific discovery, yet they suffer from inefficiencies due to noisy observations. Existing principle-evolving agents act on every anomaly, leaving its source unresolved: a new mechanism or an evaluation artifact, so invalid revisions bias evolution. To address this, we formulate revision as evidence admission: anomalies steer search immediately, but only independently verified evidence may rewrite persistent principles, and contradicted principles are retracted. We realize this in TRIAGE-PIEVO , which pairs an anomaly-responsive search portfolio with a staged verification gate that alone controls what enters principle state. Across four scientific tasks with three backbones, TRIAGE-PIEVO (1) attains the highest average solution quality with +7.5%~11.8% over the strongest baseline per backbone, matching the strongest baseline’s full-budget quality with one-third of the evaluation budget on Qwen3-235B; and additional stress tests and cross-harness deployment demonstrate that (2) in constructed stress tests, it admits none of the corrupted, hidden-source, or exploit evidence that ungated variants admit, while missing no planted mechanism, and (3) plugged unchanged into three external research harnesses, it cuts false evidence admissions by over 60% and raises verified scores by 31%~43%.
[LG-290] Ascent Conditions for Length-Normalized Single-Rollout Policy Updates
链接: https://arxiv.org/abs/2610.14670
作者: Levin David Schwab
类目: Machine Learning (cs.LG)
*备注: 16 pages, 1 figure, 3 tables; code and data included as ancillary files
Abstract:Dividing each trajectory’s score contribution by its realized length can change the direction of a reward-centered policy update. In a finite trajectory model, we derive an exact ascent threshold for a three-outcome softmax policy and exhibit a binary-reward example in which the expected Euclidean update decreases task success. Natural-gradient ascent holds for fully controllable outcome distributions with deterministic binary labels, but can fail under a one-parameter restriction. A third-order U-statistic estimates the first-order return change without bias from independent current-policy trajectories, one per request. In the example, a same-sample plug-in estimator instead has the opposite sign in expectation for batch sizes two through forty-nine. Exact policy-grid enumeration confirms the local sign reversal. In a branching process, length-normalized learning initially loses return, then recovers and exceeds the unnormalized comparator. The analysis concerns the expected update field at the current policy.
[LG-291] agBO: LLM -Driven Task-Aware Graph Bayesian Optimization for Scientific Discovery NEURIPS2026
链接: https://arxiv.org/abs/2610.14626
作者: Xinzhe Yuan,Zimu Zeng,Zhuo Chen,Jianshu Zhang,Jinzong Dong,Jingyi Chen,Nanyang Ye,Huan Xiong,Qinying Gu
类目: Machine Learning (cs.LG)
*备注: NeurIPS 2026 spotlight poster accepted
Abstract:Mixed-variable optimization in scientific discovery often requires exploring large combinatorial spaces under costly experimental evaluations, making sample?efficient Bayesian optimization (BO) essential. However, existing BO methods typ?ically rely on structural descriptors to define similarity among categorical variables, implicitly assuming that structural similarity reflects similarity in task performance. In scientific optimization, this assumption is often violated, leading to surrogate representations that are poorly aligned with the optimization objective. To address this limitation, we propose TagBO (Task-aware Graph Bayesian Optimization), a method that integrates LLM-derived relational priors with descriptor-based struc?tural information for mixed-variable Bayesian optimization. Rather than directly using LLM outputs as predictive scores, TagBO treats LLM outputs as a potentially noisy source of task-aware relational priors and extracts coarse ordinal relational signals from LLM reasoning to construct task-aware similarity structures for scien?tific optimization. TagBO then performs constrained Gaussian process Bayesian optimization over the resulting joint discrete-continuous representation space. Ex?periments on diverse chemistry and materials formulation tasks demonstrate that TagBO consistently improves sample efficiency under fixed evaluation budgets across LLMs of varying capabilities.
[LG-292] From Rainfall to Road-Network Functionality: Transferable Graph Forecasting for Flood-Disrupted Transportation Networks
链接: https://arxiv.org/abs/2610.14622
作者: Junyan Wang,Naiyu Wang
类目: Machine Learning (cs.LG)
*备注: 31 pages, 16 figures, 5 tables
Abstract:Extreme rainfall can rapidly degrade road-network functionality (RNF), disrupting emergency mobility, accessibility, evacuation, logistics, and infrastructure restoration. Existing flood-transportation impact models are often difficult to deploy across regions because they rely on local calibration, intensive simulation workflows or target-region retraining. This study proposes G-R2R, a transferable spatiotemporal graph-learning framework that reframes rainfall-induced road-network disruption as a region-conditioned, topology-flexible graph forecasting problem for unseen transportation networks. The framework integrates ego-centered subgraph sampling, terrain-conditioned graph representation, region-conditioned dual-level static encoding, and dynamic rainfall-network feature learning. Static geographic and infrastructure attributes are fused with rainfall forcing and recent RNF states to forecast node-level RNF across heterogeneous road networks. G-R2R is evaluated using simulation-derived RNF sequences from ten flood-prone regions in Wenzhou, China, under unseen-region testing protocols. It outperforms representative forecasting baselines, achieving a stabilized MAPE of 2.33% and R^2 of 0.855 under the fixed unseen-region benchmark. Leave-one-area-out evaluation yields stabilized MAPE values of 2.30%-3.54%. Ablation analysis supports the contributions of the terrain-conditioned graph representation and dual-level static encoding. A historical-event replay of Typhoon Fitow further shows that G-R2R reproduces principal temporal trajectories and node-level spatial patterns of the simulation-derived RNF reference under observed rainfall forcing without target-region retraining. These results support rapid and transferable node-level RNF forecasting across the regional road-network domains represented in this study.
[LG-293] LLM -Augmented Bayesian Optimization with Adaptive Policies for Chiplet Design Space Exploration
链接: https://arxiv.org/abs/2610.14612
作者: Hang Liu,Zhengfei Chen,Zhengrui Chen,Hao Geng,Yu Li,Qi Sun
类目: Machine Learning (cs.LG)
*备注:
Abstract:Chiplet technology is a key approach for extending Moore’s Law. However, its design process is constrained by two primary challenges: the scarcity of early-stage platforms for rapidly evaluating chiplet-based heterogeneous systems, and the complex relationships between design parameters and optimization metrics. To address these challenges, we propose a synergistic framework that integrates a Large Language Model (LLM) with Bayesian Optimization (BO), enabling architectural-level simulation and automated optimization for the 2.5D/3D CPU-GPU heterogeneous chiplets. We extend the open-source chiplet simulator to evaluate 3D designs by supporting graph topologies. Further, to enhance the exploration efficiency of BO, we introduce an LLM that generates high-quality policies to guide the optimization process, enabled by our proposed adaptive directional policy reasoning (ADPR) and an exploration gain-aware fine-tuning method. Finally, we use the policy-guided BO to perform an automated and efficient exploration of the complex chiplet design space. Experiments on 2.5D and 3D chiplet reveal that our framework improves design space exploration efficiency.
[LG-294] Dissecting the Carbon-SLO Tension in Geo-Distributed LLM Serving
链接: https://arxiv.org/abs/2610.14571
作者: Burak Topcu,Musa Cim,Zuhal Ozturk,Mahmut Taylan Kandemir
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG); Performance (cs.PF)
*备注: 13 pages, 8 figures. Code: this https URL
Abstract:Large Language Model (LLM) inference is driving a growing share of datacenter carbon emissions. Although carbon-aware scheduling can reduce these emissions in geo-distributed datacenters by exploiting the spatiotemporal variations in grid carbon intensity, the latency-sensitive nature of LLM serving constrains this design space. While prior studies investigate the carbon savings achievable through spatial scheduling, they oversimplify the dynamics of modern inference engines, such as queueing and continuous batching, which strongly shape inference latency. We address this issue by designing and evaluating a latency-centric, geo-distributed inference simulator, calibrated with hardware-grounded latency and power measurements, that reproduces these dynamics and enables the study of carbon savings under strict service-level objectives (SLOs). We then formulate a series of SLO- and carbon-aware schedulers of increasing sophistication, from per-request heuristics to batch-level optimization: (i) a greedy heuristic, (ii) an integer linear programming (ILP)-based scheduler jointly assigning batches of requests, (iii) an iterative ILP-based solution projecting its own assignments back into the load estimates to capture load coupling, and (iv) a facility mixed-integer linear programming (MILP) scheduler pricing the energy of batched requests collectively rather than in isolation. Across three LLMs under MLPerf SLOs and varying system loads, the MILP-based option offers the best carbon-SLO balance, reducing carbon emissions by up to 60% relative to the round-robin policy, with an average SLO violation of 3%. Our characterization further shows that savings grow with the spread of carbon intensity across datacenters and shrink as system load increases, while SLO attainment depends on how frequently schedulers receive datacenters’ load updates.
[LG-295] Chunk-wise Model-based Policy Optimization
链接: https://arxiv.org/abs/2610.14564
作者: Weixi Song,Donglin Wang,Yong-Lu Li
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: Accepted to CoRL 2026 Spotlight
Abstract:Model-based reinforcement learning has shown superior sample efficiency by leveraging the learned latent dynamics either in the form of derivative-free planning/imagined rollout training or dynamics gradient-based policy updates. While these approaches achieve comparatively fast learning, they all suffer, to varying degrees, from issues of recurrent compounding errors, high inference-time latency, and instability in backward propagation through time (BPTT). In this work, we propose a model-based RL approach that uses the chunk world model as a differentiable training-time evaluator to optimize the chunk actor without recurrent model invocation. Through the collaborative chunk-wise prediction of the world model and policy, we can generate H-step hypothetical rollouts in a single forward pass and directly optimize value expansion, avoiding the costly and unstable BPTT. We evaluate our method on 19 challenging high-dimensional humanoid tasks from HumanoidBench and deploy our method to the real-world for quadruped bipedal walking. Project Website: this https URL
[LG-296] PGSB: Pretrained-Guided Shared Basis for LoRA Model Merging NEURIPS2026
链接: https://arxiv.org/abs/2610.14558
作者: Muqing Liu,Chongjie Si,Zhuoya Liu,Yuheng Jia
类目: Machine Learning (cs.LG)
*备注: NeurIPS 2026
Abstract:Large pretrained models have achieved remarkable success across diverse domains, yet adapting them to multiple downstream tasks remains challenging: full fine-tuning is costly, while maintaining separate task-specific models is storage-inefficient and fails to produce a unified multi-task model. Low-Rank Adaptation (LoRA) offers an efficient alternative for Model Merging by representing each task adaptation as a low-rank update over frozen pretrained weights. However, existing LoRA merging methods often treat task-specific updates as directly comparable objects and merge them in their original low-rank parameter spaces. This overlooks two key factors: which update directions provide reliable shared support across tasks, and how these directions should be re-parameterized with respect to the pretrained weights before merging. To address this limitation, we propose Pretrained-Guided Shared Basis (PGSB), a two-stage framework for LoRA model merging. PGSB first identifies shared directions among task-specific LoRA updates, and then re-parameterizes these directions using their interaction with the pretrained weights, producing a more suitable basis for merging. Extensive experiments demonstrate that PGSB consistently outperforms state-of-the-art baselines across multiple benchmarks. Codes are available at this https URL .
[LG-297] DEFT: Disentanglement-Enhanced Fine-Tuning for EEG Foundation Models NEURIPS2026
链接: https://arxiv.org/abs/2610.14497
作者: Yuqiu Deng,Muyun Jiang,Zongpeng Zhang,Mingqing Xiao,Yansen Wang,Peiliang Gong,Xinliang Zhou,Yi Ding,Ziyu Jia,Cuntai Guan,Chenyu Liu
类目: Machine Learning (cs.LG)
*备注: Accepted at NeurIPS 2026
Abstract:EEG foundation models (EFMs) have improved EEG representation learning by transferring knowledge from large-scale multi-subject pretraining to downstream tasks. However, their pretrained embeddings often retain substantial subject-related variation, placing an additional burden on the downstream classifier, especially in cross-subject settings. To address this issue, we propose Disentangled EFM Fine-Tuning (DEFT), a plug-and-play adapter for EFM fine-tuning. DEFT encourages branch disentanglement of pretrained tokens into task-oriented content and subject-consistent style, then performs downstream prediction from the content branch. It instantiates this factorize-before-classify design with recursive Gaussian refinement, temporal-variance regularization, subject-aware contrastive learning, and reconstruction. Experiments across multiple EEG datasets and four EFM backbones show that DEFT improves matched fine-tuning baselines. Further ablations and analyses support the intended content/style branch organization.
[LG-298] A Gated Catchment Hydrology Model (GCHM) for Physically-Interpretable Rainfall-Evapotranspiration-Runoff Simulation Across Diverse Hydroclimatic Regimes
链接: https://arxiv.org/abs/2610.14493
作者: M Jawad,HV Gupta,MA Farmani,A Behrangi,GY Niu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Recent advances in machine learning have improved rainfall-runoff prediction, yet most data-driven models remain difficult to interpret and do not explicitly enforce physical conservation laws. We present the Gated Catchment Hydrology Model (GCHM), a mass-conserving architecture based on the Mass-Conserving Perceptron (MCP). GCHM uses five storage states representing canopy interception, snow, soil moisture, groundwater, and streamflow generation. Constrained gates govern precipitation partitioning, evapotranspiration, infiltration, seepage, runoff, and baseflow while learning context-dependent process relationships. We evaluate GCHM across 50 CAMELS catchments spanning snowmelt-dominated, recent-rainfall, historical-rainfall, and arid regimes, using NOAA AORC meteorological forcing and remotely sensed vegetation and land-surface data. Performance is assessed using the Jawad-Kling-Gupta Efficiency (JKGE) metric, which accounts for stationary and non-stationary hydrologic behavior. We compare the full architecture with ablated variants and recurrent neural network benchmarks. GCHM reproduces streamflow dynamics across diverse hydroclimatic regimes, including arid catchments, while maintaining interpretable states and fluxes. Ablation experiments show that simplified architectures may perform adequately under wet conditions but struggle to represent flow persistence and recession during dry periods, when groundwater storage sustains low flows. Learned gates recover hydrologically plausible relationships governing snowmelt, runoff, seepage, and evapotranspiration. GCHM combines competitive predictive performance with mass conservation and interpretability, providing a foundation for hydrologic modeling and counterfactual analysis. Subjects: Machine Learning (cs.LG) Cite as: arXiv:2610.14493 [cs.LG] (or arXiv:2610.14493v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.14493 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-299] Learning to Extrapolate in Depth with Looped Transformers
链接: https://arxiv.org/abs/2610.14456
作者: Huitao Yang,Jianhao Huang,Zixuan Wang,Baharan Mirzasoleiman
类目: Machine Learning (cs.LG)
*备注:
Abstract:Looped transformers reuse a single block across iterations, raising a natural question: how far can a block trained at one depth be safely extrapolated to greater depths? We study this question for all-pairs graph connectivity, where harder graphs require longer-range information propagation. We show that under exchangeability and symmetric positive initialization, population training at depth K converges to an update rule governed by a single effective parameter that balances graph propagation against fitting structure unresolved at training depth. This trade-off creates a finite extrapolation window: too few test-time loops provide insufficient propagation, while too many eventually create false connections. We derive explicit endpoints for this window and bound the graph hardness solvable beyond training as a function of training depth and hardness. In particular, a test depth chosen solely from the training distribution can guarantee recovery on substantially harder graphs, with a certified depth gain that grows exponentially as training depth outpaces training hardness. Experiments validate the predicted windows and show how compute budget and initialization affect the optimal training depth.
[LG-300] From Transfer Probes to Expert Grouping: Evidence from LoRA Post-Training
链接: https://arxiv.org/abs/2610.14437
作者: Xueyang Wu
类目: Machine Learning (cs.LG)
*备注: Technical report for training resources allocation for specialization of LLMs
Abstract:A platform that post-trains one base model for many business lines often has more lines than experts it can maintain, so lines must share experts, yet lines seldom come with a taxonomy saying which should share. We ask whether transfer probes (train on one source, measure paired changes on all sources) can discover a grouping when the standard one is unknown. To test probes against a reference, we use seven public sources that do have a label tree (five mathematics subfields and two kinds of programming, standing in for business lines) and treat the tree’s grouping as a label reference that the evaluator holds and the method does not. With four LoRA experts under fixed per-source exposure and per-expert budgets, we compare a purely probe-derived grouping and an uninformed uniform default (whose tied routing serves every line from one expert) against this reference. Trained once for one interval from a shared warm-up state and routed by a rule fixed by each allocation, neither reached the prespecified practical gain of 2 percentage points, and both had lower overall utility than the reference ( -2.43 and -2.54 pp; 95% intervals below zero). In a post hoc comparison, the probe-derived grouping was indistinguishable from the uniform default ( +0.11 pp, [-1.25,+1.54] ). In fixed-budget LoRA post-training, a grouping built from initial-state single-source probes thus did not deliver the advantage predicted from the probes once carried to a later training state, a finite budget, and specified routing.
[LG-301] ByteMark: A Unified Byte-Level Watermark for Detection Across Language Models
链接: https://arxiv.org/abs/2610.14427
作者: Xiaopu Wang,Runze Li
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:As watermarking of LLM-generated text enters deployment, users increasingly need to check text whose source is unknown. Existing detectors read the watermark through a specific tokenizer or embedding pipeline, and even if all providers adopted the same watermarking method, differences in these representations would leave their watermarks incompatible, and the text alone does not reveal which representation produced it. We propose ByteMark, a unified watermarking scheme that defines its signal on the UTF-8 bytes of published text. Models with different tokenizers embed this signal under one public specification and key, and a single detector verifies the text from the text and key alone. When the tokenizer or embedding model of a baseline detector is replaced, its detection degrades sharply, whereas ByteMark, which uses neither, retains its detection rate; ByteMark also remains detectable when participating models continue or rewrite the text.
[LG-302] PotentialEP: Backprop-Free Training of CIFAR-Scale Generative Models via Equilibrium Propagation ICLR2027 ATC
链接: https://arxiv.org/abs/2610.14409
作者: Alex Gower
类目: Machine Learning (cs.LG); Emerging Technologies (cs.ET); Neural and Evolutionary Computing (cs.NE)
*备注: Under review at ICLR 2027. Code: this http URL
Abstract:Generating an image with a diffusion or flow-matching model typically involves integrating a velocity field for hundreds of sequential steps. That computation suits a GPU poorly and an analog neuromorphic substrate well, where one fast, low-power physical relaxation could replace the whole sequence. Two things are needed: a generative algorithm whose computation is a relaxation, and a way to train it without backpropagation, whose global gradient transport is hard to realise in analog. Energy Matching supplies the first, sampling by gradient-descent relaxation on a time-independent potential - but obtaining the gradient that descent follows is itself a backward pass, once per integration step. We prove that a predictive coding network with a single linear injection term added to its energy relaxes to a state whose visible layer carries that gradient inherently, with an error that vanishes for small injection strengths. An Energy Matching model trained by backpropagation can therefore be loaded into this network unchanged and sampled by relaxation alone, matching the original model’s FID to within 0.04 at CIFAR-10 and ImageNet-32 scale. The network can also be trained without backpropagation: Equilibrium Propagation reproduces backpropagation’s parameter updates to a cosine similarity of 0.9998, and training end to end with those updates reaches an FID of 3.64 on unconditional CIFAR-10 against 3.54 for backpropagation. To our knowledge this is the first CIFAR-10 generative model trained with Equilibrium Propagation, and the first at competitive quality by any backpropagation-free rule. This quality does not require group normalisation or attention: an architecture of only convolutions and pointwise nonlinearities, both native to an analog crossbar, reaches an FID of 3.51 when trained by backpropagation.
[LG-303] Sparse-TD-JEPA: Improving Zero-Shot Reinforcement Learning With Sparse Latent Task Representations
链接: https://arxiv.org/abs/2610.14400
作者: Duaa Nakshbandi,Zhiyu An,Wan Du
类目: Machine Learning (cs.LG)
*备注:
Abstract:Successor-feature-based zero-shot reinforcement learning methods pre-train a task encoder and successor-feature predictor on offline, reward-free data, then recover a policy at test time by solving a closed-form linear regression over a small labeled dataset. We identify an underexplored limitation of this approach: when the task encoder produces dense representations, the inferred reward signal becomes diffuse across the state space, reducing the contrast between goal-relevant states and the background and providing a weaker signal for the policy to follow. We address this limitation by introducing an l1 sparsity penalty that concentrates reward-predictive structure into a smaller effective subspace of the task encoder’s latent space. The penalty is applied only during training and requires no architectural change or modification to the closed-form inference procedure. We instantiate this approach as Sparse-TD-JEPA, a single-line of code addition based on Temporal-Difference Joint-Embedding Predictive Architecture (TD-JEPA) that significantly improved its performance in zero-shot reinforcement learning. We evaluate Sparse-TD-JEPA on 65 tasks across the DeepMind Control (DMC) Suite and OGBench, covering locomotion, navigation, trajectory stitching, puzzle solving, and robotic manipulation. Sparse-TD-JEPA improves over TD-JEPA in all 13 domain-level comparisons. On DMC, it increases the average return from 661 ± 6 to 714 ± 3. On OGBench, it increases the average success rate from 38.0 ± 0.8 to 48.4 ± 2.3, with improvements of 13.4-18.0 percentage points on the evaluated antmaze navigation and stitching domains. Overall, Sparse-TD-JEPA achieves the highest mean performance among established zero-shot baselines in nine of the 13 domains. Analysis further shows that sparsity produces a sharper and more task-informative inferred reward landscape.
[LG-304] Clinical value claims in multimodal AI outpace supporting evidence
链接: https://arxiv.org/abs/2610.14395
作者: Jiayu Feng
类目: Machine Learning (cs.LG); Computers and Society (cs.CY)
*备注:
Abstract:Multimodal prediction models add a data source, such as imaging or an assay, to clinical information, and health systems must decide whether that added test is worth acquiring. We audited 177 studies comparing multimodal with unimodal models: 81 from a systematic review and 96 from a probability sample of 2025-2026 PubMed records. Gains in the area under the receiver operating characteristic curve (AUROC) were smaller when a study fitted a separate model for every data source (adjusted difference, -0.030; 95% CI, -0.056 to -0.005). Within 78 such studies, leaving out the strongest source’s model would have raised the average gain from 0.046 to 0.108. This complete benchmarking fell from 69.1% of review studies to an estimated 46.0% of recent ones. An estimated 22% assessed decision utility and none evaluated a model prospectively, yet most claimed potential clinical usefulness or improved care (84.0% and an estimated 87.7%).
[LG-305] DPAL: A Test-Time Adaptation Loss for Financial Time Series Forecasting
链接: https://arxiv.org/abs/2610.14376
作者: Wuzhou Lian,Haochun Wang,Yuefan Deng
类目: Machine Learning (cs.LG)
*备注:
Abstract:Test-time adaptation (TTA) enables time series forecasters to update after deployment as new observations become available. However, adaptation objectives that focus primarily on prediction errors at individual time steps do not fully address the needs of financial time series forecasting. In addition to pointwise return fitting, return direction and the price trajectories reconstructed from sequences of returns are also important. To better account for these properties, we introduce Direction- and Price-Aware Adaptation Loss (DPAL), which combines pointwise return fitting with directional and price-calibration terms. We apply DPAL within existing TSF-TTA frameworks without modifying the source forecaster or the adapter architecture. Across a diverse collection of financial datasets, DPAL reduces reconstructed-price mean squared error in most evaluated settings.
[LG-306] When More Memory Fails: Hidden Feasibility Holes in Dynamic Tensor Rematerialization
链接: https://arxiv.org/abs/2610.14369
作者: Mahesh Reddy Pagadala
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
*备注: 8 pages, 2 figures. Simulator study. Code and execution records: this https URL
Abstract:Dynamic Tensor Rematerialization (DTR) trains neural networks under a memory budget by evicting tensors and recomputing them on demand. A natural expectation is that a budget large enough to complete training stays sufficient when it is increased. Using DTR’s reference simulator, simrd, to replay the public DTR training traces, we show that this expectation fails. We call the exceptions feasibility holes: memory budgets at which an eviction policy runs out of memory although both a smaller and a larger budget complete, a form of non-monotone feasibility. Under a written sweep protocol with dated amendments, a 0.01-step budget grid shows no such hole for DTR on any of six confirmatory traces, although unresolved timeouts prevent it from establishing monotonicity. Finer sampling finds holes on InceptionV4 (three regions) and U-Net (two regions), in addition to one on the ResNet-32 development trace. Every hole is reproduced by three repeats and by unmodified simrd. For one InceptionV4 pair (0.2343 versus 0.235 of peak memory) we trace the failure to individual allocations. The larger budget avoids one eviction at an operator where the smaller budget was 173,479 bytes short. The extra data it carries forward forces a later eviction of a tensor needed three operators afterwards, and rebuilding that tensor triggers a recomputation cascade that cannot fit. Keeping that one tensor resident under the same budget prevents the failure. The two simple score changes we tested do not remove holes across traces; both add failing budgets on some traces. All results are from simulation, with no allocator model; behaviour on GPU hardware is untested.
