本篇博文主要内容为 2026-10-19 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 (2026-10-19)
今日共更新826篇论文,其中:
- 自然语言处理共147篇(Computation and Language (cs.CL))
- 人工智能共176篇(Artificial Intelligence (cs.AI))
- 计算机视觉共128篇(Computer Vision and Pattern Recognition (cs.CV))
- 机器学习共223篇(Machine Learning (cs.LG))
- 多智能体系统共17篇(Multiagent Systems (cs.MA))
- 信息检索共11篇(Information Retrieval (cs.IR))
- 人机交互共24篇(Human-Computer Interaction (cs.HC))
多智能体系统
[MA-0] Self-Building Labs
【速读】：该论文旨在解决自驱动实验室（Self-driving Laboratories, SDLs）在面对新实验时缺乏灵活性与可扩展性的问题。现有SDL系统仅能执行预设的实验流程，若需适应新的实验任务，仍需大量人工进行硬件配置与软件编程，导致实验室的可扩展性受限，成为科学发现加速的瓶颈。其解决方案的关键在于提出自构建实验室（Self-Building Laboratory, SBL），该系统能够根据高层次的科学目标，自主设计、配置并操作完整的实验环境。SBL从现有基础（如空房间或已有运行中的SDL）出发，自动构建缺失的要素，包括实验室布局、数字孪生模型、机器人技能以及实验工作流。通过在不同机器人平台和实验设置上的验证，SBL实现了从单工作站到多机器人协同实验室的可扩展部署，并在真实世界量子点合成实验中展示了其端到端的自动化能力。该方法将科研人员的角色从繁琐的设备配置与程序编写转变为科学目标的定义，显著提升了科学自动化系统的可访问性、灵活性与可扩展性。
链接: https://arxiv.org/abs/2610.23826
作者: Jiaru Bai,Yuchi Zhao,Han Hao,Miroslav Bogdanovic,Yang Cao,Varinia Bernales,Alán Aspuru-Guzik
机构: 未知
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA); Robotics (cs.RO)
备注:
Abstract:Self-driving laboratories (SDLs) accelerate scientific discovery by automating experimental execution and optimization. AI agents can increasingly propose new experiments, but existing SDLs can only run the procedures they were built for. Adapting them to a new experiment still takes extensive manual engineering, which makes laboratory flexibility the bottleneck for scaling discovery. We introduce the self-building laboratory (SBL), which designs, configures, and operates its own laboratory setup given a high-level scientific goal. SBL starts from whatever already exists, from an empty room to a working SDL, and builds the missing capabilities: the lab layout, its digital twin, robot skills, and experimental workflows. We demonstrate the system working across different robots and setups, scaling from a single workstation to laboratories with several robots working together. We validate SBL’s end-to-end capabilities through a real-world quantum-dot synthesis experiment. By shifting the role of human researchers from manually configuring and programming device automation to specifying scientific objectives, SBL makes scientific automation more accessible, flexible, and scalable. Project website: this https URL
[MA-1] CocoRNA: Collective RNA Design with Cooperative Multi-agent Reinforcement Learning ICML2026
【速读】：该论文旨在解决RNA序列逆向设计（RNA inverse design）中如何高效生成能够稳定折叠为特定二级结构的序列这一关键挑战。传统方法在处理高维、非线性且约束复杂的序列-结构映射关系时，往往面临搜索空间巨大、优化效率低及泛化能力差等问题。本文提出的CocoRNA框架采用协作式多智能体强化学习（cooperative multi-agent reinforcement learning）机制，将复杂的逆向设计任务分解为多个可并行求解的子问题，由多个智能体协同完成。其核心创新在于引入一个集中式价值函数（centralized critic），利用全局结构信息对各智能体进行策略指导，从而实现联合优化与高效探索。该设计显著降低了问题复杂度，提升了策略学习的质量与泛化能力，使模型在未见结构上无需额外训练即可生成高质量序列。实验结果表明，CocoRNA在Rfam数据集上的成功率和设计速度均显著优于现有先进方法，并展现出在其他生物序列设计任务中的广泛适用性，验证了其在复杂设计问题中的强大潜力。
链接: https://arxiv.org/abs/2610.23820
作者: Tianmeng Hu,Biao Luo,Ke Li
机构: 未知
类目: Machine Learning (cs.LG); Multiagent Systems (cs.MA)
备注: ICML 2026
Abstract:Designing RNA sequences that reliably fold into specific secondary structures is essential for understanding their biological functions but remains a challenging computational problem. We propose CocoRNA, a cooperative multi-agent reinforcement learning framework for RNA inverse design. CocoRNA simplifies the design task by decomposing it into smaller sub-problems, each solved collaboratively by multiple agents. This approach reduces the complexity of the problem and improves the exploration of design policies. During training, a centralized critic uses global structural information to guide the agents, enabling them to jointly optimize their design strategies. As a result, CocoRNA learns high-quality RNA design policies that generalize effectively to unseen structures without additional training. Experiments on the Rfam dataset demonstrate that CocoRNA substantially outperforms state-of-the-art methods in both success rate and design speed. Further experiments on other biological sequence design tasks highlight the effectiveness and broad potential of CocoRNA for complex design tasks.