[LG-307] History Adaptation and Hidden Gain: Two Prospective Tests of Learned Predictive Control in a Synthetic Neural Population
链接: https://arxiv.org/abs/2610.14368
作者: Mahesh Reddy Pagadala
类目: Machine Learning (cs.LG); Systems and Control (eess.SY)
*备注: 13 pages, 4 figures. Synthetic closed-loop neural rate-control study; code, frozen protocols and results at this https URL (archived: this https URL )
Abstract:We study learned one-step predictive control of a synthetic population of 64 leaky integrate-and-fire neurons whose light sensitivity is scaled by a hidden factor s, using two prospective confirmations on plants simulated only after the analysis was frozen. In the first (50 plants, 11 conditions), removing the 200-ms input-output history substantially worsened a parameter-count-matched neural predictor at low and moderate gain (s = 0.6, 0.8; RMSE rose by 2.47 Hz at s = 0.6, 97.5% CI 2.43-2.51). In secondary comparisons, recursively adapted ARX had the lowest mean error of the deployable controllers in 10 of 11 conditions, tuned PI with anti-windup matched or beat the network for s = 0.7, and training for 10,000 rather than 900 epochs lowered prediction error but worsened control. Development work suggested the network’s learned action sensitivity did not track the hidden gain. A second protocol (50 new plants, 9 sensitivities, 2 noise levels) confirmed three pre-specified predictions at the primary noise level. Re-estimating a single multiplicative input gain recovered 87% (95% CI 86-88%) of the benefit of full adaptive ARX. Supplying that online estimate to the network as an extra input reduced high-gain tracking error by 1.84 Hz [1.81, 1.88]. Longer training worsened the plain network by 1.09 Hz but improved the gain-input network by 1.07 Hz, a differential of 2.16 Hz [2.13, 2.20]. Under high process noise (sigma = 2.0) the effects shrank to about 0.04 Hz or less and gain-only adaptation recovered only 24% of the adaptive-ARX benefit. In this synthetic plant, at the primary noise level, one-parameter online gain estimation accounts for most of the gap between a history-based network and full adaptive ARX, and for its prediction-control mismatch.
[LG-308] ConformalGuard: False-Positive-Controlled Action Gating for LM Classifiers
链接: https://arxiv.org/abs/2610.14365
作者: Bishnu Bista,Dhanraj Haibatrao Jadhav,Vibha Choudhary,Gaurav Saxena,Arun Pandiyan Perumal,Arjun Subedi,Jignesh Ammineni
类目: Machine Learning (cs.LG)
*备注: 12 pages. Published at COLM 2026. Code: this https URL
Abstract:Language model classifiers are increasingly deployed for high-stakes decisions such as phishing screening, content moderation, and fraud triage, yet validation-set thresholds provide no finite-sample control of deployment-time false positives. We present ConformalGuard, a model-agnostic calibration layer that wraps a fixed scoring function with finite-sample control over the rate at which legitimate inputs trigger an intervention tier. Our contribution is empirical and operational, not a new conformal theorem: zero-shot LM API scores can be coarse enough (10–11 unique values here) to make temperature scaling and isotonic regression operationally identical to raw thresholding for low-FPR action gating. We compare these methods with conformal calibration for fine-tuned BERT, Claude Sonnet 4, and GPT-4o on public email corpora. For Claude, conformal calibration reduces FPR from 0.040 to 0.008 while recall falls from 0.938 to 0.875. For deterministic GPT-4o scores, the 1% operating point is too conservative to deploy (recall 0.564); three-run stochastic averaging raises recall to 0.732 at FPR 0.005. At 2% prevalence with BERT, conformal calibration removes 25 false-positive reviews per 10,000 emails (95% CI [10, 43]). The guarantee is legitimate-event-rate control under exchangeability; discrimination, workload effects, and robustness to shift remain empirical.
[LG-309] he Sample Complexity of Multicalibration with an ERM Oracle
链接: https://arxiv.org/abs/2610.14363
作者: Princewill Okoroafor
类目: Machine Learning (cs.LG); Data Structures and Algorithms (cs.DS); Computer Science and Game Theory (cs.GT)
*备注: 58 pages
Abstract:A predictor is multicalibrated with respect to a class of subpopulations (formally, of functions on the inputs, called auditors) if, for every subpopulation in the class and every value the predictor outputs, the average outcome among the members of that subpopulation who receive that value is close to the value itself. For finite classes, about \varepsilon^-3 samples are necessary and sufficient to learn a predictor with multicalibration error \varepsilon , but the algorithms that attain this rate enumerate the class, which is impossible for the large classes of interest. We ask how many samples suffice when the learner accesses the class only through an empirical risk minimization (ERM) oracle. We give an algorithm that uses \tildeO(d/\varepsilon^3) samples and polynomially many ERM calls, where d is the VC dimension of the class. Its dependence on \varepsilon is optimal for all learners. When d is at least about 1/\varepsilon , every learner that can only access the class through polynomially many ERM calls needs \tilde\Omega(d/\varepsilon^3) samples. Learners with unlimited computation need only \tildeO(\varepsilon^-3+d\varepsilon^-2) samples, so the extra factor 1/\varepsilon on the dimension term is the price of oracle efficiency. For swap multicalibration, which allows a different function from the class for each predicted value, \tilde\Theta(d/\varepsilon^3) samples are necessary and sufficient. For estimating the error of a given predictor, the ordinary notion is the harder one: the swap error can be estimated with an ERM oracle, whereas the ordinary error is NP-hard to approximate even for classes with linear-time ERM. Consequently, there are classes for which it is computationally easier to learn a multicalibrated predictor than to compute the multicalibration error of a given predictor.
[LG-310] Class-Conditional Conformal Prediction for Signal Temporal Logic Inference with Many Classes
链接: https://arxiv.org/abs/2610.14352
作者: Yixuan Wang,Danyang Li,Matthew Cleaveland,Roberto Tron,Mingyu Cai
类目: Machine Learning (cs.LG)
*备注:
Abstract:Signal Temporal Logic (STL) inference provides interpretable formulas for time series classification. For multiclass classification, however, accurate predictions do not necessarily ensure that the inferred formulas are satisfied consistently with the true class. We introduce the positive satisfaction rate and codeword satisfaction rate to quantify satisfaction of required formulas and consistency with the complete class codeword, respectively. We develop a two-stage procedure that first learns discriminative formulas using a margin-based objective and then refines predicate thresholds toward a small positive robustness target while fixing the formula structure and temporal intervals. Satisfaction and ranking penalties discourage violations of required formulas and loss of class separation. The decoding loss is used as a nonconformity score and calibrated separately for each class, providing finite-sample class-conditional coverage under intra-class exchangeability. Experiments on naval, synthetic, and robot manipulation trajectories demonstrate discrepancies between classification accuracy and formula satisfaction. The results further reveal a data-dependent tradeoff between robustness tightness and satisfaction, while class-conditional calibration mitigates undercoverage that can be hidden by pooled calibration.
[LG-311] On the Depth Transfer of Neural Optimizers via L-Smoothness
链接: https://arxiv.org/abs/2610.14344
作者: Maximilian Wong,Chenyu Xu,Jiajin Li,Yiping Lu
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:Scaling deep learning models efficiently requires predictable performance transfer across compute regimes. Establishing reliable hyperparameter transfer allows optimal training dynamics and structural choices to be discovered on low-cost proxy models, avoiding catastrophic failures during massive-scale training. Existing depth-transfer analyses are architecture-specific and heuristic: hyperparameter transfer is only justified through toy blocks, and both rely on infinite-depth dynamics, a specific initialization, and feature-learning criteria. Building on L -smoothness framework of Xu et al. (2026), we understand hyperparameter transfer as how do the Lipschitz and L -smoothness constants of the loss scale with the depth K ? We show that inverse-depth residual branch scaling \alpha=\Theta(1/K) is sufficient to make both constants depth-independent at fixed width, uniformly over the bounded parameter domain and for arbitrary p\to q geometries. Because the bounds depend on each residual branch only through its Lipschitz and smoothness constants, the argument applies to general residual networks (including branches of block depth \geq 2 such as transformer and ResNet blocks) without assumptions on initialization or correlations between layer weights. Since the smoothness constant controls the curvature penalty in the descent inequality, our bounds motivate a learning-rate scale without an additional depth-dependent curvature correction for updates bounded in the prescribed norm. This provides a geometric explanation of inverse-depth residual scaling and its application to the operator-norm geometries associated with AdamW, Muon, and MOGA. Learning-rate sweeps on GPT-2 (depths 2–128) and CIFAR-100 ResNets (8–32 residual blocks) confirm that Muon and MOGA exhibit stable learning-rate transfer under these prescriptions, with a common optimal rate across a 64-fold increase in depth.
[LG-312] Continuous-Time Cloth Dynamics with Physics-Informed Neural ODEs
链接: https://arxiv.org/abs/2610.14342
作者: Yingdong Ru,Lipeng Zhuang,Zhuo He,Gerardo Aragon Camarasa
类目: Machine Learning (cs.LG); Robotics (cs.RO)
*备注:
Abstract:We present ContinuousCloth, a continuous-time, physics-informed neural solver for deformable cloth dynamics that enforces a spring–mass ODE within a neural trajectory model. Unlike classical time-stepping simulators that return trajectories only on a discrete integration grid, ContinuousCloth supports querying at arbitrary timestamps and computing displacement, velocity, and acceleration via automatic differentiation. We evaluate ContinuousCloth along two dimensions. In a physics-only setting, ContinuousCloth achieves trajectory accuracy comparable to high-precision solvers across six diverse boundary-condition scenarios while reducing runtime by up to 17.4 times. When fused with sparse and noisy observations, the spring–mass ODE acts as a strong inductive bias for downstream tasks: interpolation and parameter estimation. That is, ContinuousCloth reduces interpolation error by 7.3 times over a data-only neural baseline in simulation and 2.4 times on real-world cloth sequences, and achieves 2.6 times lower parameter error than a differentiable-solver baseline under high-noise conditions, demonstrating consistent robustness under realistic measurement imperfections.
[LG-313] Decentralized Bandit Learning of Markov Perfect Equilibria
链接: https://arxiv.org/abs/2610.14333
作者: Felix Höfer,H. Mete Soner
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注: 23 pages, 1 figure
Abstract:We study decentralized bandit learning in discounted infinite-horizon, general-sum Markov games with continuous actions. Agents observe the public state, but only their own realized rewards, and use simultaneous perturbation stochastic approximation to update their actions. Under strong monotonicity of the negative running-reward gradients and sufficiently small time steps, we prove that the fully decentralized procedure converges almost surely to the unique stationary Markov perfect equilibrium. We obtain an O(n^-1/2) convergence rate and present a natural extension to episodic finite-horizon learning.
[LG-314] Learning to See through Action: Predictive Curiosity for Robots
链接: https://arxiv.org/abs/2610.14302
作者: Stelios Zarifis,Panagiotis Filntisis,George Retsinas,Petros Maragos
类目: Machine Learning (cs.LG); Robotics (cs.RO)
*备注:
Abstract:Humans and animals do not learn vision from a fixed set of data: they move, interact, and learn from the visual consequences of their actions. Motivated by this developmental view, we study whether robots can improve visual learning by actively collecting experience through curiosity-driven interaction. We introduce DySCOVER, a self-supervised framework for active robot visual learning. A robot observes RGB frames, learns a visual autoencoder that encodes and reconstructs them, and trains a recurrent world model to predict the next visual latent state conditioned on the current state and action. The agent’s policy is trained only with intrinsic rewards: reconstruction error encourages visually novel observations, while the world model’s error encourages transitions whose consequences are not yet predictable. This prediction error biases the policy toward more intricate trajectories that enrich the agent’s set of experiences with observations beyond those encouraged by visual novelty alone. Our experiments conducted with robotic arms as the embodied agents show that adding predictive curiosity improves the vision model over using visual curiosity alone. Across variations in initial conditions, experience capacity, embodiment, and scene content, adding predictive curiosity consistently improved visual learning. These results show that predictive curiosity can improve active visual learning beyond reconstruction-based curiosity alone.
[LG-315] Diagnosing Generalization Boundaries in AI-Based Vulnerability Detection via Calibrated Embedding Distances
链接: https://arxiv.org/abs/2610.14296
作者: Konstantin Ievtushenko,Ralf Lämmel,Jan Jürjens
类目: oftware Engineering (cs.SE); Machine Learning (cs.LG)
*备注: Submitted for review
Abstract:High benchmark scores in learning-based vulnerability detection mask severe degradation under distribution shifts. This paper demonstrates that calibrated embedding geometry exposes these hidden generalization boundaries on syntactically similar vulnerable and patched code pairs. An evaluation of CodeBERT and Qwen2.5-Coder across pattern, project, and temporal shifts on a strictly paired dataset reveals three main findings. First, whitened intermediate-layer embeddings with local nearest-neighbour scoring separate unfamiliar samples with a mean out-of-distribution AUROC of 0.739 (RQ1). Second, distance scores reveal a shift-specific reliability gradient where accuracy drops by over 35 percentage points and cross-project shifts cause the sharpest degradation, even though temporal shifts drift geometrically farthest (RQ2). Third, false positives on patched code systematically correlate with geometric proximity to vulnerable training neighbourhoods, increasing from 9.4% in the lowest-risk decile to 58.8% in the highest (RQ3). These calibrated distances establish a diagnostic signal that maps representation geometry directly to model degradation without requiring auxiliary models.
[LG-316] Layer-wise Input Recoverability in Deep Neural Networks
链接: https://arxiv.org/abs/2610.14290
作者: Maxwell Spiridonov,Alireza Rahmansetayesh,Pierre Baldi
类目: Machine Learning (cs.LG)
*备注:
Abstract:The inner workings of deep neural networks are still not well understood. The information bottleneck framework, which characterizes deep networks by the mutual information between their hidden representations and the network’s input, has been used to identify compression of input information with respect to an input image and output labels. We instead study how much of a layer’s own input is recoverable from the layer’s output in convolutional neural networks (CNNs). A convolutional layer applies its filters to one input patch at a time, so we evaluate recoverability on sampled input patches with a reconstruction-based probe and not on the entire feature map. The probe’s size thus depends on channels and kernel size and not on feature map size, which lets our method scale to deep CNNs with wide feature maps. For each layer, we compare pretrained, random, and reconstruction-optimized encoders, using linear and linear-plus-residual-MLP decoders. We measure recoverability as R^2 between input patches and their reconstructions on a held-out set. Across nine ImageNet CNNs, recoverability generally decreases with depth but is more consistently predicted by the compression of the layer. Learning-attributed recoverability, the difference between pretrained and random encoders, is not uniform across architectures. We introduce single-property random controls, which transfer one pretrained property onto a random encoder, and find that each control changes recoverability with a consistent sign across models: copying pretrained singular vectors tends to raise recoverability, while copying the pretrained singular-value spectrum or activation sparsity lowers it.
[LG-317] Learning to Sync at the Edge: Client-Side Adaptive Synchronization for Federated Learning
链接: https://arxiv.org/abs/2610.14268
作者: Ebasa Temesgen,Ammar Ahmed,Nathan Lechamo,Ali Anwar,Maria Gini
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG); Networking and Internet Architecture (cs.NI)
*备注: 9 pages, 8 figures, 8 tables. Accepted to IEEE SEC 2026
Abstract:The scalability and performance of edge federated learning are fundamentally constrained by synchronization bottlenecks. Clients operate under heterogeneous compute, time-varying communication conditions, and, in some settings, privacy constraints, yet synchronization is often still handled with fixed schedules. However, synchronization decisions are often made outside the client, even though the client directly observes local progress, resource state, and communication conditions. We present a client-side reinforcement learning synchronization framework in which each client decides during training whether to upload, download, synchronize bidirectionally, or defer communication. The framework places a lightweight scheduler alongside the local model and formulates synchronization as a resource-aware control problem driven by local learning and system state. We study this framework on two representative edge workloads, spatiotemporal traffic forecasting and image classification. Across both settings, adaptive client-side synchronization improves the accuracy-cost trade-off, reducing system cost in traffic forecasting while maintaining competitive cost and improving final accuracy in image classification. We also consider privacy-constrained settings, where synchronization must account for not only latency and energy but also privacy expenditure. Overall, our results show that synchronization in edge federated learning is better treated as an adaptive client-side control problem than only as a server-side coordination mechanism.
[LG-318] Variance Fingerprints: Component-Wise Identification of Content Factors by Augmentation Design
链接: https://arxiv.org/abs/2610.14260
作者: D. Hudson Smith
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 28 pages, 4 figures
Abstract:Augmentation-based self-supervised learning can recover the content shared across views without separating its underlying factors into individual coordinates. We ask whether augmentations can be designed to make those factors identifiable. Our approach allows content factors to fluctuate across augmented views while preserving their mean for each source instance. Each instance receives a fixed, recorded set of augmentation strengths that controls how much its factors fluctuate. When factors respond differently to changes in these strengths, their patterns of variation provide information that distinguishes them. Under suitable decoder and support assumptions, we prove that these differences identify individual factors in a conditional Gaussian model, up to permutation, scaling, and translation. We then develop a learner that uses the recorded strengths to estimate these patterns jointly across instances, enabling learning from few views per instance. Synthetic experiments demonstrate near-perfect factor recovery through nonlinear mixing. Controlled dSprites experiments show improved coordinate alignment using augmentations that target known factors, even without telling the learner which augmentation corresponds to which representation coordinate. These results suggest a route to component identification in self-supervised learning: augmentation design can supply identifying information beyond shared content alone.
[LG-319] Do Artificial Neural Networks Need Dendritic Nanotubes? Two Reason s for Shrinking Importance Estimates in Continual Learning
链接: https://arxiv.org/abs/2610.14258
作者: Saurav K. Aryal
类目: Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
*备注: Github: this https URL 21 pages, 3 figures, 10 tables. Simulation and figure scripts are included as ancillary files
Abstract:Regularization-based continual learning protects each parameter in proportion to an estimate of its importance, yet these estimates come from finite samples and heuristics and are used as if they were exact. We report four studies that start from a biologically inspired mechanism and end with a statistical explanation of it. Motivated by dendritic nanotubes, we let transient lateral links diffuse importance tags between hidden units. Links between co-active units gave no significant gain, but random links with lifetimes of 25 or 200 steps raised final accuracy on 20-task Permuted MNIST by 1.5 to 2.6 points. We show that random diffusion implements linear shrinkage of importance toward the layer mean in expectation. A simulation of the link process alone predicts shrinkage weights of 0.23 and 0.52 for these lifetimes, and graph-free shrinkage at 0.25 and 0.5 reproduces both the accuracy and the dispersion of the consolidated importance, while every activity-dependent rule did worse than random links. A five-seed screen across EWC, SI, and MAS found nominal gains for SI and MAS on Permuted MNIST and losses on task-incremental Split CIFAR-100, where forgetting was small and shrinkage cost plasticity. Varying the number of importance samples then separates two mechanisms on Permuted MNIST. For EWC, the gain fell by about ten points per tenfold increase in samples, from 19.6 points with 10 samples to -0.4 with 1000, a variance effect. For MAS, a gain of about one point persisted at 1000 samples, consistent with a bias in the importance measure rather than noise. Shrinkage had no detectable effect for MAS on a domain-incremental CIFAR-100 benchmark. Finally, we show that a standard heteroscedastic empirical Bayes rule shrinks the largest Fisher-type estimates most, which may explain its poor results, and point to a log-scale alternative.
[LG-320] Adaptive-Depth Is a Finishing Stage NEURIPS2026
链接: https://arxiv.org/abs/2610.14252
作者: Andrei Piterbarg,Sean McLeish,Dhruv Pai,Benjamin Keigwin
类目: Machine Learning (cs.LG)
*备注: Accepted for poster presentation at the AXIOM Workshop (Foundations of Efficient Deep Learning) at NeurIPS 2026. 9 pages, 8 figures
Abstract:Looped transformers reuse the same layer multiple times, allowing inference compute to be adjusted without changing the model. Recent adaptive-depth methods train the model across multiple loop depths from the beginning of training. We introduce fixed-first training, which varies model depth only during a short final phase of training. At 200M parameters, training a model of fixed-depth 2 for the first 67% of the run reduces modeled training compute by 45% while keeping mean negative log-likelihood (NLL) across inference depths within 0.5% of training across depths throughout. At 1B parameters, training initially at depth 4 reduces modeled compute by 31% to 37% and wall-clock time by 32% to 39%, while keeping final NLL within 0.8% and downstream differences within the observed seed variation. These results show that training over the full range of depths can be reserved for a short finishing phase with only small changes in performance.
[LG-321] Learning Non-Gaussian SEMs with Conditional Risk and Why the Direction of Recursion Matters
链接: https://arxiv.org/abs/2610.14246
作者: Ujas Shah,Panagiota Birmpa,Rebecca Morrison
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 15 pages, 5 figures
Abstract:Variance-based methods recover a Gaussian linear structural equation model (SEM) with equal or suitably constrained error variances, either by recursively identifying the source nodes with smallest conditional variance (top-down method) or sink nodes with highest conditional variance (bottom-up method). We show that the use of conditional variance is incidental: replacing it with the conditional risk, i.e., the minimal expected loss of linear prediction under the negative log-likelihood of the noise distribution, extends both methods to non-Gaussian noise. Existing methods are the least-squares special case. This yields two main results. First, exploiting the noise distribution improves structure recovery: the matched-loss procedures recover the SEM more accurately than their least-squares counterparts and the LiNGAM methods, which are designed for non-Gaussianity. Second, the two methods, interchangeable in the Gaussian setting, separate under non-Gaussian noise: the top-down method is correct for every noise density with finite differential entropy, whereas the bottom-up method requires a regularity condition and can fail without it, for instance under mean-zero exponential or Cauchy noise. Simulations confirm both results.
[LG-322] Embedding Graph Kernel Entropy (EGKE): Fine-grained Hallucination Detection for LLM s from Embeddings
链接: https://arxiv.org/abs/2610.14214
作者: Dilrangi Sankalpana,Pragatheeswaran Vipulanandan,Uthayasanker Thayasivam,Kamal Premaratne
类目: Machine Learning (cs.LG)
*备注: Accepted for ICKG 2026
Abstract:Hallucination detection is critical for ensuring the reliability of Large Language Models (LLMs) in real-world deployments. While recent works, such as Semantic Entropy (SE) and Kernel Language Entropy (KLE), identify hallucinations by measuring semantic disagreement across multiple generated responses using Natural Language Inference (NLI) models, these methods depend on separately loaded NLI classifiers to estimate pairwise semantic similarity, introducing significant computational overhead. We propose Embedding Graph Kernel Entropy (EGKE), a variant of Kernel Language Entropy (KLE) that replaces its NLI-derived semantic graph with a Radial Basis Function (RBF) kernel applied to dense sentence embeddings generated by a BGE embedding model, while retaining KLE’s graph Laplacian, heat kernel, and von Neumann entropy formulation unchanged. By computing similarities directly in a continuous embedding space, the proposed method provides a symmetric and computationally efficient alternative to NLI-based semantic comparison. We evaluate our approach across 30 experimental setups (covering multiple question-answer datasets and open-weight LLM families) using both AUROC and AUARC, and demonstrate that this RBF-based method achieves competitive hallucination detection performance against state-of-the-art methods.
[LG-323] Refined Grid Sampling for Efficient Implicit Neural Representations of Large-Scale Scientific Data
链接: https://arxiv.org/abs/2610.14205
作者: Guang Zhao,Xihaier Luo,Sean McCue,Luke P. Van Roekel,Seungjun Lee,Yihui Ren,Shinjae Yoo,Wei Xu
类目: Machine Learning (cs.LG)
*备注: 7 pages
Abstract:Implicit neural representations (INRs) offer a powerful framework for modeling high-dimensional physical fields, but their training remains inefficient because evaluating dense coordinate grids is computationally expensive. Existing coordinate-pruning strategies rely on local error functions that can oversample correlated points and incur substantial overhead from full-grid error evaluation. We introduce Refined Grid Sampling (RGS), which combines stratified grid sampling, error-based refinement, and dynamic grid-resolution adjustment. This design reduces spatial redundancy without evaluating errors over the full grid. On high-resolution Navier–Stokes fields, RGS achieves higher reconstruction accuracy within a fixed wall-clock budget and reaches the accuracy of full-grid training up to 6\times faster than the evaluated baselines.
[LG-324] ULPack: Reference-Conditioned Lossless Compression of Checkpoint Trajectories
链接: https://arxiv.org/abs/2610.14184
作者: Aniket Srinivasan Ashok(University of Waterloo)
类目: Machine Learning (cs.LG)
*备注:
Abstract:Studying model training requires checkpoint trajectories, whose storage cost limits temporal resolution. Although adjacent checkpoints are numerically close, existing lossless codecs either operate on XOR bit patterns or ignore trajectory-specific conditional structure. We show that a reference weight’s exponent predicts the bit length of its displacement in ordered floating-point space without side information. We introduce \textbfULPack, a lossless codec which exploits this dependence by conditioning entropy models of ordered-integer residuals on the reference exponent. Across bfloat16, float16, and float32 trajectories, ULPack reduces storage by up to 27% against ZipNN, 21% against zstd-19, and 11.5% against FM-Delta. Across two dense trajectories, compression improves as updates shrink and reference conditioning strengthens, while near uniform low-order bits set the remaining compression floor and earlier checkpoints add little information.