[MA-2] Annotrace: A Workbench for Step-Level Annotation of Agent Traces AAAI2027
【速读】：该论文旨在解决大语言模型（LLM）智能体在实际应用中失败原因分析时所面临的标注瓶颈问题。当前研究普遍依赖人工手动标注智能体执行轨迹（agent traces），且每项研究均需自行开发临时性标注工具，导致重复劳动与标准不统一。现有可观测性平台虽可记录轨迹，但缺乏结构化标注能力；通用文本标注工具则无法理解智能体执行中的“步骤”“任务委派”或“工具调用”等关键粒度，难以界定最小标注单元。针对此问题，Annotrace提出一种统一的解决方案：支持六种主流轨迹格式（包括OpenTelemetry GenAI、LangSmith和Langfuse），将复杂轨迹解析为有序嵌套的执行步骤列表。其核心创新在于构建人机协同的标注流程——用户可自定义标签体系，前沿生成式模型基于样本轨迹自动提出初始标注建议，以与人类相同的提问方式参与标注，实现真正的联合标注。同时，Annotrace内置标注一致性评估机制，可量化任意两名标注者（人或AI）间的协议程度。在多智能体问答数据集上的实验表明，Annotrace能精确复现先前人工标注的235个标注单元，并使两个模型对同一轨迹的标注一致性达到91%（κ = 0.82），显著提升了标注效率与可靠性。
链接: https://arxiv.org/abs/2610.23284
作者: Arun Josephraj Arokiaraj,Chenyi Huang,Sahan Bulathwela,Zekun Wu,Bin Wu
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: 4 pages, 4 figures, 1 table. Under review at the AAAI 2027 Demonstration Program
Abstract:Every study of why LLM agents fail begins by labelling traces by hand, and every such study builds its own throwaway tool. Observability platforms record agent traces but offer no structured labelling; annotation tools label text but have no notion of a step, a delegation or a tool call, leaving it unclear what the smallest unit of execution to label should be. Annotrace reads six trace formats, including OpenTelemetry GenAI, LangSmith and Langfuse, and turns each run into an ordered, nested list of steps. The annotator writes their own labelling codebook; a frontier model proposes a starting set from a sample of their own traces, answers first, and is asked exactly the questions the human is asked, making annotation a joint human-AI process. Annotrace measures agreement between any two annotators, human or AI. On a multi-agent question-answering corpus it reproduces a previous manual effort’s 235 annotation units exactly, and two models labelling the same traces agree 91% of the time at \kappa = 0.82 .
[MA-3] owards High Semantic Fidelity: Hyperdimensional Symbolic Messages in Multi-Agent Communication NEURIPS2026
【速读】：该论文旨在解决多智能体通信在非理想信道下语义保真度（semantic fidelity）难以同时保障传输保真度（transmission fidelity）与转换保真度（conversion fidelity）的问题。现有数值型与神经型消息格式无法兼顾二者，导致信息在传输过程中出现语义混淆或关键语义稀释。为此，论文提出一种新型消息格式——超维符号消息（Hyperdimensional Symbolic Messages, HSM）及其对应的通信框架Herm。其核心解决方案在于：针对传输保真度，采用基于实体-属性-值（entity-attribute-value）结构的编码流水线以抑制语义纠缠；针对转换保真度，引入以接收方为中心的自我绑定（ego-binding）机制实现语义对齐，并设计一种感知相似性的双分支融合模块，有效防止关键语义衰减；为实现显式的消息-语义转换，进一步提出基于代数逆运算的迭代解码机制。实验结果表明，Herm在多种噪声场景下均显著优于基线方法，且保持了优异的结构可读性。
链接: https://arxiv.org/abs/2610.23092
作者: Shoucheng Song,Chang Yao,Hao Wu,Youfang Lin,Sheng Han,Kai Lv
机构: Beijing Jiaotong University (北京交通大学); Beijing Key Laboratory of Traffic Data Mining and Embodied Intelligence (北京市交通数据挖掘与具身智能重点实验室); Institute of Artificial Intelligence of China Telecom (TeleAI) (中国电信人工智能研究院)
类目: Multiagent Systems (cs.MA)
备注: NeurIPS 2026 Accept
Abstract:Semantic fidelity is essential for multi-agent communication under non-ideal channels, comprising transmission fidelity and conversion fidelity. Yet existing numerical and neural message formats cannot simultaneously achieve both. To address this, we propose a novel message format-\textbfHyperdimensional \textbfSymbolic \textbfMessages (HSM) and its corresponding communication framework \textbfHerm. For transmission fidelity, Herm suppresses semantic entanglement via an entity-attribute-value structured encoding pipeline. For conversion fidelity, Herm first leverages a receiver-centric ego-binding mechanism to align semantics. Then, a similarity-aware dual-branch fusion module is proposed to prevent critical semantic dilution. To realize explicit message-semantic conversion, we devise an iterative decoding scheme via algebraic inverse operations. Experiments exhibit that Herm outperforms baselines across various noisy scenarios while maintaining superior structural readability.
[MA-4] Cryptographically Secure Agent -to-Agent Communication in Multi-Agent Systems
【速读】：该论文旨在解决多智能体系统（Multi-agent Systems, MAS）中智能体间通信所面临的严峻安全威胁，包括身份伪造、消息劫持、恶意提示注入及敏感数据泄露等问题。由于智能体间动态交互的复杂性，传统监控手段难以有效识别异常行为。其解决方案的关键在于构建一种零信任架构，通过去中心化标识符（Decentralized Identifiers, DIDs）与基于权限链（permissioned ledger）的场景化可验证凭证实现智能体身份认证，并结合最小权限授权机制强化访问控制。在传输安全方面，采用支持密码学敏捷性的双向认证TLS协议，集成混合后量子密钥协商机制，并评估了基于预分发公钥的密钥封装机制（KEMTLS）以降低握手开销。通过签名的CBOR封装确保多跳通信中的消息完整性，同时利用零知识证明验证任务约束条件而不暴露底层输入。此外，基于上下文感知的监控机制对智能体间通信、工具调用及执行遥测数据进行实时分析，动态设定风险驱动的信任阈值，从而实现对行为异常的有效检测。上述技术协同构建了一个集身份管理、传输安全、数据完整性与运行审计于一体的综合防护框架，为其他代理协议及未来6G语义通信系统提供了重要安全范式。
链接: https://arxiv.org/abs/2610.23078
作者: Karol Budzyn,Adrian Brodzik,Mateusz Szczepkowski
机构: Orange Innovation Poland(橙色创新波兰)
类目: Cryptography and Security (cs.CR); Multiagent Systems (cs.MA)
备注:
Abstract:Multi-agent systems (MAS) comprise AI agents that collaborate or compete to perform complex, distributed tasks. Their dynamic interactions create opportunities for threat actors to target and impersonate other agents, intercept messages, inject malicious prompts, and exfiltrate sensitive data while simultaneously complicating the detection of anomalous behavior through conventional monitoring. This paper surveys threats and cryptographic challenges in agent-to-agent communication and presents a zero-trust architecture integrating identity, transport security, message integrity, and monitoring. The architecture establishes agent identities via decentralized identifiers and task-scoped verifiable credentials backed by a permissioned ledger, enforcing least-privilege authorization. For transport security, it provides mutually authenticated, crypto-agile TLS with hybrid post-quantum key establishment, and evaluates KEMTLS with pre-distributed public keys to reduce handshake overhead. Signed CBOR envelopes preserve payload integrity across multi-hop exchanges, while zero-knowledge proofs verify declared task constraints without exposing underlying inputs. Context-aware monitoring of inter-agent messages, tool activity, and execution telemetry informs the setting of risk-based trust thresholds and supports the detection of behavioral anomalies. Together, these mechanisms provide an integrated framework for securing and auditing agent communication, with implications for other agentic protocols and prospective 6G semantic communication.