[LG-325] From Prototype to Massing: Learning Graph Correspondence for Scalable Facade Modeling
链接: https://arxiv.org/abs/2610.14098
作者: Yiliang Shao
类目: Machine Learning (cs.LG); Graphics (cs.GR)
*备注: Accepted at ACADIA 2026 (Humanism Recoded, Detroit). Author’s version. 10 pages, 13 figures
Abstract:Facade articulation at building scale is usually resolved manually or through parametric systems that encode design intent as explicit rules. This paper investigates a complementary workflow in which a designer-authored prototype supplies transferable articulation logic. Prototype and target massing are represented as heterogeneous attributed graphs, and a Siamese graph attention network with instance-local triplet supervision embeds their segments in a shared metric space. Many-to-one nearest-neighbor retrieval then routes prototype Patterns to the larger target. On the original 85/15 within-corpus holdout, prototype-to-massing top-1 accuracy reached 96.5%, compared with a reported 76.5% raw-feature baseline. A Rhino demonstrator converts predicted correspondence into reusable facade Patterns, and qualitative applications on four built projects show the workflow across distinct massing organizations. The contribution is not automated facade generation, but an authoring model in which resolved geometry serves as a computational specification for project-scale propagation.
[LG-326] A Physics-Informed Hybrid Forward-Inverse Machine Learning Framework for High-Dimensional Sparse Recipe Reconstruction
链接: https://arxiv.org/abs/2610.14097
作者: Md. Muqtadir Fuad,Md. Istiak Adnan,Ridwan Al Aziz
类目: Machine Learning (cs.LG)
*备注:
Abstract:High-dimensional sparse recipe formulation in process industries involves nonlinear association of various covariates, bounded by sparsity constraints and multi-criteria tolerances. Traditional physics-based classical methods (such as Kubelka-Munk) fail to generate sparse candidate recipes properly due to their simplified assumptions and industrial process variations. This study proposes a hybrid physics-informed uncertainty-aware framework that is structured into three modules: (a) Module A, a multi-target forward quality model; (b) Module B, a sparse inverse reconstructor; and © Module C, a feasibility-constrained ranking layer. Then it has been validated in an industrial textile dyeing context with recipe-grouped cross-validation under multi-illuminant quality assessment comprising process and production information, target reflectance, K/S values, and spectral descriptors. Histogram Gradient Boosting (HGB) with spectral-only features ( R_\lambda , K/S) and Gradient Boosting Regressor (GBDT) with recipe-process features outperforms other configurations with out-of-Fold MAE of 0.1655 and 0.1523 in the forward quality model. For inverse reconstruction, a component-wise median blend achieved a recipe-vector MAE of 0.0295. Exploratory conditional treatment effect (CATE) suggests an observational association that alkali adjustment reduces error for red and pale shades (CATE = -0.196). This framework provides a human-in-the-loop decision-support tool that minimizes costly laboratory trials across industrial formulation domains.
[LG-327] Beyond a Better Score: Long-Horizon Agent ic ML Development and Evaluation Protocol for Physics Time Series
链接: https://arxiv.org/abs/2610.14024
作者: Yue Ma,Lucas Venetoulias,Akbota Assan,Aobo Li
类目: Machine Learning (cs.LG); High Energy Physics - Experiment (hep-ex)
*备注: 43 pages, 10 figures, 15 tables
Abstract:Physics time-series data are central to fundamental discoveries involving dark matter, neutrinos, and gravitational waves. Machine learning has shown strong potential to recover weak scientific signals buried in complex noise. Recent LLM-agent systems further automate ML development, but existing approaches are often driven primarily by one or a few numerical scores. In pursuit of a better score, agents may exploit shortcuts, trigger mode collapse, or game the evaluation procedure, producing models that achieve high scores but are scientifically invalid. We introduce SIDERIUS, an LLM agent system designed to move beyond a better score toward scientifically valid, long-horizon ML development. SIDERIUS combines contract-defined composable capabilities for flexible orchestration and human-in-the-loop, a multi-layer scientific evaluation protocol that distinguishes valid models from unhealthy or gaming models, and resource-aware exploration for computationally expensive scientific ML. In our TIDMAD benchmark, SIDERIUS with a GPT-based orchestrator achieves the highest Valid-model proposal rate of 63.8% and the best valid denoising score of 2.744, compared with 1.6% and -0.647, respectively, for the same GPT model using the general-purpose Codex coding agent. We further benchmark SIDERIUS on three additional physics time-series datasets, demonstrating its reuse and performance across different tasks. Together, these results suggest that progress in agentic scientific ML should be measured beyond a better score, by how an agent explores and by whether the models it produces can be trusted as science.
[LG-328] Historical Anchors for Scientific Forecasting on Human-Aware Knowledge Graphs
链接: https://arxiv.org/abs/2610.14023
作者: Siu Tin Wu,Ryan C. Barron,Phoebe Upton Sanderbeck,Kim Ø. Rasmussen,George Tompkins,Maksim E. Eren
类目: Machine Learning (cs.LG)
*备注:
Abstract:Predicting which scientific connections will form next can point researchers toward the most promising leads. One successful approach learns representations by sampling paths through publication networks, represented by a hypergraph, that connect scientific entities, concepts, and researchers, approximating how knowledge and expertise move through the literature. Sampled paths are then used to train an embedding model to forecast scientific connections such as a property of a material. Performance, however, depends strongly on how many paths are sampled: too little sampling yields incomplete representations, while further sampling can eventually weaken prediction quality. The best sampling budget is therefore often identifiable only after future discoveries are known. We introduce Query–Anchor Agreement (QAA), a diagnostic requiring no future labels or separately trained model. QAA can be used to select a sampling budget, and refine or re-rank the resulting forecast to improve forecasting scientific connections. QAA compares the candidates a model predicts for a target with those most similar to the target’s known historical examples, and identifies when these two views converge. Across thermoelectric-material and COVID-19 drug-repurposing benchmarks, repeated random-seed runs, and three sampling policies, QAA-selected budgets fall on average 1.86 percentage points below hindsight-optimal selection for broad candidate recovery. Further, promoting candidates supported by both views, under a capped fusion that bounds the anchor contribution, improves broad recovery across all six benchmark-policy settings, which define the types of network walks used for sampling.
[LG-329] DShuttleNet: A Badminton Doubles Dataset for Evaluating Team Coordination
链接: https://arxiv.org/abs/2610.14016
作者: Guan-Yi Jhang,Chiencheng Chiang,Wei-Kai Chen,Hui-Ching Yu,Hsu-Chao Lai,Hwai-Jung Hsu,Wang Chih-Chuan,Jhy-Cheng Huang,Jiun-Long Huang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Most existing sports datasets are individual-centric or ball-centric, focusing on individual player actions while ignoring interactions among teammates. In badminton doubles, tactical success is governed by rapid team formation and coordination, yet standard resources for modeling these player interactions are scarce. To bridge this gap, we present DShuttleNet, a comprehensive dataset for badminton doubles curated from elite BWF World Tour tournaments. Beyond trajectory data, DShuttleNet uniquely incorporates expert-labeled “Loss Reasons,” transforming the dataset into a diagnostic tool for causal tactical evaluation. We formally define two geometry-driven representations, Teammate-Relative Angle (TRA) and Vertical Shot Angle (VSA), to characterize coordination and shot aggressiveness. To quantify interpersonal synergy, we introduce novel metrics, Court Coverage Efficacy Index (CCEI) and Threat Load Asymmetry Index (TLAI), which evaluate the often-overlooked non-hitting partner’s role and defensive coordination. Experimental results demonstrate that conventional models suffer from underfitting when failing to incorporate these coordination-centric features, highlighting the critical importance and potential impact of releasing DShuttleNet. Dataset is available at: this https URL.
[LG-330] Perfect Reconstruction Cannot Show Whether Two Codebooks on a VAEs Latent Space Agree
链接: https://arxiv.org/abs/2610.14012
作者: Yusuke Hayashi
类目: Machine Learning (cs.LG); Information Theory (cs.IT)
*备注: 28 pages, 4 figures, 4 tables. Supplementary material (per-run results, precommit documents and verification scripts) is provided as ancillary files
Abstract:Many systems read the latent space of a variational autoencoder (VAE) as discrete codes, and rely on each latent being read back as the code it was written as. Shannon (1948) took such a shared codebook as given; inside a VAE nothing enforces it. We prove that no item of a standard training report can reveal whether the written and read codes agree on encoded data. We construct two VAEs that reconstruct every input perfectly, infer exactly, and match on every per-example ELBO and every code histogram, yet agreement is one in the first and zero in the second. We then show that the ELBO’s own terms bound how far a measured agreement can move: the variational gap bounds the shift to exact inference, and the rate bounds the shift to generation from the prior. For Gaussian posteriors, a mean-field encoder pays an exact price in the gap that depends on its choice of axes. On probabilistic PCA the gap certifies a trained encoder in all 75 settings we test, at agreement between 0.76 and 0.97. On 240 trained VAEs, a floor set by the rate and the dataset size rules out rate-side certification in advance for 66 models, and none of them certifies. The results yield a simple protocol, Certify-or-Measure, for any pipeline that reads a VAE as codes.
[LG-331] SIRNET: Causal Adaptation of Pretrained Embeddings for Semiparametric Treatment-Effect Estimation Under Surrogacy
链接: https://arxiv.org/abs/2610.14011
作者: Lei Shi,Qiuran Lyu
类目: Machine Learning (cs.LG); Methodology (stat.ME)
*备注:
Abstract:Modern experiments increasingly produce rich post-treatment objects—such as text, images, and video—that may provide early surrogate signals for outcomes that are delayed, expensive, or observed only for a subset of participants. A pretrained encoder can turn such an object into a strong generic embedding, but the embedding is not trained for the estimation tasks required by a causal estimator and may emphasize information irrelevant to the causal target. We propose SIRNET, a Surrogate-Informed Representation adaptation framework for neural-NETwork-based causal estimation in the presence of unstructured post-treatment surrogates. Starting from pretrained surrogate embeddings, SIRNET learns a shared adapter jointly for outcome prediction, treatment propensity, and arm-specific surrogate marginalization, beyond the generic pretraining objective. The resulting covariate and surrogate representations provide a common interface for these nuisance tasks. The final treatment-effect estimate and confidence interval are obtained from cross-fitted orthogonal scores that attain the semiparametric efficiency bound under the stated nuisance-rate conditions. We also validate SIRNET’s performance in synthetic and semisynthetic experiments, as well as the Anti-Amyloid in Asymptomatic Alzheimer’s Disease (A4) trial, using frozen pretrained brain-MRI embeddings as surrogates for a long-term outcome.
[LG-332] Generalized PINN-Based Inverse Estimation of Spatially Varying Heat Transfer Coefficients with Transfer Learning
链接: https://arxiv.org/abs/2610.14008
作者: Pradipan Maitra,Jariful Hassan,Arijit Hazra
类目: Machine Learning (cs.LG); Computational Engineering, Finance, and Science (cs.CE); Numerical Analysis (math.NA); Optimization and Control (math.OC); Fluid Dynamics (physics.flu-dyn)
*备注: 35 pages
Abstract:This study investigates generalized boundary representations and transfer learning for physics-informed neural-network (PINN) estimation of spatially varying convective heat transfer coefficients (HTCs) in jet impingement cooling. Piecewise-linear (PWL) and neural-network boundary estimators are inferred from transient interior-temperature measurements and evaluated against CFD-CHT reference HTC profiles at Reynolds numbers of 5000, 10000 and 15000. A discrete Adjoint-FEM formulation with gradient regularization based on the H^1 -seminorm provides an independent inverse-method comparison. Transfer learning is investigated under two source-target conditions: transfer between noise levels at the same Reynolds number, and transfer from a controlled Gaussian-profile problem to CFD-CHT-derived target problems at different Reynolds numbers. For the representative Re = 5000 target with 5% measurement noise, quadratic, PWL, and neural-network PINNs achieve relative HTC-profile errors of 5.08, 5.31 and 5.19%, respectively, while generalized representations reduce the error for the Re = 15000 target from 16.76% for the quadratic PINN to 14.41% and 14.46%. Transfer learning reduces target-training time by 57.86-81.71% for fixed-Reynolds-number transfer and by 50.59-84.94% for Gaussian-to-CHT transfer. Its effect on reconstruction accuracy is target dependent: fixed-profile PWL transfer remains within 0.30 percentage points of scratch training, whereas Gaussian-to-CHT transfer produces differences of 0.10-1.52 percentage points across targets and representations. These results demonstrate that generalized PINN representations can accommodate spatially varying HTC profiles without a prescribed quadratic form, while transfer learning can substantially reduce target-training cost for repeated inverse estimation, with the resulting reconstruction accuracy depending on source-target compatibility.
[LG-333] DoSA: Dots Sparse Attention
链接: https://arxiv.org/abs/2610.13990
作者: Fanxu Meng,Taosong Fang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Sparse attention visits few keys, yet finding them can still require scoring every key with every indexer head. We introduce DoSA (Dots Sparse Attention), a lossless method for making this computation conditional while preserving exact token selection. Its central observation is that Top-K selection requires sufficient evidence to reject a candidate, not necessarily its full score. A seed set supplies a lower bound on the winning threshold. Nonnegative head projections and joint residual bounds upper-bound positive contributions, while activation-informed aggregation bounds the magnitude subtracted by negative heads. The certificate directs full scoring toward candidates that remain eligible; exact-score filtering further reduces writeback and selection work. We establish the certificate and its exact-selection conditions, and analyze how geometry and selection margins govern its effectiveness. Our GPU implementation accelerates the complete Indexer by up to 2.74 times over SGLang and 1.69 times over LiteTopK across English, Chinese, and code workloads. End-to-end evaluation on GLM-5.2 and DeepSeek-V3.2 achieves up to 1.94 times and 1.50 times time-to-first-token speedups over the two baselines, respectively.
[LG-334] PRIME: Frequency-Aware Unified Pruning and Merging for MoE Compression
链接: https://arxiv.org/abs/2610.13982
作者: Shanfeng Huang,Rongqian Zhang,Zhou Fang,Song Xiao,Hai Du
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Mixture-of-Experts (MoE) models have emerged as an effective paradigm for scaling large language models, yet their massive parameter footprint poses severe memory challenges for deployment. Existing compression methods typically treat expert pruning and merging as disjoint operations and rely on greedy heuristics, often leading to irreversible information loss or representational pollution. In this paper, we propose PRIME, a unified optimization framework for MoE compression via adaptive PRunIng and MErging. PRIME casts the discrete prune-merge assignment step as a global Integer Linear Programming over a frequency-weighted surrogate objective, jointly selecting pruning and merging decisions while accounting for expert usage. We motivate the frequency-aware objective through an error-bound analysis under fixed representative assignments, and show that the formulation favors high-frequency experts as centroids, promoting routing stability. Extensive experiments on Qwen3-30B-A3B and Qwen3.5-122B-A10B demonstrate that PRIME achieves strong accuracy retention across knowledge, coding, math, and long-context tasks, while serving benchmarks of merged models quantify the memory and throughput benefits of expert compression in vLLM.
[LG-335] LACRM: LLM -Guided Mining of Directed Advertiser Competition Relationships in E-Commerce Advertising
链接: https://arxiv.org/abs/2610.13959
作者: Xiang Shi,Yitian Kou,Bing Li,Shengsheng Niu,Chun Gan,Linghui Meng,Jie He,Zhangang Lin,Ching Law
类目: Machine Learning (cs.LG)
*备注:
Abstract:In e-commerce advertising, ad platforms typically fail to model the externalities among advertisers when making optimization decisions: an advertiser’s actual performance is largely shaped by rivals competing for the same users; these rivals typically operate in overlapping product categories and at comparable budget scales. However, capturing this rivalry as an explicit, reusable signal is challenging because it is directed, multi-faceted, and partially observable. Existing paradigms such as interaction graphs, taxonomy rules, and single-embedding models do not fully capture these properties. In this paper, we propose LACRM (LLM-driven Advertiser Competition Relationship Mining), a novel framework that constructs directed competition graphs with continuous edge intensities via a three-stage pipeline. LACRM moves beyond symmetric similarity through LLM-driven role-separated profiling, which enables asymmetric multi-route retrieval to uncover both realized and latent rivals. The framework further incorporates listwise learning-to-rank aligned with temporal auction behavior to accurately model directed competition intensity. In offline retrieval experiments on an industrial benchmark, LACRM outperforms strong directed GNN baselines and ablation variants. Multi-LLM and human studies provide complementary evidence that the recovered latent rivals reflect meaningful market competition rather than merely model-generated artifacts. In production A/B tests, LACRM-guided bidding and incentive allocation increase advertising revenue by 1.73% and 1.14%, respectively. The results demonstrate the practical value of explicit competition modeling in large-scale advertising systems.
[LG-336] MDN-VJEPA: Multimodal Probabilistic Joint-Embedding World Models
链接: https://arxiv.org/abs/2610.13956
作者: Yongchao Huang
类目: Machine Learning (cs.LG)
*备注: 69 pages
Abstract:Joint-Embedding Predictive Architectures (JEPAs) learn world models by predicting future representations rather than reconstructing observations. Variational JEPA (VJEPA) gives this prediction probabilistic semantics and permits flexible conditional predictive families. We propose MDN-VJEPA, a multimodal probabilistic JEPA that instantiates the latent predictor as a mixture density network (MDN), representing future states with a Gaussian mixture whose components encode alternative futures, their probabilities, and within-mode uncertainty. For sequential prediction, we address two additional structural requirements: multi-modal futures should remain temporally coherent across a trajectory, and predictive beliefs should remain consistent when propagated to the same horizon through different temporal path compositions, as motivated by the Chapman–Kolmogorov equation. Together, these considerations define a three-level hierarchy of multimodal prediction, coherent multimodal trajectories, and compositionally consistent multimodal dynamics. Controlled experiments show that MDN-VJEPA closely recovers known branching predictive laws, that retaining the complete multimodal belief rather than its conditional mean can lead to materially different and substantially better planning decisions under branching futures, and that trajectory-level coupling and distributional consistency improve temporal coherence and multi-horizon agreement. These results support a simple principle: a probabilistic JEPA should preserve not only how uncertain the future is, but which distinct futures constitute that uncertainty, how they form coherent trajectories, and how those beliefs compose through time.
[LG-337] BandPC: Learning Residual-Band Preconditioner Combinations for Flexible Conjugate Gradient Solvers
链接: https://arxiv.org/abs/2609.37140
作者: D. M. Li,Shang-Tian Yang,Xin Qiu
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG)
*备注:
Abstract:The conjugate gradient (CG) method is a classical iterative solver for sparse symmetric positive definite (SPD) linear systems, but its convergence strongly depends on the spectral properties of the system matrix. Preconditioning can improve these properties; however, designing preconditioners that generalize across diverse and irregular sparsity patterns remains challenging. We propose BandPC (Band Preconditioner Combinations, where “Band” denotes residual-norm bands), a data-driven framework that leverages graph neural networks (GNNs) to predict multi-stage preconditioning strategies for the flexible conjugate gradient (FCG) method. BandPC partitions the FCG iteration into three residual-norm-based bands and defines a structured search space of 125 candidate combinations over five classical preconditioners. By exploiting FCG’s ability to switch preconditioners across iterations, BandPC learns to map matrix structure directly to a promising preconditioner sequence. To improve training and generalization, we introduce a soft-labeling mechanism that retains near-optimal combinations and normalizes their scores into a label distribution, together with a hierarchical feature representation that captures node-level attributes, edge-level algebraic coupling strengths, and global matrix statistics. Experiments on a hybrid benchmark of synthetic SPD problems and diverse matrices from the SuiteSparse Collection show that BandPC achieves the best iteration count and solution time on 41.3% and 33.1% of test matrices, respectively, even when compared with each matrix’s individually best traditional preconditioner. These results demonstrate that learned residual-band preconditioner scheduling can effectively accelerate flexible CG solvers.
[LG-338] SlideDP: Scaling Host-Resident LLM Fine-Tuning Across Multiple GPUs
链接: https://arxiv.org/abs/2609.34162
作者: Ruijia Yang,Shiyuan Lin,Yulong Ao,Zhiyu Li,Yingli Zhao,Xianduo Li,Yonghua Lin,Zeyi Wen
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注: 14 pages, 15 figures, 4 tables
Abstract:Host-resident layer streaming enables full-parameter LLM fine-tuning beyond GPU memory, but data-parallel ranks compete for shared host resources. Replicated transfers amplify traffic, while strong scaling can expose host work as computation windows shrink. We present SlideDP, a synchronous data-parallel runtime for shared-host multi-GPU systems. It maintains one authoritative host state, decouples communication routes from state layout, and pipelines parameter delivery, gradient aggregation, and CPU updates across ranks and chunks. An analytical step-time model characterizes resource bottlenecks and pipeline exposure; runtime measurements guide communication, chunking, and activation policies under a GPU memory budget. In matched-batch sweeps, SlideDP achieves geometric-mean throughput ratios of 1.46-2.64 \times over SlideFormer, MegaTrain, and ZeRO-Offload. On four H100s, SlideDP approaches GPU-resident FSDP2 throughput for Qwen3-14B at a smaller batch size. With a larger batch, it processes over 1M tokens per step and exceeds FSDP2’s measured peak throughput by 11.2%. Separately, it supports 256K-token sequences for the same model and fine-tunes Qwen2.5-72B on four RTX 4090 GPUs. Project page: this https URL.
[LG-339] Data Free Training of Diffusion Models for Lattice Field Theories
链接: https://arxiv.org/abs/2610.18863
作者: Dennis Grinwald,Ankur Singha,Khaled Kahouli,Karl Jansen,Shinichi Nakajima
类目: High Energy Physics - Lattice (hep-lat); Machine Learning (cs.LG)
*备注:
Abstract:Sampling equilibrium field configurations remains a major computational bottleneck in lattice field theory, motivating the development of generative samplers that can produce independent configurations. Diffusion models are promising for this purpose, but their standard denoising score matching objective requires samples from the target distribution, seemingly reintroducing the original sampling problem. Here, we show that diffusion models can instead be trained from their own generated samples using the target score identity, provided access to the unnormalized probability density. Importance-weighted formulations of this identity yield a consistent target throughout training, while unweighted variants define a self-consistent fixed-point training scheme. We benchmark the different approaches in two-dimensional (\phi^4) theory and identify effective choices that enable efficient training and sampling across the symmetric, critical, and broken phases. Ultimately, our data-free training approach achieves comparable performance to training on HMC reference data.
[LG-340] Joint Neural Posterior Estimation and Model Selection for Chemical Reaction Networks
链接: https://arxiv.org/abs/2610.18825
作者: J. Arturo Esquivel,Jorge A. Campos-Gonzalez-Angulo,Sergio Pablo-García,Alán Aspuru-Guzik,Vianey Leos-Barajas
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Chemical Physics (physics.chem-ph)
*备注: 40 pages total (11 pages main text, 29 pages appendix), 18 figures. Code available at: this https URL
Abstract:Simulation-based Bayesian inference for Chemical Reaction Networks (CRNs) must simultaneously solve two computationally demanding problems: discriminating between overlapping reaction pathways and estimating reaction rate constants. Classical approaches, such as Sequential Monte Carlo ABC, rely on handcrafted summary statistics and scale poorly across large parameter spaces. To overcome this, we introduce a neural inference architecture. Our pipeline utilizes shared 1D-CNNs and a permutation-invariant Set Attention mechanism to learn low-dimensional summary statistics from variable-length, multi-experiment time-series data. These statistics simultaneously condition a model selection network to compute posterior model probabilities and model-specific conditional Invertible Neural Networks (cINNs) that estimate parameter posterior distributions. We demonstrate the practical utility of our approach on a real protodeboronation example, illustrating how it handles structural ambiguity and extracts kinetic parameters from competing catalyst deactivation pathways.
[LG-341] Emergent Loop Geometry in Gauge-Equivariant Graph Neural Networks
链接: https://arxiv.org/abs/2610.18742
作者: Ali Rayat,Gia-Wei Chern
类目: atistical Mechanics (cond-mat.stat-mech); Strongly Correlated Electrons (cond-mat.str-el); Machine Learning (cs.LG); High Energy Physics - Lattice (hep-lat)
*备注: 16 pages, 8 figures
Abstract:Gauge-equivariant graph neural networks generate ordered products of gauge links through local message passing, but the geometry of these internal structures is not captured by the receptive field alone. We investigate the statistical ensemble of closed lattice walks generated by recursive node- and edge-feature operations, independently of network training. Analytical and numerical results reveal a separation between rapid growth of contour length and much slower spatial expansion: deeper compositions generate increasingly long paths through repeated traversal of relatively compact regions. Loop expansions of fermionic observables provide a physical interpretation of this hierarchy, showing how spatial locality remains compatible with high-order dependence on nearby gauge fields. These findings identify complementary roles of network depth in extending spatial reach and enriching the representations constructed within that reach, providing a geometrical perspective on learning lattice gauge observables.
[LG-342] Minimax rates for pointwise CATE estimation: the role of design smoothness
链接: https://arxiv.org/abs/2610.18623
作者: Yuxuan Hou,Zhenghan Li
类目: atistics Theory (math.ST); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:We study the role of design smoothness in the minimax estimation of conditional average treatment effects at a fixed covariate point. For design densities bounded above and away from zero, without further regularity, we derive upper and lower risk bounds with an arbitrarily small gap in the convergence-rate exponent. Under uniform design, we obtain matching upper and lower bounds at interior points in the regime of low nuisance smoothness. These results show that rough covariate designs can lead to slower rates than uniform designs, even under the same smoothness assumptions on the treatment effect and nuisance functions, highlighting the role of the covariate distribution in pointwise causal estimation.
[LG-343] Understanding Circulant Permutation to Extend the CMinHash Estimator NEURIPS2026
链接: https://arxiv.org/abs/2610.18551
作者: Keegan Kang,Avery Hood,Benedict H.W. Wong
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: Accepted to NeurIPS 2026 as poster
Abstract:CMinHash promises a new direction for designing MinHash algorithms with reduced variance and storage space via circulant permutations. However, the original variance analysis is difficult to extend to other MinHash estimators. We present a new approach that gives explicit and easily comparable approximate variance expressions, and use it to analyze the Minner estimator. Numerical experiments show that our variance expression matches the empirical mean square error (MSE) of estimates. Our work thereby demonstrates how to apply circulant permutation to other MinHash estimators and compute their approximate variance.
[LG-344] How Low Can You Go: Fine-Tuning LLM s with Bayesian Optimization (LlM BO)
链接: https://arxiv.org/abs/2610.18504
作者: Aditya Kulkarni,Davide Sartor,Carlos Soto,Nathan Wycoff
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
*备注:
Abstract:Recently, Morris et al. introduced TinyLoRA, a strategy for LLM fine-tuning which introduces arbitrarily few parameters, and demonstrated accuracy improvements using less than 10 parameters rather than the \gtrsim 10,000 typically demanded by LoRA-based methods. However, that study continued to use gradient-based local search methods of the kind deployed for training multi-billion parameter LLMs in the first place. This article is predicated on the observation that, in the low dimensional setting, global optimization methods become tractable, unlocking Large language Model Bayesian Optimization (LlMBO). LlMBO is sample-efficient, gradient-free, memory-light, parallelizes across devices without synchronization, and, unlike policy-gradient methods, can optimize greedy-decoding accuracy directly. We describe numerical experiments comparing LlMBO against GRPO on fine-tuning tasks for a variety of models on a math benchmark suite which show that LlMBO can be a suitable alternative in this regime, matching or exceeding GRPO’s fine tuning gains, while using a fraction of the compute and memory.
[LG-345] Steering the Fidelity in State Preparation on Noisy Bosonic Systems with Risk-Seeking Optimization
链接: https://arxiv.org/abs/2610.18495
作者: Riccardo Piantoni,Riccardo Pellini,Maurizio Ferrari Dacrema,Paolo Cremonesi
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG)
*备注: 12 pages, 1 table, accepted at International Workshop on AI for Quantum and Quantum for AI (AIQxQIA 2026), co-located with AIxIA 2026, Perugia, Italy
Abstract:In order to execute quantum algorithms on bosonic quantum computers, it is necessary to encode their inputs into a desired target quantum state, with a process called state preparation. In real settings, state preparation methods operate in stochastic scenarios, where the prepared state is represented as a mixed state, that is a probabilistic ensemble of pure states. Therefore, existing state preparation methods optimize a variational ansatz by averaging over all possible sequences of photon loss events contained in the prepared ensemble, yielding an output state which approximates the desired target with a certain average fidelity. Such an approach, however, does not take into account directly the distribution of fidelities across states in the ensemble. In this work, we aim to maximize the fidelity of only a subset of pure states sampled from the prepared ensemble, at the cost of reducing the average fidelity overall. This trade-off is relevant when the goal is to obtain high-fidelity states with a certain probability, rather than maximizing the average fidelity of the prepared ensemble. We borrow and adapt objective functions from risk-sensitive optimization, a framework that optimizes specific characteristics of a distribution rather than only its expectation, and that has been successfully applied in Reinforcement Learning. In the considered experimental settings, our results show that the infidelities of some pure states in the prepared ensemble can be reduced by up to four orders of magnitude when using risk-seeking objectives, compared to those achieved using the standard optimization of the average fidelity. Furthermore, we show that it is possible to control the trade-off between the probability of obtaining high-fidelity states and the lowest infidelity threshold reached, providing the flexibility to select the operating point that best suits a certain application.
[LG-346] Lifting: Quantum State Discrimination Framework for Mid-Circuit Measurements
链接: https://arxiv.org/abs/2610.18482
作者: Chi Zhang,Neel Vora,Yilun Xu,Gang Huang,Fan Chen,Lei Jiang
类目: Quantum Physics (quant-ph); Emerging Technologies (cs.ET); Machine Learning (cs.LG)
*备注:
Abstract:Quantum computing promises new capabilities beyond classical computation. Mid-circuit measurements (MCMs) enable error correction and outcome-dependent control, but their reliability is limited by noise on superconducting hardware. These operations depend on accurate quantum state discrimination (QSD), which assigns measured signals to state labels. Neural network-based (NN-based) QSD improves this assignment, yet its error remains bounded below by the Bayes risk of the available readout distributions. A complementary challenge is state deterioration as successive measurements accumulate and circuit execution advances, making static calibration less representative of the workload. We present Lifting, a QSD framework that addresses both challenges through repetition encoding, neural decoding, and in situ calibration. Encoding supplies joint in-phase and quadrature (I-Q) readout signals from the target and auxiliary qubits, while contextual calibration trains the decoder for the conditions encountered during execution. Our theoretical analysis shows that retaining additional channels from a fixed joint distribution cannot increase Bayes risk and establishes sufficient conditions for contraction of an overlap-based upper bound. Experiments across three IBM processors evaluate the framework on repeated stabilizer measurement, parity-check, and teleportation workloads.
[LG-347] Convergence-Rate-Guided Algorithm Design of Expected Improvement for Continuous Simulation Optimization
链接: https://arxiv.org/abs/2610.18434
作者: Songhao Wang,Jianglin Xia,Haowei Wang,Szu Hui Ng
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Expected improvement (EI) based on stochastic kriging (SK) surrogate is widely used in continuous simulation optimization, but unknown heteroscedastic noise on simulation output makes three basic design choices consequential: the incumbent used to define improvement, the replication level at each sampled point, and the terminal recommendation. Existing methods typically either assume the noise levels are known or choose these components separately through heuristic rules or empirical calibration. We instead derive a high-probability upper bound on the simple regret and systematically choose the three components within general rule classes to tighten this bound. We first establish a finite-sample confidence bound for the SK predictor with unknown and estimated noise levels as a foundation for the three design choices. We then analyze general posterior-quantile families for both the incumbent and the terminal recommendation. For the incumbent, minimizing the regret bound selects a lower quantile coefficient that matches the coefficient in the SK predictor confidence bound. For the terminal recommendation, minimizing the same bound instead selects the sampled point with the best upper posterior quantile under the same coefficient. These opposite quantile choices reflect the different roles of search and implementation in simulation optimization. Intermediate simulations are exploratory with an optimistic incumbent, while the terminal recommendation is implemented in the real system and is therefore relatively conservative about the remaining posterior uncertainty. Under mild conditions satisfied by common kernels, minimizing a tractable kernel-specific bound yields \Theta(\log T) replications. To our knowledge, this is the first simple-regret guarantee for a continuous-domain EI-type simulation optimization method that learns unknown heteroscedastic variances through replications.
[LG-348] Overcoming Catastrophic Forgetting: Continual Learning for Radar Waveform Classification
链接: https://arxiv.org/abs/2610.18404
作者: Mohamed Rabie,Chinthana J. Panagamuwa,Konstantinos G. Kyriakopoulos
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注: 15 pages, 5 figures, 16 tables. Submitted to IEEE Transactions on Aerospace and Electronic Systems
Abstract:Automated radar waveform recognition classifies the waveform and modulation type of a received signal. As new waveform variants appear over time, a deployed classifier must learn them without retraining from scratch. Catastrophic Forgetting occurs when learning new waveforms overwrites prior knowledge. Continual Learning (CL) methods tackle this challenge, but prior radar studies only evaluate a few methods on narrow benchmarks. Practitioners lack evidence on which method to deploy or which mechanisms are effective. This work presents the first large-scale benchmark of continual learning for radar waveform recognition. Eighteen CL methods are evaluated on a sequential ten-task problem, drawn from four public datasets spanning 54 waveforms, under task- and class-incremental (TIL, CIL) settings and single- and multi-epoch regimes. Methods that retain data from old tasks lead both TIL and CIL settings. Plain replay alone closes 77-88% of the gap between the lower and upper practical limits under TIL, but only 46-55% under CIL. The internal mechanisms of the leading methods are ablated and statistically tested to identify effective mechanisms. Each ablation is paired against its full method over n=6 training seeds, topped up to n=12 when inconclusive, and p-values are Holm-Bonferroni corrected within each method’s family. Ablation results show that several mechanisms confer smaller but statistically significant gains. Combining these mechanisms improves the leading F1 score by 3.4-5.5% and 2.7-4.9% under TIL and CIL, respectively, and the combined models exceed every benchmarked method in both settings for all SNRs. These results give researchers of radar waveform classification systems a mechanism-level basis for design.
[LG-349] Beyond RMSE and Granger Gain: Multi-Horizon Forecast-Error Entropy for Nonlinear Predictive Coupling
链接: https://arxiv.org/abs/2610.18390
作者: Andrei Velichko
类目: Chaotic Dynamics (nlin.CD); Machine Learning (cs.LG)
*备注: 35 pages, 13 figures, 12 tables. Software and benchmark package: Zenodo, this https URL
Abstract:We investigate whether the past of an auxiliary time series provides predictive information about a target beyond that contained in the target’s own history. We compare matched out-of-sample target-only and target-plus-source forecasts and introduce the forecast-error distribution entropy (FEDE) predictive-causality framework, where predictive causality is understood in the Granger sense: auxiliary history is informative when it improves prediction beyond the target’s own past. FEDE quantifies how the full signed forecast-error distribution changes across prediction horizons. Controlled nonlinear benchmarks hide source-target dependence behind nonlinear transformations not supplied to the predictors. LightGBM recovers these hidden channels, whereas Ridge does not. Across four equal-power nonlinearities and four noise levels, the mean response-sensitivity statistic for FEDE remains 29.00, 27.47, 23.22, and 28.67, compared with 14.09, 10.68, 12.67, and 12.71 for mean log-RMSE gain. Directional and source controls separate the true source-to-target channel from the reverse direction and from independent periodic, quasiperiodic, chaotic, and autoregressive auxiliaries. Chronological Granger-style comparisons show strong model-form dependence: a degree-2 polynomial model is exceptionally sensitive to the matched quadratic channel but becomes weak or oppositely aligned for other nonlinearities, while basic convergent cross mapping is weak or directionally inconsistent in this stochastic, partly non-invertible setting. FEDE remains positively aligned across all 108 tested combinations of window length, history length, and maximum forecast horizon. Together, these results establish FEDE as a sensitive and robust horizon-resolved tool that extracts predictive information beyond conventional error-magnitude measures when the nonlinear interaction form is unknown. Code available in Zenodo.
[LG-350] On the Optimality of Random Features under General Conditions: Misspecified Cases and Robust Learning
链接: https://arxiv.org/abs/2610.18333
作者: Caixing Wang
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 61 pages, 10 figures, 2 tables
Abstract:Random features provide a standard approach to scaling kernel methods to large datasets by approximating the kernel with a finite-dimensional feature map. Existing works of random feature methods primarily focus on well-specified models and light-tailed noise conditions, leaving a gap in understanding their performance under more general conditions, such as misspecified models and heavy-tailed noise. In this paper, we provide a comprehensive and improved analysis of kernel ridge regression with random features (KRR-RF) under general conditions. In the light-tailed noise setting, we improve upon the worst-case uniform bounds of the feature map used in the previous works and lead to the minimax optimal learning rates with 2r+\gamma\alpha_0 , where (r,\gamma,\alpha_0) denote the regularity of the target function, the effective dimension of the kernel, and the embedding index of the RKHS, respectively. In the heavy-tailed noise setting, we leverage a recently developed Hilbert-valued Fuk-Nagaev inequality and extend the same optimal results under finite weak moment conditions. The additional cost to achieve robustness is a mildly stronger explicit sample-size requirement than in the light-tailed case. For both settings, our results recover the conditions random features budgets (up to log factors) required in the previous works. In the most benign case when \alpha_0=\gamma , KRR-RF are minimax optimal for all r \in (0,1] . To the best of our knowledge, this is the first work to establish optimal learning rates for KRR-RF under weak moment conditions in the misspecified setting. Numerical experiments and a real data analysis validate the derived learning rates and demonstrate their practical relevance in the presence of mis-specification and heavy-tailed noise.
[LG-351] Hierarchical Time Series Forecasting with Operator-Valued Kernels
链接: https://arxiv.org/abs/2610.18196
作者: Junjie Yang(LPSM (UMR_8001), SU),Adeline Fermanian,Claire Boyer(LMO, IUF),Gérard Biau(LPSM (UMR_8001), SU, IUF)
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Hierarchical time series forecasting requires generating accurate predictions that are coherent across multiple levels of aggregation. Existing methods either rely on two-stage pipelines with post-hoc reconciliation or on end-to-end models that often lack statistical guarantees. We introduce a unified framework that formulates hierarchical forecasting as a multivariate regression problem in a vector-valued reproducing kernel Hilbert space. By designing structured operator-valued kernels, our approach enforces coherence directly within the hypothesis space, ensuring that all predictions satisfy hierarchical constraints by construction. We further propose tailored kernel constructions for hierarchical data and derive excess risk bounds that characterize generalization performance. Empirical results on synthetic and real-world datasets demonstrate the relevance of the proposed method.
[LG-352] Link Both Views: Cross-Image Routing Attention for Dual-Endpoint V2I Beam-Pair Power Prediction
链接: https://arxiv.org/abs/2610.18195
作者: Shiliang Lu,Lu Bai,Ziwei Huang,Xiang Cheng
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注:
Abstract:For sensing-aided communications, single-endpoint sensing offers only a partial view of propagation, motivating link-centric sensing across both endpoints. We propose cross-image routing attention (CIRA) for sparse association and mutual enhancement of propagation-relevant features from dual-endpoint RGB images. Building on CIRA, CiFormer predicts millimeter-wave (mmWave) beam-pair power matrices for vehicle-to-infrastructure (V2I) communications. Experiments show that CiFormer reduces mean absolute error and improves top-1 beam-pair selection accuracy by 3.15 percentage points over the strongest baseline. Visualizations and ablation studies support dual-endpoint sensing and selective cross-view interaction in capturing propagation-relevant dependencies, suggesting broader applications in wireless tasks beyond beam prediction.
[LG-353] Distance Prediction for Quantum Codes
链接: https://arxiv.org/abs/2610.18130
作者: Nadav Schneider,Eliya Nachmani
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG)
*备注:
Abstract:The distance of a quantum code is a fundamental metric that dictates its error-correction capability. While quantum low-density parity-check (qLDPC) families such as Bivariate Bicycle (BB) codes offer a promising pathway to low-overhead fault-tolerant quantum computing, computing their distance is challenging. Current exact solvers and even search heuristics that provide a distance upper bound suffer from severe computational bottlenecks, restricting the discovery of optimal codes. In this paper, we construct a dataset of BB codes with corresponding distance upper bounds obtained through high-budget heuristic searches. We introduce BB-Net, an attention-based neural network architecture for predicting the distance of BB codes. It achieves 96.39% accuracy on the reference distances in the test set, compared with 94.86% for state-of-the-art heuristics with a budget of up to 10^4 times more computational runtime. On a subset with certified exact distances, BB-Net achieves 98.94%. By relieving the distance-calculation bottleneck, BB-Net enables faster screening of large code spaces, particularly in ML-driven search pipelines, and thereby can accelerate the discovery of novel quantum error-correcting codes, an important capability for advancing scalable fault-tolerant quantum computing. Our code is publicly available: this https URL
[LG-354] Reliable Scale-Free Bandits via Empirical Likelihood
链接: https://arxiv.org/abs/2610.18055
作者: Pratham Gupta,Alhad Sethi,Shubhada Agrawal
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:We study stochastic bandits without knowledge of reward scales or tail parameters. Almost-sure regret guarantees describe eventual learning but do not quantify expected performance or the risk of large regret in a finite run. We introduce \textttEL-F , an anytime empirical-likelihood (EL) extension of the g -Forcing policy of Cowan and Katehakis (2020), and characterize its performance at three levels. First, for integrable rewards and non-decreasing, diverging exploration schedules g with vanishing increments, \textttEL-F preserves the g -Forcing asymptotics for pseudo-regret R(T) : R(T)/g(T)\to\sum_i:\Delta_i0\Delta_i almost surely as T\to\infty , where \Delta_i is the suboptimality gap of arm i . Second, on every sub-exponential instance, for non-decreasing g(t)=\omega(\log t) satisfying a mild feasibility condition, expected pseudo-regret is at most g(T)\sum_i:\Delta_i0\Delta_i+C_\underlineP , where C_\underlineP is an instance-dependent constant independent of T . Moreover, we also establish finite-time tail bounds with super-polynomial decay in regret beyond the forced-exploration budget. For slowly varying g , we determine the exact exponent of linear regret deviations on two-armed Gaussian instances with common variance and establish a matching lower bound among policies satisfying the same expected-regret budget on every such Gaussian instance. These quantitative guarantees also cover the greedy variant obtained at zero EL radius. A fixed positive radius introduces local optimism while preserving the leading asymptotic regret. We establish coupled pathwise dominance over the greedy variant in two-armed bandits with a deterministic suboptimal arm. Experiments demonstrate reductions in prolonged suboptimal allocation compared with g -Forcing on challenging instances.
[LG-355] Weighting-Based Strategies for Classification: Analysis and Local Peak Dynamics with CART Trees
链接: https://arxiv.org/abs/2610.18037
作者: Abdoulaye Sakho(LPSM (UMR_8001)),Emmanuel Malherbe,Erwan Scornet(LPSM (UMR_8001))
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Tabular learning tasks often suffer from class imbalance, which is often addressed via weighting-based strategies. In this work, we compare the naive approach (unweighted) with the weighting-based rebalancing strategies. Somewhat surprisingly, we demonstrate that, asymptotically, as the predictor complexity increases with sample size, weighting strategies yield predictive performance identical to unweighted approaches across all evaluation metrics, up to a shift in the final threshold. To go beyond, we theoretically study the estimation of a linear regression function through weighted and unweighted CART trees with limited depths. In this setting, we prove that the points in the PR space for both strategies are distributed along the same curve, thus illustrating the similar nonasymptotic performance of the two strategies. Then, we numerically identify more complex synthetic settings where performance gaps, evaluated via fixed classifier complexity, favor weighting strategies. Thus, to guide practitioners, we introduce an indicator that aims at signaling whether a data set may benefit from weighting.
[LG-356] All-electron machine-learned orbital-free density functional for periodic CHNO systems
链接: https://arxiv.org/abs/2610.17975
作者: Marc K. Ickler,Andreas Hermann,Fred A. Hamprecht
类目: Chemical Physics (physics.chem-ph); Materials Science (cond-mat.mtrl-sci); Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
*备注: 37 pages, 7 figures, 3 tables (including Supplementary Information). Supplementary Data 1 will be available with the journal version
Abstract:We introduce STRUCTURES-P, a machine-learned all-electron orbital-free density functional for periodic CHNO systems. Electrostatics are evaluated analytically in an atom-centered Gaussian basis, while an equivariant network learns the kinetic and exchange-correlation energy from energies and gradients of non-ground-state densities. Minimizing the functional from a superposition of atomic densities recovers the Kohn-Sham ground state: on held-out crystals, the density error falls from 6.7% to 0.4% of the electron count and the energy error reaches 0.25 mHa per atom, half that of a same-architecture interatomic potential. A single functional can be trained for both isolated molecules and periodic systems, and the model generalizes to larger unit cells. It largely preserves relative-energy rankings and reproduces convex-hull vertices on the tested H-N composition slice. Together, these results demonstrate the viability of machine-learned all-electron orbital-free DFT for variational density optimization and energy evaluation in fixed-geometry solids.
[LG-357] Learning Distributions from Samples and Qualities
链接: https://arxiv.org/abs/2610.17956
作者: Marc Jourdan,Nicolas Flammarion
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 50 pages, 9 figures, 1 table
Abstract:We study parameter estimation when each sample from a continuous parametric distribution is accompanied by a binary quality label governed by the same unknown parameter. By imposing hard feasibility constraints, deterministic labels can accelerate estimation from the classical n^-1/2 rate of smooth quality-aware M-estimators to n^-1 . Combining PAC guarantees with local error-to-parameter conversions, we establish this rate for both identifiable and finitely ambiguous quality classifiers, with likelihood information resolving the latter ambiguity. When the quality model is known only up to an additive constant, we construct threshold-free pairwise preferences from oppositely labeled samples and retain the accelerated rate despite the unknown threshold. We further show that pointwise qualities and pairwise preferences are incomparable: either can be strictly more informative than the other. Our assumptions hold for Gaussian, Laplace, and exponential families under thresholded log-likelihood rewards. Finally, a minimax lower bound establishes the optimal n^-1 dependence on sample size, up to logarithmic and dimension-dependent factors, while experiments confirm the benefits of constraint-based estimation.
[LG-358] Representation Geometry and Inductive Biases in Quantum Classifiers across Classical Visual Benchmarks
链接: https://arxiv.org/abs/2610.17722
作者: Yu Yamaoka,Weng Ian Chan,Kosuke Mitarai
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG)
*备注:
Abstract:Quantum Machine Learning (QML) has a theoretical advantage over classical algorithms for specific synthetic data that is computationally hard. However, it remains unclear whether such theoretical advantages play a meaningful role for realistic classical data, such as natural images. Quantum structural properties induced by unitary operations or entanglement introduce inductive biases that differ from those of classical models, potentially leading to distinct learning behavior on real-world data. In a fully quantum learning pipeline, classical real-world data are first encoded into quantum states via a quantum feature mapping, analogous to a classical feature extractor, and a quantum classifier then classifies the resulting states. Quantum feature mappings generate representations that are difficult to realize classically, while also inducing representation-level biases that do not necessarily match the structure of realistic data. To enable a fair comparison at the classifier level, we focus on the classifier and use classical feature extractors that already provide powerful representations for natural images. By controlling the data scale and supervision strength, we analyze how the unitary structure of quantum classifiers affects generalization, and show that their effectiveness depends strongly on the geometry of the input representations. These results indicate that quantum classifiers benefit from classical features that are aligned with quantum feature spaces. Our analysis shows that kernel alignment identifies representations compatible with quantum feature spaces, providing insights into translating classical models into quantum algorithms.
[LG-359] xtGrain watermarking is harder to detect: can sum-tests still do it optimally?
链接: https://arxiv.org/abs/2610.17655
作者: Swapnaneel Bhattacharyya,Soham Bonnerjee
类目: Machine Learning (stat.ML); Information Theory (cs.IT); Machine Learning (cs.LG); Statistics Theory (math.ST); Methodology (stat.ME)
*备注: Working draft; comments are welcome
Abstract:Watermarking embeds an imperceptible signal in text from large language models that is statistically detectable with a secret key. The recent textGrain watermark (Li et. al. (2026b)) calibrates this signal against the model’s sampling entropy through an explicit entropy budget, combining the stochasticity of Red-Green and the unbiasedness of Gumbel-max watermarks, and its accompanying report proposes a detector that sums pivotal statistics. We study when, and how efficiently, textGrain can be detected after human edits, modeled as a sparse mixture in which a fraction \varepsilon_n\asymp n^-p of positions retain the watermark and token uncertainty decays as n^-q . First, entropy calibration comes at a cost: the minimax detection boundary of textGrain is p+q=1/2 , so its detectable region is strictly smaller than that of Gumbel-max and maximal-coupling watermarks, whose boundary is 2p+q=1 . This cost stems from textGrain’s construction rather than from entropy calibration itself: an unbiased scheme with the same entropy loss remains detectable where textGrain is not. Second, perhaps surprisingly, this loss makes the sum detector optimal at the boundary when the number of transport columns is fixed, unlike for Gumbel-max. When the number of columns grows and the entropy budget does not vanish, the detectable region expands while that of the sum detector does not. Third, for a fixed edit fraction, we characterize the optimal exponential rate of type II error decay and show that the sum detector falls strictly short of it at every fixed alternative. Finally, we propose a weighted multiscale binomial scan, valid in finite samples and needing no knowledge of the next-token distributions, entropy budget, or edit fraction; it attains every detection boundary above and the optimal exponent.
[LG-360] Risk-Averse Inference-Time Diffusion Model Alignment with Finite-Sample Guarantees
链接: https://arxiv.org/abs/2610.17579
作者: Yuchen Liang,Xian Yu
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:Inference-time alignment steers diffusion models toward downstream objectives without costly fine-tuning, but existing methods typically optimize expected cost (or reward) without explicitly controlling tail risk. In this work, we propose a risk-averse framework for inference-time alignment of diffusion models that balances expected cost and Conditional Value-at-Risk (CVaR) under trajectory-level Kullback-Leibler regularization. We characterize the optimal path distribution and reverse transitions as functions of the CVaR threshold, reducing the alignment problem to a one-dimensional optimization problem over the threshold. We show that this one-dimensional optimization problem can be nonconvex in general, establish sufficient conditions for convexity, and introduce quadratic regularization to enable efficient derivative-free ternary search. Under suitable assumptions, we provide finite-sample guarantees for threshold optimization and value estimation, and combine them into a terminal total-variation bound relative to the regularized target under exact resampling. Experiments on synthetic data and image generation using Stable Diffusion demonstrate improved tail performance over risk-neutral baselines while maintaining competitive average cost.
[LG-361] Selective Inference for Clustering Under Unknown Covariance
链接: https://arxiv.org/abs/2610.17458
作者: Yiqun T. Chen,Lucy L. Gao
类目: Methodology (stat.ME); Machine Learning (cs.LG); Applications (stat.AP)
*备注:
Abstract:Unsupervised clustering is widely used to discover underlying group structure, after which researchers often seek to validate these groups via post-hoc hypothesis testing on the population means. However, using the same data both to form clusters and to test for a difference in means inflates the Type I error rate. Recent work in selective inference has proposed solutions when the underlying data are drawn from a multivariate Gaussian distribution with a known covariance matrix or an unknown isotropic covariance matrix. We propose a new selective test for the difference in means between two clusters that provably controls the Type I error rate under an arbitrary unknown covariance matrix. Our test statistic does not require estimating the feature covariance matrix; instead, it leverages a Monte Carlo sampling procedure over the set of orthogonal matrices. We show that this new selective test can be computed efficiently, and we derive closed-form solutions when clusters are formed using k-means or hierarchical clustering. In simulations, we demonstrate that our proposed p-value controls the selective Type I error rate and achieves substantial power, and we illustrate its application on gene expression and drug review data sets.
[LG-362] Monitoring for Shifts at Unknown Times: Parameter-Free Conditional Conformal Martingales
链接: https://arxiv.org/abs/2610.17343
作者: Tuo Liu,Drew Prinster,Francesco Orabona
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Conditional conformal test martingales provide sequential tests for detecting distribution shifts, using a fixed reference dataset to avoid contamination from test-time samples. Prior work used a penalized betting function via online optimization with smoothing, clipping, and other hyperparameters, which can adapt slowly when a shift occurs after a long stationary period. In this work, we introduce an approach that requires no hyperparameters and provably adapts quickly to shifts occurring at unknown times. Our key idea is to frame the penalized betting problem as portfolio selection in a three-asset market. Then, we use a restarting universal portfolio, with a heavy-tailed wealth allocation over time. We establish anytime false-alarm control and derive optimal finite-sample bounds on detection delay under detectable alternatives. We also develop an FFT-based implementation that computes the portfolio sequence over T rounds in \mathcalO(T^2\ln T) time, instead of \mathcalO(T^3) of the naive implementation. Synthetic and real-world experiments suggest that the proposed approach adapts rapidly after long periods of null samples, with detection delays depending substantially less on shift time than prior baselines.
[LG-363] Accounting for model error in ensemble variational data assimilation through learned generative dynamics
链接: https://arxiv.org/abs/2610.17333
作者: Marc Bocquet,Tobias S. Finn,Sibo Cheng
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Atmospheric and Oceanic Physics (physics.ao-ph)
*备注:
Abstract:Many approaches have recently been proposed to combine data assimilation with machine learning for improved forecasts with geophysical models. Here, we focus on combining advanced sequential data assimilation schemes with a generative forecasting model which accounts for complex model errors. This generative model provides a stochastic approximation of the conditional distribution of the successor state given the current state. The proposed variational formulation is a novel one-step weak-constraint 4D-Var scheme in which a learned conditional stochastic forecast replaces the prescribed additive Gaussian model error representation. We derive and implement this formulation as an ensemble variational algorithm, the generative iterative ensemble Kalman filter (IEnKF-G), which generalises the iterative ensemble Kalman filter with additive Gaussian model error (IEnKF-Q). We extend the method to a smoother with an assimilation window spanning multiple time steps and propose a simplified variant. The IEnKF-G is evaluated on a stochastic variant of the Lorenz 1963 model to assess its robustness. It is then tested on a multi-scale variant of the one-dimensional Lorenz 1996 model, in which the resolved variables are affected by model error originating from the hidden small-scale variables. The algorithm is then compared to an ensemble Kalman filter that adaptively mitigates model and sampling errors, and to the IEnKF-Q. The IEnKF-G is generally more accurate in the tested configurations, with substantial gains under partial observations and significant model error. This demonstrates the potential of generative data-driven models when paired with classical data assimilation methods, particularly when model error is substantial. Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Atmospheric and Oceanic Physics (physics.ao-ph) Cite as: arXiv:2610.17333 [stat.ML] (or arXiv:2610.17333v1 [stat.ML] for this version) https://doi.org/10.48550/arXiv.2610.17333 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-364] Deeply unrolled alternating projections for phase retrieval
链接: https://arxiv.org/abs/2610.17316
作者: Jiayi Yao,Vasileios Charisopoulos,Rebecca Willett
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注: 9 pages, 2 figures
Abstract:Deep unrolling methods for inverse problems augment traditional reconstruction algorithms with learned priors to improve their performance. These methods have been developed under the implicit assumption that the geometry of the measurement operator does not change between training and deployment; when this is no longer the case, the quality of obtained reconstructions can degrade significantly. In this paper, we argue that this is not an inherent limitation of the deep unrolling paradigm and can be mitigated by a careful choice of the underlying iterative algorithm. We propose a deep unrolling framework for the phase retrieval problem that augments the alternating projections method with a learned denoising step. Our method obtains high-quality reconstructions without the need for retraining in settings where conventional model-based alternatives struggle, such as when the dynamic range of the measurement operator changes significantly at test time.
[LG-365] Colored Graph Signal Diffusion Models ICASSP2027
链接: https://arxiv.org/abs/2610.17282
作者: Yiğit Berkay Uslu,Samar Hadou,Shirin Saeedi Bidokhti,Alejandro Ribeiro
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注: Submission under review for ICASSP 2027
Abstract:Standard diffusion processes with isotropic noise corruption are not spectrally neutral. They inherit a coarse-to-fine generation hierarchy from the nonuniform data spectrum, because each frequency band loses signal-to-noise ratio (SNR) at its own rate. For signals on a graph, the relevant spectrum is that of the graph shift operator (GSO). A natural construction, therefore, builds both the diffusion dynamics and the denoiser from graph filters of the GSO, and any such design also takes a position on this hierarchy. We expose this hierarchy as a tunable axis through a family of colored score-based diffusions. Fixing the terminal prior to a Gaussian colored by the GSO couples the drift and diffusion through a positive spectral rate, which acts as a clock for each graph-frequency band. The clock sets when a band loses and recovers information, and the forward processes of standard and certain graph-aware diffusions appear as members of this family. A GSO-valued EDM preconditioning splits the denoiser into an analytic Wiener head that carries the Gaussian second-order structure and a graph neural network (GNN) that learns the remaining non-Gaussian dependencies. Numerical experiments show that colored diffusions reach a given generation quality in far fewer sampling steps than standard diffusion.
[LG-366] Solvating machine-learning interatomic potentials for solution-phase chemistry
链接: https://arxiv.org/abs/2610.17226
作者: Anton Z. Ni,Joonho Lee
类目: Chemical Physics (physics.chem-ph); Machine Learning (cs.LG)
*备注: 19 pages, 7 figures, 11 tables
Abstract:Most machine-learning interatomic potentials (MLIPs) are trained on gas-phase data, which limits their use for chemistry in solution. Here, we introduce AnySolv, a transferable implicit solvation potential that provides solvent-dependent energy and force corrections to existing gas-phase potentials without retraining them. It is trained on 1.4 million solvation energies and forces computed with density functional theory (DFT) and SMD in 36 solvents, spanning charged, open-shell, and transition-state structures. Because solvents are represented by continuous physical descriptors, AnySolv supports 179 solvents, including those absent from training. On external benchmarks, AnySolv reproduces DFT-SMD solvation energies to 2-4 meV mean absolute error at more than three orders of magnitude lower cost, and it matches the accuracy of DFT-SMD against experimental solvation free energies. This enables screening of the mechanisms of 120 nucleophilic aromatic substitution reactions in six phases as well as molecular dynamics simulations that reproduce the experimentally observed solvent-dependent conformational preferences of alanine dipeptide. By providing transferable solvation corrections to existing MLIPs, AnySolv enables scalable simulations of solvent-dependent chemical reactivity and molecular dynamics.
[LG-367] Comprehensive Workflow for Uncertainty-Aware Ionic Liquid Solvent Discovery ICDM2026
链接: https://arxiv.org/abs/2610.17171
作者: Alexander Lobo,Urvi Awasthi,Leonid Zhukov
类目: Materials Science (cond-mat.mtrl-sci); Machine Learning (cs.LG)
*备注: 10 pages, 5 figures, 5 tables. Accepted at IEEE ICDM 2026. Copyright 2026 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses
Abstract:Designing synthesizable materials with targeted properties remains one of the central objectives in materials sciences. Ionic liquids (ILs) are well suited to this goal due to their “tunability”, where physical and chemical properties can be precisely customized by modifying their molecular structure, specifically by varying the combination of cations and anions. At the same time, current data-driven property models for ionic liquids often degrade in scarce, noisy, and uneven data regimes, where measurements cover only a small fraction of chemical space and a limited range of thermodynamic conditions. This makes out-of-distribution screening brittle: even when point predictions look plausible, it is difficult to compare candidates consistently, quantify downstream implications, or trust rankings enough to guide molecular discovery. We present a property-prediction and optimal candidate selection workflow that treats data processing, deep learning modeling, and physics-based calculations as a coupled system, linking property prediction with uncertainty estimation and physics-based filtering. By tightening the link between predicted properties and the criteria used for down-selection, the workflow supports reliable out-of-distribution ranking and enables prioritization of previously untested ionic liquid candidates with favorable predicted trade-offs. We demonstrate the application of our workflow by designing novel ionic liquids for CO2 capture from flue gas, predicting their properties and selecting Pareto-optimal candidates based on CO2 solubility, IL viscosity and working capacity.
[LG-368] A Generalized McDiarmid Inequality: Sample Reuse and Non-Uniform Unbounded Differences
链接: https://arxiv.org/abs/2610.17159
作者: Jeremiah Birrell
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 20 pages
Abstract:We develop a novel generalization of McDiarmid’s concentration inequality and use it to obtain a uniform law of large numbers result and error bounds on stochastic optimization problems which simultaneously incorporate two key features: 1) They capture the benefit of training sample reuse in algorithms that also employ easily-sampled auxiliary random variables. 2) They allow for one-component mean-difference bounding functions which are both unbounded in the component of interest and have non-uniform dependence on another component. We demonstrate the utility of our method by deriving concentration inequalities for score-based generative modeling. This application naturally leverages both key features of our new tools: 1) It utilizes auxiliary Gaussian random variables in conjunction with the training samples; accounting for training sample reuse results in tighter error bounds by effectively integrating out the dependence on the auxiliary random variables. 2) The ability to handle non-uniform mean-difference bounding functions allows us to treat unbounded score models and data distributions with unbounded support.
[LG-369] Implicit and explicit reparameterizations of the Folded Normal and Rice distributions
链接: https://arxiv.org/abs/2610.17151
作者: Luis Aldama,Kevin Dalton,Minhuan Li,Doeke Hekstra
类目: Methodology (stat.ME); Machine Learning (cs.LG); Data Analysis, Statistics and Probability (physics.data-an)
*备注: 45 pages, 14 figures, 3 tables. Includes supporting information (pages 29-45)
Abstract:Variational inference is a flexible tool for modeling experimental data and has proven useful in crystallography. Pathwise gradient estimators provide low-variance gradients for reparameterized variational inference, but are not readily available for some distributions central to crystallography. This work derives and implements implicit and explicit reparameterized gradient estimators for the folded Normal and Rice distributions in \progPyTorch, and compares their variance analytically and empirically. In each case the implicit estimator achieves lower variance for the location derivative, particularly at small location-to-scale ratios, while the scale-derivative variances are nearly equal. For both distributions, implicit estimator is proved to be the conditional expectation of the explicit estimator given the sampled magnitude, so its variance is never larger. As a practical application, French-Wilson scaling is formulated as a variational inference problem, using the folded Normal and Rice distributions as variational families for centric and acentric reflections, respectively. Both estimators recover the exact posterior to better than 1% on the amplitude, and converge at the same rate. The implementations are available in the open-source package \progrs-distributions.
[LG-370] SCOPE Before You Sample: Conformal Scores Guide Posterior Exploration for Fast and Calibrated Bayesian Predictive Inference
链接: https://arxiv.org/abs/2610.17146
作者: Zahra Moslemi,Ziyi Liang,Babak Shahbaba
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 44 pages, including appendices
Abstract:Bayesian predictive regions account for parameter uncertainty but can be poorly calibrated under model misspecification, even with expensive full MCMC. Cheaper approximations such as variational inference reduce the cost but often yield less accurate predictive distributions. We propose SCOPE (Score-guided Conformal Posterior Exploration), which retains the predictive quality of full MCMC while achieving calibrated coverage at a fraction of its cost. The key observation is that a prediction region can depend on the posterior mainly through a few parameter directions that move the region’s boundary, and these are generally not the directions of largest posterior variance. SCOPE gives nonconformity scores a new role in guiding posterior computation: score gradients on a held-out guide set, weighted toward the target boundary, identify the relevant directions. MCMC explores this subspace, while an inexpensive projected variational approximation carries the remaining uncertainty. The same guide set calibrates the region’s final density level. We establish asymptotic coverage for SCOPE and provide a split variant with finite-sample validity. Controlled simulations show that score-guided exploration yields more accurate calibrated regions than non-targeted direction choices at low active rank. Across simulated and real benchmarks, SCOPE has a median test-NLL gap of only 0.002 nats relative to full Hamiltonian Monte Carlo (HMC), and on the real benchmarks it runs 12-15 times faster.
[LG-371] In-context classification: Calibration and noise in linear attention with softmax readout
链接: https://arxiv.org/abs/2610.17061
作者: Arman Rysmakhanov,Mary Letey,Yue M. Lu,Jacob Zavatone-Veth,Cengiz Pehlevan
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:We study in-context binary classification in a simple attention model trained to minimise a logistic loss. Each prompt has a latent logistic teacher, and the learner must infer its direction from Bernoulli-labeled context examples. We derive an asymptotic prediction for the test logit in a high-dimensional regime with context length proportional to token dimension, and total training prompts proportional to dimension squared. Our formula cleanly separates calibration, test-time context noise, and non-isotropic parameter-estimation noise as interpretable effects on in-context performance. Our analysis also identifies the ridgeless separability regime and addresses the effects of context length, pretraining-set size, regularisation, and sampling temperature on in-context classification performance. The theory’s predictions closely agree with finite-dimensional simulations and fully trained linear-attention models, while experiments with standard softmax attention show qualitatively related behaviour beyond the regime assumed by our analysis.
[LG-372] Measuring Variable Importance Using Proportional Inclusive Marginal Effects
链接: https://arxiv.org/abs/2610.17053
作者: Youngseo Cho,V. Roshan Joseph
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
*备注: 57 pages, 14 figures, including supplementary material
Abstract:Shapley-based measures are widely used for measuring variable importance in statistics and machine learning. However, these measures have a serious issue–a nonactive variable can get a positive importance measure when it is correlated with an active variable. Herin et al. recently proposed a measure called proportional marginal effects (PME) that mitigates this spurious effect. However, PME is developed for deterministic models and has high computational cost. In this article, we propose a new measure called proportional inclusive marginal effects (PRIME) that overcomes these two limitations of PME. The key idea in PRIME is to use a novel model-free variable selection technique to identify the active variables and include only them in the computation of the importance measure. Analysis of test functions and real datasets show that PRIME can accurately and reliably quantify variable importance from noisy input-output data.
[LG-373] Honeycomb Crystal in a Mass-Asymmetric Electron-Hole Bilayer
链接: https://arxiv.org/abs/2610.17030
作者: Shengdu Chai,Wangqian Miao,Yuqiang Li,Lei Wang,Xi Dai,X.C. Xie
类目: rongly Correlated Electrons (cond-mat.str-el); Materials Science (cond-mat.mtrl-sci); Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
*备注:
Abstract:A two-dimensional electron gas in a uniform neutralizing background forms a triangular Wigner crystal at low density. In a strongly mass-asymmetric electron-hole bilayer, localized heavy holes instead provide a nonuniform positive background that supports an electron crystal at densities two orders of magnitude higher than for a homogeneous electron gas in the same material. Here we show that correlated mobile electrons can overturn the preference for triangular order, stabilizing a honeycomb electron-hole crystal. Using the powerful real-space neural-network variational Monte Carlo in the Born-Oppenheimer limit with fixed holes, we find that increasing the density drives a transition from a triangular electron-hole crystal to a honeycomb crystal, which eventually melts into itinerant electrons coexisting with the honeycomb or the triangular hole lattice at fixed interlayer separation. Relative to triangular order, honeycomb order emerges when gains in electronic kinetic energy and electron-hole attraction outweigh its higher intralayer Coulomb repulsion, a threshold never reached in the rigid point-dipole model. Moreover, the honeycomb hole lattice supports paramagnetic, antiferromagnetic, and stripe-ordered spin states. Complex polarization indicates a metal-insulator transition, while the states on the honeycomb hole lattice are consistent with a stripe-ordered Dirac semimetal and paramagnetic or antiferromagnetic Mott insulators via mapping to the half-filled Hubbard model. We further propose a BLG/hBN/WS _2 -WSe _2 heterostructure as a platform for the predicted honeycomb crystal, at an experimentally feasible density n\sim10^11~\textcm^-2 .
[LG-374] Minimax Optimality of Shampoo for Nonstationary Stochastic Optimization
链接: https://arxiv.org/abs/2610.16923
作者: Abir Sarkar,Nirmalya Mandal,Martin T. Wells
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:We develop a comprehensive theoretical analysis of Matrix Shampoo for tracking time-varying optima in nonstationary stochastic optimization. We derive finite-horizon high-probability bounds for average tracking error that disentangle initialization, optimizer drift, momentum-induced drift penalty, stochastic noise, and adaptive preconditioning, with explicit dependence on momentum parameter (\beta_1) and stabilization parameter (\varepsilon). These bounds reveal an explicit drift–noise and stability–geometry tradeoffs: noise-dominated regimes favor larger \beta_1 , which provides stronger first-moment averaging and greater suppression of gradient noise, whereas drift-dominated regimes favor smaller \beta_1 , allowing faster adaptation to moving optimizer. Larger \varepsilon reduces the preconditioner-perturbation term while weakening Shampoo’s ability to exploit matrix geometry, thereby reducing adaptivity. We establish a matching minimax lower bound under squared path-variation constraint, proving the information-theoretic optimality of Shampoo over the class of causal optimization procedures. Under a constant stepsize, we establish high-probability bounds for the average projected stationarity gap along general nonstationary nonconvex objectives. We validate our theoretical findings on both synthetic and multiple real-world problems.
[LG-375] Feature Normalization under Finite Training Budgets
链接: https://arxiv.org/abs/2610.16845
作者: Johan Larsson
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注: 45 pages
Abstract:Standardizing a rare feature can help stochastic gradient descent learn its effect within a finite budget, but it also changes the noise cost of learning. We characterize the lowest rare-subgroup risks attainable by standardized and unscaled training when each chooses a nonnegative constant learning rate and a deterministic stopping time under the same observation cap. For two independent binary features, we derive both optimized risks as prevalence vanishes while the expected number of rare observations remains finite. Uniform control over rates and stopping times reduces each limiting oracle to a scalar minimization. With positive noise, the preferred scaling changes exactly once as we move signal to the rare feature while the limiting risk at initialization stays fixed and positive. The formulas also explain how noise and budget affect the comparison. Separate finite-prevalence bounds favor standardization on a rare-dominant signal family and unscaled training on a common-dominant family. The results concern final iterates with population preprocessing and fresh observations. Exact-moment searches assess the approximation at finite prevalence. In a controlled study, both orderings persist under observation reuse, estimated preprocessing, and intercept fitting when tuning choices are fixed before evaluation. Selecting a predictor after observing training can reverse an ordering, even when selection uses exact population risk.
[LG-376] Sample-Size Extrapolation in Simulation-Based Inference via Statistical Scaling
链接: https://arxiv.org/abs/2610.16820
作者: Shiyi Sun,Lei Fang,Geoff K Nicholls
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Networking and Internet Architecture (cs.NI)
*备注:
Abstract:In simulation-based inference (SBI), neural posterior estimators trained on a bounded range of sample sizes can become inaccurate on larger datasets because their parameterizations do not specify how posterior location, uncertainty and shape must change as observations accumulate. We propose center-estimated NPE (CE-NPE), which combines a statistical center with an explicit contraction rate and learns a local center correction and residual densities in posterior-scale coordinates. The center is a statistical estimate, such as a closed-form statistic or a simulator-based moment match, which does not require an evaluable likelihood. Our analysis explains how contraction amplifies center error and decomposes Gaussian posterior error into shape, mean and covariance contributions. On six regular benchmarks, CE-NPE gives the best reference-posterior agreement of all tested amortized estimators at every out-of-range size, up to five times the largest training size, without retraining; it also fits better within the training range on most of them. Matched controls and nonregular examples characterize the scope of these gains and identify the conditions on which they depend: an appropriate contraction rate, an accurate center, and a stable residual representation.
[LG-377] Calibrated α-Barycenters for Multiobjective f-Divergence Optimization
链接: https://arxiv.org/abs/2610.16819
作者: Dier Tang,Peijie Li,Xiaoyuan Zhang
类目: Optimization and Control (math.OC); Information Theory (cs.IT); Machine Learning (cs.LG)
*备注: 51 pages, 38 figures, 8 tables, comments are welcome!
Abstract:Matching one distribution to many targets can require repeated evaluations of every target. CAL- \alpha -MODM aggregates a nonlinear predictor into one barycenter, calibrates its residual derivatives, and checks steps against the original f -divergence objective. Under smoothness assumptions, calibration controls parameter-step model error without requiring targets to be close to the model. Exact-oracle stationarity and local convergence guarantees are established. At common matching, canonical geometry cancels the leading quadratic iteration error, giving cubic convergence with second-order calibration and sufficiently accurate inner solves. High-precision experiments support this prediction. Sensitivity-based initialization also accelerates preference continuation along regular stationary branches. On 300-target Jensen–Shannon problems, CAL-p2 achieves median paired speedups of 3.15\times over L-BFGS and 1.10\times over second-order Direct-TR, across 23/25 and 24/25 objective-comparable stationary pairs. The savings arise when cheaper predictor queries offset calibration overhead.
[LG-378] Statistical Downscaling of Daily Precipitation from ERA5 Atmospheric Fields to IMERG over West Africa
链接: https://arxiv.org/abs/2610.16816
作者: David Sagna,Redouane Lguensat,Samo Diatta
类目: Atmospheric and Oceanic Physics (physics.ao-ph); Machine Learning (cs.LG)
*备注: Submitted
Abstract:Global reanalyses are too coarse to resolve the convective systems that produce most West African monsoon rainfall. In this work, we use U-Net based architectures to downscale daily ERA5 fields (0.25°) to IMERG V07 satellite precipitation (0.1°) over West Africa for the June to October seasons of 2000 to 2024, using geopotential, temperature, humidity and wind at three pressure levels together with elevation and land fraction. Adding ERA5 precipitation as input slightly improves performance for this particular challenging region. We find that all the deterministic networks investigated for this work are too dry and lose almost all observed variance below a few hundred kilometers, which motivated our use of a GAN-based alternative. We also investigated permutation importance for the U-Net that ranks 500 hPa specific humidity first, consistent with the sensitivity of deep convection to mid-tropospheric moisture.
[LG-379] On the Calibration of Crowdsourcing Aggregators
链接: https://arxiv.org/abs/2610.16815
作者: Gabriel Singer,Reda Arab,Maïssa Frikha,Samuel Gruffaz,Olivier Vo Van,Nicolas Vayatis,Argyris Kalogeratos
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Calibration has been extensively studied for classifiers, but remains largely unexplored for crowdsourcing aggregation methods, which are increasingly used for making probabilistic decisions. To the best of our knowledge, we provide the first theoretical and empirical study of calibration in label aggregation. To our great surprise, we find that widely used aggregation methods can be severely miscalibrated, and in opposite directions: Majority Vote is underconfident while model-based aggregators are overconfident. We prove that the Bayesian aggregator is perfectly calibrated under its data-generating model. We also show that exact calibration of weighted majority votes requires highly restrictive conditions, which helps explain why Majority Vote is generally miscalibrated in realistic crowds. Empirically, we study calibration across 39 human crowdsourcing datasets and LLM-based crowds. In both cases, we observe systematic miscalibration for aggregation models such as Majority Vote, Dawid–Skene, GLAD, and CrowdFM. Finally, we propose a post-hoc calibration procedure that \emphaggregates the aggregators: for any aggregation function \phi , we introduce a transformation that first combines its predictive distribution with those of complementary aggregators, before applying temperature scaling. This simple and interpretable transformation makes temperature scaling more effective than when applied directly to \phi . It improves calibration, especially in the high-redundancy regime where there are many annotations per item, while generally preserving or even improving accuracy.
[LG-380] When Repeated Measurements Hurt Tabular Foundation Models
链接: https://arxiv.org/abs/2610.16801
作者: Yixiang Liu,Ruizhe Deng,Bibhas Chakraborty,Qiong Zhang,Yan Shuo Tan
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 44 pages, 12 figures
Abstract:Tabular foundation models (TFMs) achieve state-of-the-art performance on IID benchmarks, but how do they respond to repeated measurements at the same inputs? Across six TFMs, synthetic and semi-synthetic experiments show that additional measurements can improve predictions at measured inputs while degrading them elsewhere. Harm is most frequent at low noise, and irrelevant features amplify it in several models; real repeated-measurement datasets show similar patterns. Statistically, the IID rule targeted by PFN pretraining treats each repeated input as fresh evidence about the input distribution. When the prior links this distribution to the response function, the resulting protocol mismatch can outweigh the benefit of additional measurements, even under exact inference. Computationally, interventions identify row-count dependence in TabPFN-v3’s attention scaling as a contributor to harm. Resetting the count largely reverses distortion in a noiseless linear diagnostic and attenuates amplification by irrelevant features. Finally, we evaluate ways to retain the benefits of repeated measurements while limiting their harm. Averaging measurements usually outperforms supplying them separately at larger repetition counts, whereas count reset can suppress both harm and useful gains. Code, saved results and reproduction instructions are available at this https URL
[LG-381] Amortised adaptive importance sampling
链接: https://arxiv.org/abs/2610.16736
作者: Arran Carter,Víctor Elvira,Esmeralda S. Whitammer
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: Code: this https URL
Abstract:Monte Carlo sampling algorithms aim to produce samples from a target distribution with a given unnormalised probability density function. The high cost of running Monte Carlo algorithms, with many simulation steps or particles needed to obtain a good empirical approximation to the target, can be reduced by introducing learnable components into the sampler – a form of amortisation. Once these components are trained, amortised samplers can produce high-quality samples with less computation at sampling time than training-free algorithms would require. We propose an amortised sampler built upon the framework of adaptive importance sampling – a hierarchical importance sampling procedure where the proposal is iteratively adapted. Our amortised adaptive importance sampling algorithm fits a lightweight neural network that adapts the proposal using information obtained over the course of sampling. By training our sampler over many different targets taken from the same family of distributions, we obtain a sampler that can adapt to a specific target at sampling time without a priori access to the target’s parameters. We show that our procedure can efficiently sample from targets within the family the sampler was trained on, giving improvements over target-specific amortised samplers and training-free adaptive importance sampling algorithms. We also explore the potential for generalisation, where the trained sampler is used to sample from targets outwith its training family.
[LG-382] Spectral Costs of Multinomial and Residual Resampling in High-Dimensional Particle Ensembles: A Free-Probability Analysis
链接: https://arxiv.org/abs/2610.16680
作者: Deming Song,Chi Kin Lam
类目: atistics Theory (math.ST); Machine Learning (cs.LG)
*备注: 25 pages, 5 figures. This is a preprint of an article submitted to IEEE TPS, and a version has been posted on engrXiv
Abstract:Resampling is the standard remedy against weight degeneracy in particle filters, yet its effect on the spectrum of the particle second-moment matrix is poorly understood. We study a single resampling step for a d-dimensional ensemble of N particles with d/N - c. For any unbiased scheme, the resampled empirical second-moment matrix dominates its pre-resampling counterpart in conditional convex order, while the conditional mean is preserved; the excess second spectral moment equals the Frobenius energy of the resampling perturbation. For multinomial resampling, the empirical count law converges to the mixed-Poisson law determined by the limiting weight law. For residual resampling, the count law is determined by the joint limit of floor-fraction pairs: we classify all residual limits of weight sequences converging to delta_1 in law as a one-parameter family between the deterministic identity and the full multinomial count law. Each count law feeds a common covariance-side spectral map, yielding exact consequences: a closed-form zero-eigenvalue mass, a rank-and-spectral-gap transition with a universal lower edge benchmark, computable regular edges, unbounded right support, and scheme-specific second-moment costs. The effective sample size determines one bulk statistic but cannot determine the limiting spectrum. Pre-registered numerical experiments confirm the theory at every level.
[LG-383] Amortized Variational Inference of Latent Intensities from Replicated Point Patterns
链接: https://arxiv.org/abs/2610.16669
作者: Wenqing Liu,Xenia Miscouridou,Déborah Sulem
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Bayesian inference for latent point process intensities often relies on per-dataset sampling or optimization, making repeated inference computationally expensive. We propose a neural-network-based amortized variational framework that learns a reusable inference rule mapping a variable number of replicated point patterns to a distribution over their shared latent intensity. Trained from event times using the point process likelihood, the model learns an amortized inference rule without ground-truth intensity supervision. At test time, it performs inference of the latent distribution in a forward pass. On synthetic data, additional realizations improve intensity recovery and yield narrower credible intervals with broadly stable coverage. Our method is tested on bicycle-sharing data and achieves the highest mean held-out log likelihood among the evaluated methods while requiring substantially less inference time than the Log-Gaussian Cox Process baseline.
[LG-384] RFWeave: Learning Wireless Propagation Across Scenes and Transceiver Configurations
链接: https://arxiv.org/abs/2610.16656
作者: Shurui Xu,Xiao Ye,Yijie Bian,Junjie Ai,Yuxuan Chen,Weicong Chen,Wankai Tang,Shi Jin
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注: 20 pages, 12 figures, 12 tables, including appendices
Abstract:Wireless propagation modeling is foundational to the evaluation and design of modern communication systems. To achieve accurate wireless propagation prediction, recent learning-based approaches leverage scene geometry and materials, demonstrating promising generalization across transceiver locations and unseen scenes. However, these approaches remain largely confined to a fixed configuration of array size, receiver orientation, and transmit beam. In this work, we introduce RFWeave, a learned wireless propagation model designed to generalize across scenes, transceiver locations, and transceiver configurations. To accommodate varying array structures, RFWeave incorporates variable-length transceiver tokens that represent antenna elements individually. Furthermore, to capture the geometry between transmit and receive antenna elements, RFWeave aggregates their pairwise spatial relations into propagation tokens. A configuration-conditioned decoder predicts the spatial spectrum and received power. RFWeave is trained on a large-scale simulated dataset spanning 10,000 indoor scenes and more than 5 million channel records under diverse transceiver configurations. Extensive evaluations show that RFWeave achieves state-of-the-art performance across all evaluated generalization settings, improving spatial spectrum structural similarity (SSIM) by up to 15.7% and reducing received power mean absolute error (MAE) by up to 35.4% over the strongest baselines on unseen scenes. Additional experiments on measured DICHASUS data and in beam management further demonstrate RFWeave’s broader applicability to real-world channels and downstream wireless tasks.
[LG-385] textttNFDiag: High-dimensional MCMC Convergence Diagnosis with Normalizing Flows
链接: https://arxiv.org/abs/2610.16574
作者: Kutay Nazlı,Elena Sellentin
类目: Machine Learning (stat.ML); Cosmology and Nongalactic Astrophysics (astro-ph.CO); Instrumentation and Methods for Astrophysics (astro-ph.IM); Machine Learning (cs.LG)
*备注:
Abstract:Accurately assessing the convergence of Monte Carlo Markov Chains is crucial for ensuring their validity in Bayesian inference problems. However, convergence assessment in high dimensions is known to be difficult, due to most convergence diagnostics becoming insensitive beyond a few dozen parameters. Conventional convergence statistics analyze marginal distributions which in high dimensions neglects increasing levels of information from the joint and conditional probability distributions. Thus we propose \textttNFDiag , which taps into the information of the full multivariate joint posterior with a normalizing flow (NF). Our convergence diagnostic thereby also analyzes information from conditional distributions. In other words, by working in the full multivariate space, it is sensitive with respect to the geometry of the chain in high dimensions. Our diagnostic utilizes that NFs assume independent and identically distributed (iid) training samples, which is a testable assumption that will be violated by a non-converged or strongly correlated chain. Our diagnostic trains NFs on separate parts of a chain, and then assesses whether these correspond to the same multivariate target distribution. We test our diagnostic in up to 250 dimensions, and demonstrate it is universal and unbiased, and detects problematic chain quality also when Gelman-Rubin \hatR and the multivarite effective sample size do not.
[LG-386] Online Individualized Treatment Selection under Partial Feedback: False Discovery Control and Minimax Optimality
链接: https://arxiv.org/abs/2610.16464
作者: Runzhe Gu,Zinan Zhao,Wenguang Sun,Xintao Xia
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
*备注:
Abstract:We study online individualized treatment selection when only the outcome under the assigned treatment is observed. Our objective is to preferentially treat individuals whose potential outcomes under treatment exceed a covariate-specific control reference, while limiting false discoveries among treated individuals. We propose Partial-Feedback Adaptive Individualized Treatment Selection (PAITS), which initializes learning from historical data and combines sequential estimation of local null risks, randomized exploration, and feedback-driven updates of the treatment selection threshold. Under regularity conditions, we establish asymptotic false discovery rate control and sublinear gap-weighted assignment regret relative to a population oracle. We further establish a minimax lower bound on assignment regret under prescribed uniform nuisance-estimation requirements and give conditions under which PAITS attains the corresponding constrained minimax rate up to logarithmic factors. These results quantify the tradeoff between accurate treatment assignment and nuisance learning under partial feedback. Simulation studies and a real-data analysis using the Jester recommendation dataset illustrate the finite-sample performance of PAITS.
[LG-387] Explaining Heterogeneous Treatment Effects with Causal Forests and the Hoeffding Functional Decomposition
链接: https://arxiv.org/abs/2610.16447
作者: Gaëtan Patinier,Clément Bénard,Margaux Törnqvist
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:The estimation of heterogeneous treatment effects is a central objective of causal inference. However, state-of-the-art algorithms to solve this task are based on machine learning principles, and thus behave as black boxes. Consequently, the identification of covariates driving heterogeneity remains a challenging and open problem. To address this limitation, we introduce CausalTreeHFD, an interpretable extension of causal forests, based on the Hoeffding functional decomposition adapted to tree ensembles. Our approach decomposes predictions of conditional average treatment effects (CATE) into contributions from individual covariates and pairwise interactions, providing a direct characterization of the factors driving treatment effect heterogeneity. We establish the consistency of CausalTreeHFD, which converges towards the Hoeffding decomposition of the underlying CATE function, under standard assumptions for the analysis of causal forests. Experiments on simulated and real datasets show that CausalTreeHFD accurately identifies relevant covariates and interactions, and precisely recovers their contributions to heterogeneous treatment effects.
[LG-388] aming Nonconvexity with Epigraph Lifting
链接: https://arxiv.org/abs/2610.16429
作者: Clara Lage,Hok Shing Wong,Matthias Ehrhardt
类目: Optimization and Control (math.OC); Machine Learning (cs.LG)
*备注:
Abstract:We consider inverse problems regularized by a trained neural network with \emphconvex activations and \emphunrestricted weights. The resulting optimization problem is generally nonconvex, and gradient-based methods may converge to poor local minima. We introduce a \emphpenalized epigraph lifting of the network and show that the lifted problem admits a \emphdifference-of-convex (DC) decomposition. For Lipschitz continuous activations, we establish \emphfinite exactness: the original objective is recovered exactly at a finite and explicitly computable penalty threshold \bar\rho . Building on this structure, we propose \emphH-DC (Homotopic DC), a homotopic scheme that solves a sequence of DC subproblems with increasing \rho , warm-starting each level from the previous solution. Numerical experiments on synthetic problems, a manifold denoiser regularizer, and image inpainting with a learned Mumford–Shah regularizer show that H-DC consistently escapes poor local minima and reaches solutions with lower objective values than gradient-based methods from the same initialization.
[LG-389] QBFF: Deep Quantum Bottleneck for Automatic Speaker Verification Anti-Spoofing
链接: https://arxiv.org/abs/2610.16390
作者: Anton Firc,Nouhaila Innan,Vojtěch Staněk,Kamil Malinka,Martin Perešini,Zbyněk Lička,Walid El Maouaki,Alberto Marchisio,Muhammad Shafique
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG)
*备注: Accepted for publication at the 2026 IEEE 2nd International Conference on Quantum Artificial Intelligence (QAI)
Abstract:A central goal of quantum machine learning is to identify applications in which small quantum circuits can improve task performance while remaining executable on real hardware. We investigate deepfake (synthetic) speech detection, a largely unexplored application in this area, where reliable confidence scores are as important as discrimination performance because deployment decisions are made at fixed thresholds and overconfident errors can be costly. We introduce QBFF, which routes a compressed speech representation through a small 8-qubit variational circuit (VQC) used as a drop-in bottleneck, with a matched classical bottleneck as the control. The quantum circuit yields better-calibrated confidence in most matched comparisons, including out-of-distribution evaluation, though the downstream deployment-cost benefit is model-dependent. We then run the same circuit on a real IBM quantum processor. To our knowledge, this is the first anti-spoofing pipeline to run on a QPU; hardware execution demonstrates feasibility, with lower subset EER and C_llr but higher mean actDCF than the matched classical reference. Together, these results show that hardware-executable quantum bottlenecks can improve score reliability, with gains that vary by evaluation setting.
[LG-390] Moment-Based Anomaly Detection in High-Dimensional Mixed Data
链接: https://arxiv.org/abs/2610.16285
作者: Davide Ferrari,Andrea Nicolodi
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:We propose an anomaly detection method for high-dimensional mixed-type data that maps heterogeneous features to a common scale via standardized moment-based transformations, yielding interpretable feature-level anomaly scores. These scores are combined through a dependence-aware aggregation step to construct the Mixed-data Influence (MDI) score, accounting for correlations across features and providing a unified measure of how strongly an observation departs from the reference data distribution. By incorporating higher-order moments, the MDI can detect anomalies that arise not only from location shifts, but also from changes in variability, skewness, and other distributional characteristics. The aggregation mechanism allows our method to capture both sparse anomalies concentrated in a small number of features and weak, coordinated deviations distributed across many features. We empirically investigate the distribution and thresholding behavior of the MDI score across different sample sizes, dimensions, dependence structures, and data types. Simulation experiments and two real-data applications, to multi-omics and ESG data, evaluate finite-sample calibration, detection performance, and feature-level interpretability.
[LG-391] Deep learning methods for gamma event selection in TAIGA-IACT image analysis in stereo-mode
链接: https://arxiv.org/abs/2610.16269
作者: E. O. Gres,A. P. Kryukov,P. A. Volchigov,A. P. Demichev,J. J. Dubenskaya,D. P. Zhurov,S. P. Polyakov,E. B. Postnikov,A. J. Razumov
类目: Instrumentation and Methods for Astrophysics (astro-ph.IM); High Energy Astrophysical Phenomena (astro-ph.HE); Machine Learning (cs.LG)
*备注: 10 pages, 4 figures, Proceedings of The 10th International Conference in Deep Learning in Computational Physics, July 6-8, 2026, Moscow, Russia
Abstract:In astrophysical complex TAIGA there are four Imaging Atmospheric Cherenkov telescopes (IACTs), located at a distance of 300-500 m from each other, due to which extensive air showers (EAS) generated by cosmic and gamma radiation are recorded by a single telescope or by multiple telescopes simultaneously (stereo-mode). The analysis of EAS images from two or more Cherenkov telescopes together is of particular interest, since for this approach the reconstruction of physical parameters of the primary gamma ray is much more accurate than in mono-observations. However, the gamma-ray event reconstruction strongly depends on the quality of the hadron background suppression of the classifier, therefore, the gamma event selection stage is one of the important ones in the further analysis of IACT data. We propose and study several neural network methods for gamma event selection in joint TAIGA-IACT image analysis. Neural network methods we suggested involves a single-stage analysis of IACT images from two telescopes, while the second method considers the formation of autoencoder-derived essential features that contain the EAS physical parameters, including the type of primary particle.
[LG-392] Exploration strategies for Gaussian process bandit optimization with inference regret guarantees
链接: https://arxiv.org/abs/2610.16187
作者: Hwanwoo Kim,Jungwoo Hong,Kyungjae Lee
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Optimization and Control (math.OC); Computation (stat.CO)
*备注:
Abstract:Existing cumulative regret guarantees for Bayesian optimization (BO) do not necessarily characterize the quality of the algorithm’s final output. This gap is critical under noisy evaluations, where the goal is to return a single high-quality recommendation. To address it, we introduce efficient exploration strategies that can be incorporated into any BO procedure by occasionally replacing the acquisition query with an exploration query. To validate this approach, we bound the Gaussian process posterior standard deviation over the search space via the fill distance of the queries, which measures the largest region left unexplored. We then use this result to derive upper bounds on the inference regret, i.e., the gap between the global optimum and the function value at the algorithm’s returned point. Empirically, these exploration strategies improve the final recommendations of existing BO algorithms on synthetic benchmarks, hyperparameter optimization, and a control task.
[LG-393] Equilibrium Learning in Unbounded Stochastic Games with Gaussian Transitions
链接: https://arxiv.org/abs/2610.16149
作者: Jonathan Anthony Mavroforas,Fred Roosta
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Learning equilibria from data is difficult when several decision makers interact indefinitely and the system state has no prescribed bounds. We examine discounted general-sum stochastic games with compact action sets and an unknown nonlinear drift observed through Gaussian transition noise. The proposed procedure uses a single uninterrupted trajectory to update Gaussian-process models of the drift and to solve successively refined finite games. We prove a perturbation result showing that stability of equilibrium selection, together with uniform KL accuracy of Gaussian transition models, is sufficient for regional uniform convergence of the induced values and approximate Nash behavior in the original game. The region specifies where the guarantees are evaluated; strategies and dynamics continue to act on the entire Euclidean state space. For GP estimators, simultaneous confidence bounds yield a rate controlled by posterior uncertainty, kernel information gain, and joint-state dimension. Numerical studies illustrate how kernel choice and dimension appear in the predicted convergence behavior.
[LG-394] Distributional Regression for Arbitrary Response Variables Using ReLU Neural Networks
链接: https://arxiv.org/abs/2610.16142
作者: Junpeng Ren,Carlos Misael Madrid Padilla,Oscar Hernan Madrid Padilla
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST); Methodology (stat.ME)
*备注:
Abstract:This paper develops a unified framework for distributional regression with arbitrary response variables. Rather than relying on scalar thresholds or Euclidean structure, we model conditional event probabilities indexed by measurable sets, which extends distributional regression to general response spaces while retaining a direct connection to classical event-based formulations. Within this framework, we study two complementary estimators: a preliminary estimator based on separate regression of event indicators, and a constrained estimator that enforces inclusion-consistent predictions through a directed acyclic graph (DAG) over a finite set family. We establish general error bounds for both estimators and further specialize the results to dense ReLU networks, yielding explicit convergence rates overcoming the curse of dimensionality. Simulation studies and real-data experiments further demonstrate the practical effectiveness of the proposed framework.
[LG-395] CloneOT: predicting clone-specific perturbation responses with lineage-supervised neural inverse optimal transport
链接: https://arxiv.org/abs/2610.16093
作者: Yanshuo Chen,Wei Chen,Heng Huang
类目: Genomics (q-bio.GN); Machine Learning (cs.LG)
*备注:
Abstract:Predicting single-cell perturbation responses requires mapping between two unpaired populations of cells measured before and after the perturbation. Optimal transport (OT) provides a principled framework for constructing such mappings by minimizing the transport cost, and neural OT methods (\textite.g., CellOT) have been employed for this prediction task. However, these methods rely on a fixed transport cost, typically the squared Euclidean distance. This cost may connect cells with the wrong responses. Lineage tracing provides the information needed to learn the cost, as heritable barcodes link the cells of the same clone before and after the perturbation. Here, we propose \CloneOT, a neural inverse OT method that learns both the transport cost and the transport map from partially lineage-traced single-cell datasets. \CloneOT\ represents the cost as a neural network. It learns the cost by minimizing the extra cost that the lineage adds to the optimal transport, so that the cheapest transport keeps the cells of each clone together. Our experiments demonstrate that \CloneOT\ recovers the correct transport on 2D datasets, while the squared Euclidean cost fails. Moreover, \CloneOT\ outperforms five baselines on 14 lineage-tracing perturbation scenarios of cancer cells, for both seen and unseen clones. Furthermore, we show that \CloneOT\ reveals the clone-specific transcriptional programs of drug-tolerant persister cells and forecasts the fate of each clone from its cells before treatment. Our implementation is available at this https URL.
[LG-396] A Nonlinear Time-Series Theory for Constant-step SGD with Momentum
链接: https://arxiv.org/abs/2610.16034
作者: Younghun Lee,Jake Zummo,Wei Biao Wu
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 42 pages, 2 figures
Abstract:Constant-step stochastic gradient descent with momentum (SGDM) augments the parameter recursion with a momentum state that carries geometrically decaying memory of past stochastic gradients. We study this coupled recursion as a nonlinear time series and develop a framework for its stationary behavior and distributional limits. The joint parameter-momentum process forgets its initialization geometrically under a finite p th moment for p 1 , yielding a unique stationary causal solution. Under a second-moment condition, this contraction leads to an invariance principle and quenched Gaussian limit for averaged SGDM, while a small-learning-rate analysis gives a Gaussian limit for the last iterate. In a stochastic-curvature quadratic model, we further study how the rate of initialization forgetting relates to stationary variance and averaged fluctuations.
[LG-397] Limit Theory for Online Expectile Estimation under Short and Long Range Dependence
链接: https://arxiv.org/abs/2610.16030
作者: Younghun Lee,Jake Zummo,Wei Biao Wu
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST)
*备注: 36 pages, 1 figure
Abstract:Expectiles are tail-sensitive distributional summaries with important applications in financial risk measurement and forecasting, yet statistical theory for their online estimation from dependent data remains limited. We study a one-pass, stochastic-gradient algorithm with constant learning rate \gamma for estimating the marginal expectile of a stationary causal process. The recursion admits a unique stationary causal solution and geometrically forgets its initialization. Under short-range dependence, its last iterate is asymptotically Gaussian at the \sqrt\gamma scale, with variance governed by the population curvature and the long-run variance of the expectile score. For long-memory linear processes the scale changes to \gamma^\beta-1/2\ell(1/\gamma) , with a Gaussian limit arising from a fractional Brownian motion integral. We further derive the joint long-memory limit for multiple expectile levels and characterize its limiting covariance structure. Various numerical experiments corroborate the predicted Gaussian approximations and regime-specific fluctuation scales.
[LG-398] Using Line Segment Tracking to enable ML-based Track Reconstruction in CMS
链接: https://arxiv.org/abs/2610.15963
作者: Aashay Arora on behalf of the CMS Collaboration
类目: High Energy Physics - Experiment (hep-ex); Machine Learning (cs.LG)
*备注: Submitted to CHEP 2026
Abstract:High-pileup conditions in CMS during the HL-LHC era make charged-particle tracking increasingly challenging as detector occupancy and combinatorics grow. We present a hybrid approach that exploits Line Segment Tracking (LST) objects rather than individual hits to enable the first CMS ML-based track reconstruction algorithm. The LST objects, namely the triplets used in this study, are built according to geometry- and physics-driven criteria and carry richer local structure, geometrical cues, and physics quantities such as transverse momentum, which helps the model incorporate detector-level information from the start. We apply a Transformer encoder with geometrically masked sparse self-attention to embed these objects in a learned latent space, and use object condensation to assemble full tracks in a single inference step. This object-based representation reduces complexity in dense environments and makes the model less sensitive to ambiguities introduced by high pileup. Evaluated on CMS Phase-2 simulation of t\bart events at \langle \mu \rangle = 200 , the approach achieves reconstruction efficiency comparable to the LST quintuplet baseline while reducing the fake rate by roughly a factor of three and suppressing duplicates almost entirely, demonstrating the promise of advanced ML architectures for next-generation particle tracking.
[LG-399] Small area learning
链接: https://arxiv.org/abs/2610.15953
作者: Katarzyna Reluga,Nicola Salvati,Setareh Ranjbar,Nikos Tzavidis,Mark van der Laan
类目: Methodology (stat.ME); Machine Learning (cs.LG); Statistics Theory (math.ST)
*备注: 53 pages, 1 figure, 10 tables
Abstract:Small area estimation (SAE) traditionally borrows strength across domains through a prespecified working model. We introduce small area learning (SAL), a learner-agnostic framework that formulates SAE as a multi-target semi-supervised learning problem and treats outcome prediction as a nuisance task. Areas may differ through unrestricted covariate shift and structured conditional shift, with information shared through common and area-specific components. We develop standard and cross-fitted estimators for low-complexity and flexible learners, unifying synthetic, model-based, and fully residual-corrected estimation. For area means, we establish consistency and asymptotic normality and provide variance estimators and confidence intervals. We extend the theory to settings with only aggregated population covariates and characterize the resulting aggregation bias. SAL reveals a bias–variance trade-off: partial residual correction can reduce mean squared error, whereas full correction yields estimators robust to working-model misspecification, potentially at higher variance. Simulations yield RMSE reductions of 34%–49% relative to direct estimation, with the best configuration varying across settings. An empirical application favours a cross-fitted estimator using a flexible learner under the chosen validation criterion. The numerical studies also illustrate how shrinkage can produce overly optimistic confidence intervals at the expense of strong modeling assumptions.
[LG-400] wo Linear Programs are Better than One
链接: https://arxiv.org/abs/2610.15909
作者: Alexander E. Black,Christoph Hertrich,Stefan Kober,Georg Loho
类目: Combinatorics (math.CO); Computational Complexity (cs.CC); Discrete Mathematics (cs.DM); Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:The extension complexity \operatornamexc§ of a polytope P is the minimum number of facets of a polytope that projects onto it. Motivated by the complexity of neural networks, Hertrich and Loho introduced the virtual extension complexity \operatornamevxc§ of a polytope P , which is the minimum of \operatornamexc(Q) + \operatornamexc® among all polytopes Q and R such that P+Q=R . We show the following results pertaining to virtual extension complexity. First, we resolve the core open problem on the topic by showing the virtual extension complexity can be exponentially smaller than extension complexity. Second, we show there are matroid base polytopes with exponential virtual extension complexity by adapting a semi-algebraic counting argument by Hrubeš that extends the counting arguments of Rothvoss. Finally, we show that the maximum virtual extension complexity of a polygon with n vertices is \Theta(\sqrtn) . This resolves an analogue of a question on extension complexity of Fiorini, Rothvoss, and Tiwary in 2011 in the case of virtual extension complexity. Subjects: Combinatorics (math.CO); Computational Complexity (cs.CC); Discrete Mathematics (cs.DM); Machine Learning (cs.LG); Optimization and Control (math.OC) Cite as: arXiv:2610.15909 [math.CO] (or arXiv:2610.15909v1 [math.CO] for this version) https://doi.org/10.48550/arXiv.2610.15909 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-401] Minimax Rates and Improved Spectral Convergence for Multi-Manifold Clustering
链接: https://arxiv.org/abs/2610.15892
作者: Haoyu Chen,Chenghui Li
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:We study eigenpair estimation for the tensorized, density-weighted Laplace–Beltrami operator from n independent data points sampled from a multi-manifold mixture distribution where manifolds may intersect. Let d denote the largest intrinsic dimension of the manifold components and s\geq2 the smoothness of the sampling densities. For simple, nontrivial eigenpairs, we establish the minimax rate n^-s/(d+2s) for approximating the eigenvalue and eigenfunction of the tensorized Laplace–Beltrami operator, and show that a continuum plug-in estimator based on kernel density estimation attains the same rate when the supporting manifolds are known. We further analyze a manifold-agnostic graph Laplacian estimator and prove that its eigenpairs converge to those of the tensorized Laplace–Beltrami operator at rate n^-1/(2.5d-1) , up to logarithmic factors, improving upon previously known spectral convergence rates for multi-manifold learning. Numerical experiments on synthetic multi-manifold data corroborate the rates guaranteed by our theorems. These results characterize the information-theoretic difficulty of multi-manifold spectral estimation and quantify the additional cost of learning without oracle component information.
[LG-402] Entropy-Regularized Mean-Field Games with Averag e-Reward Criterion: Span-Seminorm Contractions and Model-Free Learning
链接: https://arxiv.org/abs/2610.15871
作者: Gunes Akbas,Naci Saldi,Berkay Anahtarci,Can Deha Kariksiz
类目: Optimization and Control (math.OC); Computer Science and Game Theory (cs.GT); Machine Learning (cs.LG)
*备注: 37 pages, 1 figure, 2 tables
Abstract:We study stationary discrete-time mean-field games with finite state and action spaces under the average-reward criterion, with an entropy-regularized objective and population-dependent dynamics and rewards. Under a minorization condition, the soft Q-operator is a span-seminorm contraction for the full transition kernel, so relative value iteration does not require the minorizing measure. When the Lipschitz constant of the kernel is below the minorization mass and the temperature exceeds an explicit threshold, the equilibrium fixed-point operator is a contraction: an equilibrium exists, its population distribution is unique, and a fixed-point iteration converges to it geometrically. We then learn the equilibrium model-free from a simulator, combining soft Q-learning at a fixed population distribution with an empirical transition kernel. We prove a finite-sample bound for soft Q-learning in the span seminorm, with temperature-independent constants, and derive convergence rates for the learned population distribution and policy. A congestion-aware routing game illustrates the results.
[LG-403] Inference on the Value of Policy Improvement in Adaptive Experiments
链接: https://arxiv.org/abs/2610.15789
作者: Justin Whitehouse,Emma Brunskill,Vasilis Syrgkanis
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Econometrics (econ.EM); Statistics Theory (math.ST)
*备注: 97 pages, 10 figures, 13 tables
Abstract:In this paper, we develop an asymptotically normal estimator for the value of policy improvement (VPI) under adaptively collected data. This value, which is defined as the gap between the value of the best treatment policy in a class and that of a baseline policy, gauges whether or not it is worth deploying a personalized treatment policy. Building an estimator for the VPI is naturally more difficult than estimating the value of a fixed policy. This is because, in addition to contending with the correlated data and decaying propensities associated with adaptive experiments, any estimator must also handle the non-regularity, or lack of pathwise differentiability, of the underlying estimand when the optimal policy is non-unique. Our proposed estimator works by sequentially fitting a weighted empirical welfare maximizing policy on historical data and then evaluating it using inverse propensity weighting on future samples. We neither assume that the optimal policy is unique nor make a margin assumption on the underlying heterogeneous treatment effects. A key difficulty in analyzing our estimator is showing that the regret incurred through policy learning is asymptotically smaller than the standard error of the estimator. We accomplish this by showing faster regret rates for policy learning under decaying propensities, which may be of independent interest. We not only evaluate the coverage of our estimator on a mix of synthetic and semi-synthetic data, but also employ it to estimate the VPI in an adaptive experiment evaluating the efficacy of job assistance programs in Jordan due to Caria et al. Although our estimates are generally more conservative than their hierarchical Bayesian analysis, which assumes that the optimal policy is unique, we still corroborate the benefits of personalization amongst Syrian refugees.
[LG-404] Adaptive Robust Confidence Sets in High Dimensions
链接: https://arxiv.org/abs/2610.15723
作者: Ilias Diakonikolas,Chao Gao,Daniel M. Kane,Yuetian Luo
类目: atistics Theory (math.ST); Data Structures and Algorithms (cs.DS); Machine Learning (cs.LG); Methodology (stat.ME); Machine Learning (stat.ML)
*备注:
Abstract:We study adaptive confidence sets for mean vectors of multivariate Gaussians with identity covariance under Huber contamination when the contamination proportion is unknown. We show that an information-theoretically optimal adaptive confidence set can be constructed by simultaneously quantifying uncertainty across all one-dimensional projections. The resulting confidence set achieves the optimal diameter and reveals a substantial adaptation cost compared with the minimax estimation error. Although statistically optimal, this construction is computationally infeasible. We establish a Statistical Query (SQ) lower bound showing that achieving diameter O(m^-1/2) requires either exponentially many queries or query tolerance d^-\Omega(m) . Thus, in contrast to robust mean estimation, adaptation to the unknown contamination proportion incurs both a statistical cost and a computational barrier within the SQ framework. Finally, we give a low-degree moment-matching algorithm that constructs an adaptive confidence set using d^O(m) samples and time. Its diameter bound contains an additional O(m^-1/2) term beyond the information-theoretic rate, matching the dependence on m in the SQ lower bound. Together, these results characterize the statistical limits of adaptive robust confidence sets under Huber contamination and quantify a computational barrier within the SQ framework. Subjects: Statistics Theory (math.ST); Data Structures and Algorithms (cs.DS); Machine Learning (cs.LG); Methodology (stat.ME); Machine Learning (stat.ML) Cite as: arXiv:2610.15723 [math.ST] (or arXiv:2610.15723v1 [math.ST] for this version) https://doi.org/10.48550/arXiv.2610.15723 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-405] Notes on Uniform Policy Score Bounds and Lipschitz Continuity for Truncated Gaussian Product Cauchy Boltzmann and Softmax Policies
链接: https://arxiv.org/abs/2610.15690
作者: Yanjie Dong
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Probability (math.PR)
*备注:
Abstract:We establish uniform score bounds and Lipschitz continuity for truncated Gaussian, Cauchy, Boltzmann, and softmax policies under explicitly stated parameterizations.
[LG-406] When do explicit penalties improve flow matching?
链接: https://arxiv.org/abs/2610.15689
作者: Lea Kunkel,Alberto González-Sanz,Sophie Langer
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST)
*备注:
Abstract:Flow matching can learn structure directly from data, raising the question of when additional constraints improve generation. We study penalties that encode such constraints, with physics-informed flow matching as a motivating example. Satisfying a constraint more closely need not make the generated distribution more accurate. We therefore compare penalised and unpenalised generative modelling through their expected squared 2 -Wasserstein error. Our analysis covers both vector-valued and functional data in a Hilbert-space setting. We first derive geometric criteria that account for constraint violations and the distribution of samples after projection onto the constraint set. We then solve the statistical comparison explicitly for specific velocity-field classes and constraints via flow matching. In this model, penalisation improves accuracy exactly when the reduction in estimation variance outweighs the increase in squared bias. The resulting formula shows how this trade-off depends on sample size, the model class, the noise schedule, and the penalty weight.
[LG-407] Screening Collapse under Recursive Synthetic Data
链接: https://arxiv.org/abs/2610.15617
作者: Jia Zhang,Xin Chen
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 20 pages, 5 figures
Abstract:Recursive synthetic-data training can degrade a fitted model, but its effects on the variable-screening stage that precedes model fitting are less understood. When each generation labels data for the next, a marginal screener ranks variables against the current pseudo-label target rather than the original response. Across generations, this target drift can erode or reverse the active–inactive score margin and cause screening to discard variables that clean-label screening would retain. We call this process \emphscreening collapse. In a Gaussian linear model with minimum-norm recursive generators, we establish finite-sample screening bounds, characterize a late-generation information-loss limit, and derive a minimax impossibility result. Screening collapse can precede and compound ordinary model collapse: omitted variables incur a covariance-weighted loss that no refit restricted to the selected variables can eliminate, while recursive labels further degrade estimation on the retained variables. We also examine two approaches to mitigating screening collapse. Fixed-fraction real-label retention preserves a clean-label component but leaves resolution fixed, whereas data accumulation can avoid screening collapse if the pooled target preserves a positive margin and the growing sample resolves it. Numerical experiments on gene-expression data support the theoretical findings.
[LG-408] Joint Contrastive-Adversarial Representation Learning for Unsupervised Domain Adaptation
链接: https://arxiv.org/abs/2610.15613
作者: Yuling Jiao,Lican Kang,Jerry Zhijian Yang,Zhou Yu,Fusheng Zhou
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 29 pages, 5 figures, and 9 tables
Abstract:Effective unsupervised domain adaptation (UDA) requires a representation that captures transferable structure from both the labeled source domain and the unlabeled target domain. We introduce Joint Contrastive-Adversarial Representation Learning (JCAR), a two-stage framework that learns a common representation from the union of unlabeled source and target samples by combining contrastive learning with adversarial domain alignment, and then freezes the representation to fit a source-supervised predictor. We establish a non-asymptotic excess-risk bound for this procedure that couples representation and domain-discriminator estimation while separating downstream prediction, with explicit neural-network approximation errors. Under the stated assumptions and common Hölder smoothness \beta\geq1 , the bound identifies a sufficiently large unlabeled-sample regime in which the downstream rate is \widetilde\mathcalO(m^-\beta/(2\beta+d)) , where m counts labeled source samples and logarithmic factors are suppressed. The labeled-sample exponent is governed by the latent dimension d rather than the ambient dimension p , which enters the pretraining cost. Experiments on Office-31, Office-Home, and DomainNet demonstrate clear transfer gains across diverse domain shifts, with particularly strong performance when representations are learned from scratch.
[LG-409] Latent-Space Importance Weighting under Covariate Shift
链接: https://arxiv.org/abs/2610.15571
作者: Yuling Jiao,Lican Kang,Yang Xiang,Jerry Zhijian Yang,Fusheng Zhou
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 37 pages, 2 figures
Abstract:In this work, we propose a transfer learning algorithm for covariate shift, integrating sample reweighting with representation learning. A key feature of our approach is the use of a lower-dimensional latent space for modeling, in contrast to conventional methods that operate directly in the high-dimensional ambient space. Our method estimates the density ratio of learned latent features for downstream weighted prediction. We establish explicit target-domain excess-risk rates with independent data folds for strictly sufficient learned representations with regular learned-space prediction scores and density ratios. We give dimension comparisons under stated smoothness conditions and a variance comparison for true weights and a fixed predictor. General finite-sample inequalities retain representation-quality errors. Experiments on simulated and real-world datasets demonstrate gains from combining learned representations and latent weighting, and reveal sensitivity to initialization and domain choice.
[LG-410] Beyond Aggregate Drift: Optimal Client-Specific Windows in Federated Learning
链接: https://arxiv.org/abs/2610.15543
作者: Yuriy Dorn,Dmitrii Vishovan,Kirill Novoselov,Ilya Kuruzov,Darina Dvinskikh,Alexander Gasnikov
类目: Optimization and Control (math.OC); Machine Learning (cs.LG)
*备注:
Abstract:We study federated learning from continuously arriving data: each client samples from its own distribution, these distributions drift over time, and the server must track the minimizer of the clients’ weighted average loss. Clients are not alike as they collect data at different rates, with different noise, and change at different speeds. The main question is how much history should each one keep? A single shared window must serve every client at once, and so makes a fast-changing client carry the sampling noise of a slow, nearly stationary one. We give each client a memory of its own, sized to that client’s noise, arrival rate, and drift, and refresh it by restarted accelerated method that starts from a common point. The tracking error is then bounded by [ R\left(\sum_m w_m^3/2,\sigma_m\sqrt\gamma_m/\lambda_m\right)^2/3, ] for clients with weight w_m , noise \sigma_m , arrival rate \lambda_m , and drift rate \gamma_m on a domain of diameter R . This is never worse than what any common window can achieve, and strictly better whenever noise and drift are spread unevenly across clients. A lower bound that puts different groups of clients on different drift clocks shows this quantity is the right one. The two bounds agree up to a constant whenever the clients occupy boundedly many refresh scales, and in general differ only by the sixth root of the number of scales. Experiments show the gain comes from the memory rather than from the solver: it persists across accelerated, stochastic-gradient, and one-step methods, and against a separately tuned common-memory baseline. Subjects: Optimization and Control (math.OC); Machine Learning (cs.LG) Cite as: arXiv:2610.15543 [math.OC] (or arXiv:2610.15543v1 [math.OC] for this version) https://doi.org/10.48550/arXiv.2610.15543 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Kirill Novoselov [view email] [v1] Sun, 11 Oct 2026 11:34:54 UTC (366 KB)
[LG-411] RF Fingerprinting of DSP Implementations using Machine Learning: An Over-the-Air Case Study
链接: https://arxiv.org/abs/2610.15494
作者: Stefan Scholl
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注:
Abstract:Radio frequency fingerprinting aims to identify transmitters by exploiting small differences in the emitted signal waveform. Common approaches exploit hardware-dependent imperfections for that purpose. This work instead investigates whether differences between digital signal processing implementations of a transmitter can be identified from their radiated signals. As a practical case study, five different software-based DSP implementations for generating a 2-FSK RTTY communication waveform are considered. In a monitoring receiver, a classifier can identify these DSP implementations using convolutional neural networks that are trained on augmented software-generated signals represented as raw IQ, power spectrum or instantaneous frequency. The trained networks are evaluated on over-the-air recordings obtained using a shortwave transceiver and a software-defined radio receiver. The best neural network combines IQ and the power spectrum representation and achieves an accuracy of up to 90%. The study further examines the influence of the signal SNR and the observation bandwidth on the classification accuracy. The results demonstrate that DSP implementations leave distinguishable characteristics in the transmitted signals and suggest that DSP-dependent fingerprinting can provide complementary information for specific emitter identification systems.
[LG-412] Personalized Physiological-Model-Based Neural Network for Glucose Forecasting in Type 1 Diabetes
链接: https://arxiv.org/abs/2610.15456
作者: Yaowen Zhang,Shuhao Que,Ying Wang
类目: Medical Physics (physics.med-ph); Machine Learning (cs.LG)
*备注:
Abstract:Accurate, physiologically consistent glucose forecasts could help people with type 1 diabetes respond to impending excursions. We applied our physiological-model-based neural network (PMB–NN) framework to glucose forecasting. A temporal convolutional network (TCN) used 180 min of history, constrained during training by a personalized four-parameter physiological model. The main PMB–NN combined forecast and neural–physiological agreement losses; a three-term variant additionally fitted physiological predictions to measured glucose. Data comprised 6–8 weeks of continuous glucose monitoring, pump, and meal records from 25 AZT1D participants. Person-specific models used chronological 70/10/20 train/validation/test splits and 30-, 60-, and 120-min horizons. Median mean absolute errors were 16.24, 25.00, and 30.20 mg/dL with basal insulin records and 15.38, 24.21, and 31.79 mg/dL before these records became available. In the latter setting, PMB–NN had the lowest median error at 30 and 60 min; it outperformed the physiological model alone in all six settings. In input-increase tests, glucose predictions changed in the expected direction for 85% of correction boluses and 100% of carbohydrate rescues, versus 57% for active insulin and 56% for active carbohydrates. Most standalone physiological fits showed locally distinguishable parameter effects, with glucose-related coefficients more stable under resampling than insulin- or carbohydrate-related coefficients. The three-term variant improved hyperglycaemia accuracy with basal records, but did not significantly change overall MAE and reduced directional plausibility in three tests. PMB–NN combines personalized forecasting with an explicit physiological model, allowing assessment of predictive accuracy, physiological response behavior, and participant-specific coefficients.
[LG-413] LiveAnon: Improving Privacy and Utility in Streaming Speaker Anonymization with Codec Language Models
链接: https://arxiv.org/abs/2610.15379
作者: Yejin Kim,Kyungsu Kim,Seungu Han,Kyogu Lee
类目: Audio and Speech Processing (eess.AS); Machine Learning (cs.LG); Sound (cs.SD)
*备注:
Abstract:Speaker anonymization (SA) aims to conceal speaker identity while preserving linguistic content and emotion, but existing streaming systems often lose emotion information through discrete content bottlenecks or require emotion-specific modeling. We present LiveAnon, a voice-conversion-based streaming SA system that directly conditions an autoregressive codec language model on hidden features from a pretrained streaming ASR encoder, preserving emotion without emotion-specific supervision. To reduce codec-generation latency, we introduce capped RVQ decoding, which preserves dependencies among coarse codebooks while predicting the remaining codebooks in parallel. Under VPC24, LiveAnon improves intelligibility and emotion preservation over fixed-target Stream-Voice-Anon (SVA), a strong prior streaming SA system, matches emotion-prompted SVA in emotion preservation, and achieves the strongest semi-informed privacy among streaming systems. Under VPC26, it ranks second across all metrics among the evaluated anonymization systems, demonstrating a strong privacy-utility balance.
[LG-414] Deep Neural Mixed Models for Nonlinear Longitudinal Prediction with Subject-Specific Heterogeneity
链接: https://arxiv.org/abs/2610.15323
作者: Mahmud Hasan,Niloofar Ramezani,Hrishikesh Chakraborty
类目: Methodology (stat.ME); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Longitudinal health data often exhibit within-subject correlation, subject-specific heterogeneity, and complex nonlinear predictor–response relationships that may not be adequately captured by conventional mixed models. We propose a Deep Generalized Neural Network Mixed Model (DGNMM) that extends the GLMM framework by using a deep neural network to learn nonlinear population-level relationships while incorporating directly trainable, regularized subject-specific effects to capture individual heterogeneity. The model is estimated using prediction-oriented stochastic optimization with separate regularization of the neural-network parameters and subject-specific effects. This formulation allows us to investigate when increased nonlinear representation capacity provides additional predictive benefit in the presence of subject-specific heterogeneity. A simulation study varies population-level nonlinearity and subject-specific heterogeneity and compares DGNMM with GLMM, artificial neural network (ANN), Maity–Pal (MP), and GNMM alternatives for continuous and binary longitudinal outcomes. In the real-data application, DGNMM achieved the lowest MSPE for continuous depression and anxiety prediction, while alternative models remained competitive for binary outcomes. Exploratory subject-level analyses showed heterogeneous predictive gains across individuals, but the magnitude of the estimated subject-specific effects was not significantly associated with predictive gain. Overall, increased nonlinear representation capacity was most beneficial for continuous prediction under nonlinear response structures, while greater subject-specific heterogeneity did not systematically yield larger predictive gains.
[LG-415] What a Data Release Must Preserve for Off-Policy Evaluation
链接: https://arxiv.org/abs/2610.15283
作者: Kyungbok Lee,Yating Zou,Michael R. Kosorok
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:In off-policy evaluation, merging decision-time histories that matter differently for a specified policy value or comparison can make that evaluation unidentified, even with observed actions and outcomes and known behavior and target policies. We characterize which distinctions must be preserved: identification holds if and only if the action-specific evaluation weights are constant within each released record, and the coarsest exact release groups histories by these weights. For normalized outcomes, the largest sharp width induced by the release equals a behavior-weighted within-record radius of these weights. The same radius becomes a release-design objective when each record may be refined into at most m categories. The optimum has an exact m -center characterization. When behavior does not vary within a released record, we give a finite-data repair algorithm whose maximum evaluation-weight radius is at most twice the finite-data optimum. For a fixed action-outcome distribution within a released record, a distinct sharp-identification problem remains. Under the same condition, the action-specific outcome distributions are identified, and optimizing over their couplings gives the sharp interval. We construct an AIPW envelope containing this sharp interval; it matches the interval for all such outcome distributions exactly when the extreme within-record action-coefficient vectors take at most two values in each action coordinate. In 24 KuaiRec tasks, the repair algorithm yields a 93% median reduction in mean pairwise sharp width at m=64 relative to the unrepaired release.
[LG-416] Explicit Bounds on Hoffman Constants for Sparse Polyhedral Regularizers
链接: https://arxiv.org/abs/2610.15241
作者: Faris Chaudhry,Anthea Monod,Keisuke Yano
类目: Optimization and Control (math.OC); Machine Learning (cs.LG); Statistics Theory (math.ST)
*备注: 51 pages, 4 figures
Abstract:Hoffman constants quantify the distance to the solution set of a polyhedral system in terms of its residual; they control error bounds, geometric conditioning, the convergence rate of first-order iterative algorithms, and the stability of regression estimators to target perturbations. Despite their importance, explicit values of Hoffman constants are rarely available due to their computational complexity. In this paper, we derive explicit bounds on the Hoffman constant for a class of structured polyhedral systems arising in sparse regularization. We show that the constants admit a decomposition into spectral, geometric, and interaction components, which together determine their scaling behavior. This framework explains the gap between worst-case dimension-dependent bounds and the favorable conditioning observed empirically in structured problems. As a primary application, we analyze \ell_1 -regularized least squares, demonstrating that the effective constants depend on low-dimensional active subspaces rather than the ambient dimension. We then extend to the elastic net and adaptive \ell_1 -regularized formulations, demonstrating how the underlying optimization geometry is altered by the \ell_2 penalty and reweighting scheme. Furthermore, we relate our results to classical statistical assumptions and show that the restricted eigenvalue condition bounds only the primal curvature of the active set, leaving the dual interaction geometry unconstrained and is thus insufficient alone for fast theoretical convergence rates. We further show standard \ell_\infty coherence conditions do not yield dimension-independent Euclidean bounds to close this gap. Our theoretical results show that Hoffman constants admit explicit, interpretable forms in structured regression settings and, in practice, provide actionable tools for post-hoc feature pruning and a priori optimal hyperparameter selection.
[LG-417] Non-Euclidean Gradient Descent Has Many Edges of Stability
链接: https://arxiv.org/abs/2610.15223
作者: Peter Zaika,Zakhar Shumaylov
类目: Optimization and Control (math.OC); Machine Learning (cs.LG)
*备注:
Abstract:Recent observations that deep learning models train at the edge of stability have renewed interest in understanding stability properties of classical optimization algorithms. For quadratic objectives, standard gradient descent has a sharp stability threshold: it converges from every initialization when \eta2/L , diverges from almost every initialization when \eta2/L , and diverges from every nonzero initialization when \eta2/\mu . For non-Euclidean gradient descent, the corresponding known guarantees are weaker: convergence holds for every initialization when \eta2/S , where S is the generalized sharpness, while every nonzero initialization diverges when \eta2/\mu . The behavior in the intermediate regime 2/S\eta2/\mu is not fully characterized. We show that Hilbertian norms are the only ones for which 2/S is an almost-everywhere sharp threshold for every positive-definite quadratic objective. Using linear conjugacy between non-Euclidean and Euclidean gradient descent, we show that for general norms the intermediate regime can contain arbitrarily many edges of stability associated with distinct attracting directions. We identify a forward-invariant region of exponentially divergent directions beyond the generalized stability threshold. When the directions of minimum curvature are finite and nondegenerate, we prove that, sufficiently close to 2/\mu from below, only their invariant directions remain convergent. We give smooth two-dimensional constructions with arbitrarily many distinct thresholds, arbitrarily large relative threshold separation and show these properties to be stable.
[LG-418] Physics Foundation Models for Efficient Control of Fluid Flows
链接: https://arxiv.org/abs/2610.15173
作者: Mahindra Rautela,Shaifalee Saxena,Alexander Scheinker
类目: Fluid Dynamics (physics.flu-dyn); Machine Learning (cs.LG)
*备注:
Abstract:We investigate whether a physics foundation model (PFM) pretrained for PDE forecasting can be repurposed for closed-loop fluid-flow control. We construct a PFM-based controller by attaching lightweight Gaussian actor and critic heads to frozen pretrained representations and train the policy using proximal policy optimization (PPO). We evaluate the approach on HydroGym’s two-dimensional Kolmogorov-flow environment using vorticity observations at (64\times64) and (128\times128) resolutions and Reynolds numbers (\mathrmRe=100) and (200). Across these settings, we compare the frozen PFM controller against a baseline MLP policy trained with the same PPO framework. The PFM controller achieves consistently higher reward while requiring smaller executed action magnitudes over training, indicating improved control performance with reduced control effort. These results provide initial evidence that representations learned through self-supervised next-frame prediction can be transferred directly to closed-loop fluid flow control. Project page: this https URL.
[LG-419] Data-Topology Compatibility Enables Personalization in Decentralized Learning
链接: https://arxiv.org/abs/2610.15126
作者: Veronica Xu,Lingfei Wang,Chenlei Leng,Fengxiang He
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Decentralized learning must balance collaboration with personalization across heterogeneous agents. This paper studies when cooperation can support statistically coherent groups of personalized models. We show that data-topology compatibility enables robust orbit-wise personalization: when local objectives and communication topology have (approximately) compatible automorphisms, agents within an orbit share (approximately) equal limiting distributions, while different orbits remain distinct without explicit personalization. This creates a new intermediate regime between global consensus and independent local learning. Under exact data-topology compatibility, as a reference system, we decompose the stochastic learning dynamics into orbit-level and transverse components, derive a necessary and sufficient stability criterion for distributional synchronization, and characterize the orbit-structured stationary law. Modelling an approximate data-topology compatible system as an \varepsilon -perturbed reference system, we prove the introduced errors have uniform-in-time and stationary O(\varepsilon) bounds in 2-Wasserstein distance. We also prove sufficient conditions for lower stationary average excess risk than both independent local training and the best single shared parameter. Experiments on synthetic data, MNIST, and CIFAR-10 are in full agreement with the theory. The code is available at this https URL.
[LG-420] Energy-Guided Learning for Heat Equation with Noisy Source: Optimal Estimation and Generalization Bounds
链接: https://arxiv.org/abs/2610.15111
作者: Zixiang Zhou,Xiaohui Chen,Shanghua Teng
类目: atistics Theory (math.ST); Machine Learning (cs.LG); Methodology (stat.ME); Machine Learning (stat.ML)
*备注:
Abstract:Data and the governing partial differential equations (PDEs) of continuous-time dynamical systems provide rich structure for statistical learning. A central challenge is how to integrate data with PDE constraints in a principled way to improve generalization. In this paper, we characterize the statistical estimation and generalization properties of learning solutions to the heat equation with a noisy source function from finite, randomly sampled data. We first derive a novel variational formulation, which we characterize as the minimizer of loss function for an energy regularized and ellipticity debiased Sobolev regression. By formulating the learning task as a supervised problem over space-time domains under physically meaningful norms, our approach goes beyond the traditional supervised learning framework by effectively leveraging \emphauxiliary data—samples of the source function of the heat equation—to design a \emphphysics-informed energy regularization. Our analysis reveals that learning in linear parabolic PDEs exhibits intrinsic anisotropy between spatial and temporal regularity, in sharp contrast to the static elliptic problems, leading to fundamentally different estimation and generalization behaviors. Specifically, we derive non-asymptotically rate-optimal estimation errors, as well as generalization error, which explicitly capture the interplay among temporal sampling, spatial resolution, and solution smoothness. Our numerical experiments demonstrate substantial orders of magnitude of reduction in testing errors against various physics-informed neural network methods on various parabolic and hyperbolic PDEs.
[LG-421] DisCo Algebra: Why Disintegration and Combination Are Not Inverses
链接: https://arxiv.org/abs/2610.15006
作者: Xuexing Lu
类目: Probability (math.PR); Machine Learning (cs.LG); Mathematical Physics (math-ph); Category Theory (math.CT); Statistics Theory (math.ST)
*备注: 8 pages
Abstract:We present DisCo algebra—a compositional framework for joint probability built from a symmetric cooperad (disintegration) and a symmetric operad (combination) on the bare family \mathcalP_n=\mathrmProb([0,1]^n) . No probability integral transform, no external summary functions, no enrichment of objects with conditional kernels. The disintegration side uses block-wise joint CDFs as intrinsic, measure-determined 1D summaries, yielding a lossy compression. The combination side reads the same summaries and reconstructs via regular conditional measures, yielding a lossless embedding. We prove that combination followed by disintegration is the identity on admissible inputs, but disintegration followed by combination is not—the information loss is exactly the conditional KL divergence between the true conditional distribution and the conditionally independent product. This asymmetry reveals a hard limit on how much cross-block dependence can be captured by scalar block summaries.
[LG-422] DiffGA: Steering Frozen Diffusion Models for Materials Design via Evolutionary Search
链接: https://arxiv.org/abs/2610.14999
作者: Shivang Agarwal,Rodrigo Wang,Kevin Ryczko,Aayush R. Singh
类目: Materials Science (cond-mat.mtrl-sci); Machine Learning (cs.LG)
*备注:
Abstract:Steering candidate generation toward user-defined objectives is central to inverse materials design, yet remains difficult when objectives are black-box, changing, or non-differentiable, as in many simulation- and experiment-derived properties. Generative models offer a solution but typically require retraining or fine-tuning for each new objective, demanding labeled data and substantial compute, or rely on differentiable guidance that is brittle to gradient noise and undefined for many practical targets. We present DiffGA, a retraining-free framework that injects evolutionary search into the sampling trajectory of a retrained diffusion model, evolving partially denoised states under an arbitrary black-box oracle. Because oracle evaluation often dominates campaign cost, we couple DiffGA with an on-the-fly Bayesian surrogate that reuses representations from state-of-the-art machine-learned force fields (MLFFs). A Gaussian process is fit during the campaign on the force fields’s graph embeddings and, once it demonstrably ranks an unseen generation, replaces expensive property evaluations under an uncertainty-aware acquisition rule, reducing oracle cost with negligible loss of accuracy. Across three crystal-design tasks using a frozen MatterGen backbone, DiffGA outperforms diffusion-only, compute-matched, purely evolutionary, and fine-tuned baselines in oracle-assessed rates of stable, unique, and novel hits. Rather than proposing scientific compounds, we use density functional theory to quantify MLFF oracle fidelity, showing that it is a faithful proxy in key regimes but systematically biased in others, delineating when proxy-driven generative discovery can be trusted. On a representative sample, DiffGA’s top hits survive first principles re-evaluation as physically realistic hits, while the surrogate-accelerated pipeline preserves this fidelity at a fraction of the cost.
[LG-423] An Algebraic Approach to Constrained K-means Clustering
链接: https://arxiv.org/abs/2610.14932
作者: Hiroki Oohori,Kentaro Fukushima,Michio Yamamoto
类目: Methodology (stat.ME); Machine Learning (cs.LG); Computation (stat.CO)
*备注: 21 pages, 7 figures, 4 tables; includes appendices
Abstract:Clustering is typically performed using only object features, although additional information about relationships between objects may be available and can be incorporated through constraints. Among these, pairwise constraints require two objects to belong to the same cluster or to different clusters. Several extensions of the k-means algorithm accommodate such constraints, but they sometimes fail to find a feasible assignment or incur high computational cost. Motivated by these limitations, we propose an algebraic approach to k-means clustering with pairwise constraints. The algorithm simplifies the constraints using Gröbner bases before optimization, improving the efficiency of cluster assignment. It can also detect whether the constraint set has a feasible assignment before optimization. Comparative experiments with existing algorithms showed that the proposed algorithm achieved comparable or better clustering quality at a lower computational cost, particularly when the numbers of constraints and clusters were small.
[LG-424] Minimum Volume Optimal Transport for Multivariate Conformal Prediction
链接: https://arxiv.org/abs/2610.14907
作者: Chuchen Zhang,Haojie Ren,Zhaojun Wang,Yajie Bao,Changliang Zou
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Optimal transport (OT) enables multivariate conformal prediction (CP) to construct flexible prediction sets that reflect the underlying complex distributional geometry while retaining distribution-free coverage guarantees. Existing OT-based methods typically rely on a fixed Euclidean transport cost, which can be suboptimal in terms of the volume of prediction sets. We propose Minimum-Volume OTCP, a framework that learns a Mahalanobis ground cost to directly minimize prediction-region volume over a class of positive-definite metrics. Using a metric Brenier representation, we derive a volume-based objective that depends explicitly on the Mahalanobis cost matrix and optimize it jointly with the transport map. We establish finite-sample marginal coverage of the output prediction sets and optimality of volume metrics. We further develop an individualized extension that learns covariate-dependent transport metrics, adapting the shape and orientation of prediction sets to heterogeneous conditional distributions. For this extension, we also show the asymptotic conditional coverage of our method. Simulations and real-data experiments demonstrate substantial reductions in prediction-set volume.
[LG-425] FDTD Model of Noise Pollution from an AI Data Center
链接: https://arxiv.org/abs/2610.14899
作者: Gavin Buxton
类目: Computational Physics (physics.comp-ph); Machine Learning (cs.LG)
*备注:
Abstract:The sudden growth in AI data centers is raising numerous environmental concerns. Low-frequency noise pollution is a concern for local residents, that may not just be an annoyance but also carry significant health impacts. An FDTD model of sound propagation over the topography near an AI data center is presented. The topography plays a large role in the level of noise pollution, with areas of sudden elevation on either side of the river near the AI data center more susceptible to high noise levels, while areas further afield largely protected from noise pollution by the topography. Understanding the role that topography plays in noise pollution, and capturing the noise pollution in specific topographies, can help identify areas where noise mitigation strategies might be best employed.
[LG-426] Physics-informed random neural architectures on compact boundaryless multiply connected manifolds for random-operator models in machine learning and uncertainty quantification
链接: https://arxiv.org/abs/2610.14844
作者: Christian Soize
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: Published in Computer Methods in Applied Mechanics and Engineering}, doi: https://doi.org/10.1016/j.cma.2026.119437 , 463, 119437 (2027)
Abstract:We introduce stochastic physics-informed neural networks (SPINNs) for stochastic boundary-value problems (SBVPs) with random operators. The method enforces physics through a structural probabilistic coupling: a common Gaussian stochastic germ drives both the SBVP random operator and a manifold-based stochastic neural architecture. The stochastic neural network is non-feedforward and intrinsically random. A latent Gaussian field and an inhomogeneous Poisson point process on a compact, boundaryless, multiply connected manifold generate random neuron locations, geodesically local sparse connectivity, and a sparse random weight matrix, all governed by a low-dimensional hyperparameter. For deterministic inputs, the network output is random and can represent non-Gaussian random operators. Training is performed in observation space: SBVP simulations produce random observables, while the SNN defines a predictive conditional probability density function. The supervised objective combines data fidelity, through negative log-likelihood, with a cross-entropy alignment term relative to the SBVP-induced conditional distribution. The resulting target is a convex mixture of observation-induced and SBVP-induced distributions, with a dimensionless parameter controlling robustness to model–data mismatch. The shared stochastic germ yields common-random-numbers Monte Carlo estimators and reduced-variance gradients. The method uses only evaluations of the SBVP solution mapping, without differentiating or modifying it, and avoids PDE-residual or boundary-penalty losses. A stochastic elliptic problem on a periodic cylindrical surface illustrates the approach in SBVP-consistent and SBVP-misspecified regimes. A generalized-posterior formulation is also outlined for hyperparameter inference and posterior averaging without per-query reoptimization.
[LG-427] Strong Gaussian Approximation of Constant-Learning-Rate SGD: Bias and Trajectory Inference
链接: https://arxiv.org/abs/2610.14567
作者: Joseph Keslin,Wei Biao Wu
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:This paper develops strong Gaussian approximations (KMT type) for constant learning rate stochastic gradient descent (SGD). We work with heavy tailed noise and assume that the number of iterates grows and the learning rate shrinks simultaneously. Our work constructs simple Gaussian processes on the same space and bounds, in probability, the maximum difference between the SGD errors and their respective limits. While there exist central limit theorems for constant rate SGD centered at the optimum, these only provide individual iterate inference. By contrast, our quenched KMT approximations allow for finite sample, simultaneous, whole-trajectory-level inference and imply stochastic process convergence (Functional Central Limit Theorems (FCLTs)). For accumulated errors centered at the true optimum, we first obtain a KMT-type approximation bound when the learning rate scales as \gamma_n=\fracc\sqrt n+o(n^-1/2) for a constant c \in \mathbbR^+ and n iterations. The result reveals a trade off between controlling the bias of SGD and the initialization effect, with no learning rate scaling generally making both contributions vanish in the accumulated error limit. To eliminate these contributions, we consider a Burn in scheme and Richardson-Romberg extrapolation and prove respective KMT and FCLT theorems. We also prove a finite-horizon KMT-type approximation and FCLT for the rescaled iterate trajectory to a stationary Ornstein–Uhlenbeck (OU) process. We consider applications in risk control and simultaneous inference and perform simulations to illustrate our results.
[LG-428] Learn the Dynamics Enforce the Physics: Generative Latent Operators for High-Dimensional Beam Evolution
链接: https://arxiv.org/abs/2610.14537
作者: Haohan Zou,Alexander Scheinker,Yuanyuan Shi
类目: Accelerator Physics (physics.acc-ph); Machine Learning (cs.LG)
*备注: 46 pages (9+37), preprint for discussion
Abstract:Particle-in-cell (PIC) simulation is indispensable for modeling beam dynamics in particle accelerators, but its computational cost limits many-query workflows, such as configuration exploration, virtual diagnostics, and controller design. We introduce GLOBE, a physics-constrained Generative Latent Operator for Beam Evolution, modeling the nonlinear, control-dependent, multiscale evolution of 3D beam–field dynamics covering 10^7 spatial cells. GLOBE compresses beam–field states into a compact latent space and factors their evolution into two components. A control-conditioned neural operator learns the dominant dynamics, while an amplitude-normalized generative model captures the residual distribution, recovering fine-scale structures and conditional variations. GLOBE enforces field constraints by regime-specific, parameter-free operators: a Poisson solve reconstructs the electric field in the electrostatic regime, and a spectral projection imposes Gauss’s law and a divergence-free magnetic field in the electromagnetic regime. On two high-resolution 3D particle-accelerator benchmarks, bunched-beam transport and plasma-wakefield acceleration, GLOBE lowers relative- L_2 error by 9% and 13%, low-frequency spectral error by 69% and 51%, and mid-frequency spectral error by 54% and 20% relative to the strongest baseline (DPOT), satisfies field constraints, and reduces hours of PIC simulation to a minute per rollout.
[LG-429] Identication and Estimation of Nonsymmetric Determinantal Point Processes
链接: https://arxiv.org/abs/2610.14520
作者: Christian Gouriéroux,Yang Lu
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Determinantal point processes (DPPs) provide tractable distributions for nite random sets, but the usual symmetric-kernel specication imposes nonpositive pairwise dependence. Statistical research on nonsymmetric DPPs, which can accommodate positive dependence, remains limited. We study identication and estimation for this class. First, we establish the equivalence of two sucient conditions for a nonsymmetric kernel to dene a valid DPP. Second, we develop the identication theory for nonsymmetric DPPs. Third, we develop cardinality-truncated maximum-likelihood estimation and estimation by averaged stochastic gradient descent. Under suitable identication and regularity conditions, we establish consistency and asymptotic normality for both estimators. Simulations examine identication, the eects of truncation, and the nite-sample performance of the estimators. We illustrate the methods with two cyber applications at markedly dierent sampling frequencies: weekly sector-by-breach data and second-by-second enterprise authentication data with more than ve million set observations.
[LG-430] Decomposing A Neural Learner
链接: https://arxiv.org/abs/2610.14388
作者: Zelong Bi
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 1st submission
Abstract:Neural learners for classification and regression are often criticized as black boxes. We propose a probabilistic decomposition of neural learners that accounts for heterogeneous predictability across the feature space. A training procedure combining the EM algorithm, stochastic gradient descent, and maximum likelihood estimation is developed to jointly learn the predictable component and an input-dependent reliability score. We show that the optimal decomposed learner recovers the response distribution within the predictable component, while the reliability score characterizes the predictability of individual observations. We further extend the framework to incorporate localized feature transformations. Experiments on synthetic and real-world datasets demonstrate the effectiveness of the proposed framework.
[LG-431] STREGO: A Scalable Implementation of a Trust-Region Framework for Noisy Bayesian Optimization
链接: https://arxiv.org/abs/2610.14338
作者: Youssef Diouane,Issmail El Hallaoui,Meisseme Kadri,Amina Lamghari
类目: Optimization and Control (math.OC); Machine Learning (cs.LG)
*备注:
Abstract:TREGO, a trust-region Bayesian optimization (BO) method for expensive blackbox problems, has convergence guarantees, but only for exact function values. In high dimension, its performance tends to deteriorate. This paper asks whether a trust-region BO method can be extended to handle noisy evaluations while being competitive in high dimension. We first generalize TREGO into a trust-region framework for noisy BO and extend its convergence analysis to noisy function estimates that are sufficiently accurate with sufficiently high probability. We then propose STREGO, a scalable implementation of this framework. Its global phase selects a batch of candidates by solving a bi-objective problem on the posterior mean of a Gaussian process and the integrated variance reduction. Unlike the predictive variance, this second objective does not favor boundary points merely because they are far from the data. Its local phase minimizes the posterior mean of a second Gaussian process, trained only on points close to the incumbent. STREGO is compared with TREGO, TURBO, BAXUS and NSMA-BO on synthetic and real-world problems of dimension 4 to 124 , with and without noise. It gives the best average results in every synthetic setting tested and the best or tied-best solution on four of the six real-world problems. An implementation of STREGO is available at this https URL.
[LG-432] A Stochastic Analysis Perspective on Diffusion Models
链接: https://arxiv.org/abs/2610.14133
作者: Stanislas Strasman(SU, LPSM (UMR_8001))
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Probability (math.PR)
*备注:
Abstract:Generative modelling involves two complementary tasks: learning a probabilistic approximation of an unknown data distribution from observations, and designing a tractable mechanism for sampling from this approximation. We present a stochastic-analysis perspective on score-based generative models, connecting ideas from classical sampling theory, score estimation, and diffusion time reversal. We first explain how score functions link Langevin sampling with score matching, and how Gaussian smoothing leads to denoising score matching and multiscale sampling methods. We then emphasize the distinction between dynamics preserving a fixed distribution and dynamics transporting a prescribed family of marginals. This motivates a path-space treatment of diffusion time reversal. Finally, we connect the ideal reverse diffusion with practical generative algorithms.
[LG-433] On physical implementability of reinforcement-learning-driven reasoning
链接: https://arxiv.org/abs/2610.14131
作者: Eiji Konishi
类目: Disordered Systems and Neural Networks (cond-mat.dis-nn); Machine Learning (cs.LG)
*备注: 3 pages, revtex
Abstract:In the machine learning community, the reinforcement-learning-driven reasoning process has been intensively studied since around 2025. In this article, we present a negative result on the hardware implementation of this reinforcement-learning-driven reasoning process in a physical neural network model in the presence of a reward external potential energy for a given reward. In this hardware implementation, the existence of a chain of thought in dynamics contradicts the Lyapunov stability of the system.
[LG-434] Hierarchical Multi-Agent Reasoning Systems for Knowledge-Grounded Material Substitution
链接: https://arxiv.org/abs/2610.14123
作者: Tarjei Paule Hage,Subhadeep Pal,Wei Lu,Yu-Chuan Hsu,Gayla Lyon,Jiezhu Jin,Markus J. Buehler
类目: Materials Science (cond-mat.mtrl-sci); Mesoscale and Nanoscale Physics (cond-mat.mes-hall); Machine Learning (cs.LG)
*备注:
Abstract:AI-assisted scientific discovery increasingly seeks to move beyond generating plausible ideas toward reproducible, constraint-aware workflows that can support practical engineering decisions. We present MARS, a hierarchical Multi-Agent Reasoning System for knowledge-grounded material substitution under scientific and manufacturing constraints. We apply MARS to replace per- and polyfluoroalkyl substances (PFAS), a demanding materials-design problem in which substitutes must satisfy functional performance, environmental, experimental, and processing requirements. MARS decomposes substitution into a closed-loop workflow comprising requirement analysis, grounded candidate generation, and manufacturability assessment with process design. The system integrates retrieval-augmented generation, knowledge-graph reasoning, a database of lab-available materials, persistent memory of intermediate results, and rejection-aware iterative feedback. In a PFAS-substitution case study, MARS generates structured material requirements, identifies an evidence-grounded substitute candidate, and develops a lab-scale manufacturing route. In a multi-seed ablation evaluation, MARS achieves the highest overall performance across all three subsystems, with gains that persist when the backbone model is held fixed. These results suggest that hierarchical decomposition, knowledge grounding, and manufacturability-aware refinement improve material-substitution reasoning beyond model capability alone.
[LG-435] Early Detection of Pension-Sponsor Distress from Public Filings: A Sponsor Vulnerability Index Stress-Augmented VaR and Dynamic Self-Investment Caps
链接: https://arxiv.org/abs/2610.14120
作者: Ravil Akhtyamov
类目: Risk Management (q-fin.RM); Machine Learning (cs.LG)
*备注: 9 pages, 4 figures, 4 tables. Code and data: this https URL
Abstract:A defined-benefit pension is only as safe as its sponsor, yet sponsor distress is under-instrumented relative to plan-side modelling. We present a transparent, reproducible early-warning pipeline built entirely on public data. A Sponsor Vulnerability Index (SVI), a gradient-boosted classifier over point-in-time SEC EDGAR balance-sheet fundamentals and two macro series, predicts a filings-based distress state one to two quarters ahead. On 36 U.S. sponsors in five sector groups (1,418 firm-quarters, 2015Q1-2025Q3) it reaches an out-of-fold AUC-ROC of 0.887 and PR-AUC of 0.711 (base rate 11.9%) under leave-one-firm-out cross-validation. A Monte Carlo VaR with a stylized transition-stress overlay shows that climate-type stress is negligible for a diversified plan but adds 20-43% to one-year 95% VaR when 22.5% of assets are held in sponsor securities. A dynamic self-investment cap maps the SVI to a limit that falls from 8% to below 1% in acute distress and recovers with the sponsor, which we illustrate on PGE’s 2019 bankruptcy and Boeing’s persistent fragility. We report the method’s failure modes alongside its results (book-equity labels that also fire on non-credit events such as spin-offs, positives concentrated in one firm, and stylized stress parameters) and release code, data and results under a concept DOI.
[LG-436] Disease-level drug approval forecasting from trial-agnostic public data
链接: https://arxiv.org/abs/2610.14013
作者: Samir Townsley,Pagkratios Tagkopoulos,Ilias Tagkopoulos
类目: Quantitative Methods (q-bio.QM); Machine Learning (cs.LG)
*备注:
Abstract:Drug development is an expensive and lengthy process requiring developers to select drugs for human trials and diseases to target. Computational tools that can estimate before testing which drug-disease candidates are most likely to reach FDA approval could reduce development time and costs through early program advancement and disease prioritization. Yet existing approaches predict outcomes at a single level or stage or do not distinguish between disease outcomes for the same drug. We ask (1) whether FDA approval can be forecast for a drug-disease pair using only pre-clinical trial information, and (2) whether these forecasts can help prioritize diseases for the same drug. We built a cohort of 14,134 drug-disease pairs by reconstructing programs from this http URL, extracting disease-level approvals from FDA label documentation, and enriching each pair with molecular, disease, ADMET, target, and pathway features. We divided the cohort into pre- and post-2019 programs and trained a classifier on pre-2019 data to predict FDA approval. On the post-2019 cohort (2,637 programs), the model would have prioritized a program that ultimately gained approval over one that did not approximately three out of four times. For drugs tested across multiple diseases, selecting the model’s highest-ranked disease would have identified an ultimately approved disease 61.3% of the time, compared with 40.7% under random selection. These results show that public pre-trial information contains signal for forecasting FDA approval at the drug-disease level and can help prioritize diseases for the same drug. We provide a public web tool for SMILES- and ICD-10-based drug-disease approval forecasting.
[LG-437] Deterministic denoising of long-duration gravitational-wave signal candidates with α-(de)blending
链接: https://arxiv.org/abs/2609.15308
作者: Przemysław Figura,Michał Bejger
类目: General Relativity and Quantum Cosmology (gr-qc); High Energy Astrophysical Phenomena (astro-ph.HE); Instrumentation and Methods for Astrophysics (astro-ph.IM); Solar and Stellar Astrophysics (astro-ph.SR); Machine Learning (cs.LG)
*备注: 23 pages, 12 figures. Submitted to Classical and Quantum Gravity
Abstract:We present a deterministic machine-learning method for denoising candidate signals from all-sky searches for continuous gravitational waves from rotating neutron stars. Using the time-domain F -statistic search pipeline, we post-process the resulting F(f,\dotf) candidate patterns with iterative \alpha -(de)blending, a deterministic diffusion-type generative model implemented as a U-Net, well suited to the non-Gaussian, correlated noise of the F -statistic output. Two models, trained on data based on simulated 6- and 12-day time-domain segments with software-injected signals similar to the LVK hardware injections, are tested by comparing deblended images to the library of expected signal patterns via the Structural Similarity Index Measure. In general the method recovers the correct sky-position-dependent pattern for signal-to-noise ratios \rho \gtrsim 4 , moderately below typical all-sky detection thresholds, with recovery depending strongly on pattern morphology, demonstrating that deterministic diffusion-based denoising may serve as a consistency/veto tool ahead of the semi-coherent coincidence stage. We also discuss limitations and possible improvements of the method.
附件下载