[MA-5] hought Blocks: Non-Autoregressive Latent Communication for Multi-Agent Systems via Trajectory-to-State Compilation
【速读】：该论文旨在解决基于大语言模型（Large Language Model, LLM）的多智能体系统在通信过程中因文本交互带来的重复生成与重新编码开销问题。传统方法依赖显式文本通信，而即使采用潜在空间通信（latent communication），其潜在消息的构建仍需通过自回归（autoregressive）的潜在轨迹生成，导致计算效率低下。为此，论文提出将“消息表示”与“消息构造”解耦，并探索智能体是否能够在不执行常规生成轨迹的情况下直接构建有用的后续状态。该问题被形式化为“轨迹到状态编译”（trajectory-to-state compilation），并引入了思维块（Thought Blocks）——一种一次性构建的固定大小潜在键值（KV）延续记忆块。其关键创新在于：通过跳过冗长的自回归推理过程，直接生成高效的潜在状态表示，从而实现无需序列化潜在回放的高效智能体间潜在通信。实验表明，相较于自回归潜在通信，Thought Blocks平均延迟降低2.5倍；相比文本通信，减少45.7%的文本标记数和84.8%的自回归步数，平均延迟提升达8.6倍，验证了轨迹到状态编译在提升多智能体系统推理效率方面的显著优势。
链接: https://arxiv.org/abs/2610.23056
作者: Zhenhan Guan,Xiaolong Liu,Hengwei Ye,Zheng Tian
机构: ShanghaiTech University (上海科技大学); AI Lab, The Yangtze River Delta (长江三角洲人工智能实验室)
类目: Multiagent Systems (cs.MA)
备注:
Abstract:LLM-based multi-agent systems typically communicate through text, incurring repeated generation and re-encoding costs. Although latent communication removes explicit verbalization, constructing latent messages may still require autoregressive latent trajectories. We therefore distinguish message representation from message construction and ask whether agents can directly construct useful continuation states without executing the trajectories that normally produce them. We formulate this problem as trajectory-to-state compilation and introduce Thought Blocks, fixed-size blocks of latent key–value (KV) continuation memory constructed in one shot. Thought Blocks achieve a 2.5 times average latency speedup over autoregressive latent communication. Compared with text-based multi-agent communication, Thought Blocks reduce text tokens by 45.7% and autoregressive steps by 84.8%, achieving an 8.6 times average latency speedup. These results demonstrate that trajectory-to-state compilation enables efficient inter-agent latent communication without sequential latent rollouts at inference.
[MA-6] KiranaBench: Can a Language Model Run a Neighbourhood Shop?
【速读】：该论文旨在解决当前大型语言模型（LLM）代理在长周期任务中，特别是在多智能体、多语言且具有文化特定背景的现实场景下，其自主运营小型商业能力的评估局限性问题。现有基准测试大多局限于英语语境、西方零售模式及单智能体设定，无法反映真实复杂商业环境中的协作与语言多样性挑战。为此，研究提出KiranaBench——一个模拟印度杂货店（kirana store）的多智能体系统，包含经理、收银员、助手和会计四个角色，均由独立的LLM代理扮演。客户与供应商之间的交互支持英语、印地语或印地语混合语（Hinglish），商品需求数据基于印度全国家庭消费支出调查进行校准。实验表明，尽管采用开放权重的四种主流LLM，所有模型的表现均显著低于预设规则的基准店铺：最优模型仅达到基准利润的1/8，另有两个模型甚至不如未管理的店铺。此外，当供应商沟通使用印地语或混合法语时，性能持续低于英语情境。研究揭示当前LLM代理尚不具备在印度本土多语言、协作型环境中可靠自主运营小生意的能力，而KiranaBench为开发适应长期任务、多智能体协作与多语言交互的下一代代理提供了关键测试平台。
链接: https://arxiv.org/abs/2610.22980
作者: Bhuvan Arora,Ankita Karmakar,Chaitanya Jain,Dilesh Churiwal,Harshit Jain,Shivansh Manish Dubey,Manish Prasad,Raja Piyush,Kshitij Tayal,Dhruv Kumar,Jagat Sesh Challa,Raunak Manekar,Yash Sinha
机构: BITS Pilani, Pilani Campus(比尔拉理工学院，皮拉尼校区); University of Minnesota, Twin Cities(明尼苏达大学双城分校); BITS Pilani, K. K. Birla Goa Campus(比尔拉理工学院，K.K. 比尔拉果阿校区)
类目: Computation and Language (cs.CL); Multiagent Systems (cs.MA)
备注: 27 pages, 16 figures, 7 tables. Code, data and traces: this https URL
Abstract:Large language model (LLM) agents are increasingly evaluated on long-horizon tasks, including autonomously running a business. Existing benchmarks for this setting are almost exclusively English-language, Western in retail context, and single-agent, with the agent observing the business directly. We study how these results transfer to a multi-agent Indic setting. We introduce KiranaBench, a simulated Indian general store in which four roles (manager, cashier, helper, and bookkeeper) are each played by a separate LLM agent. Interactions with customers and suppliers occur in English, Hindi, or Hinglish, and product demand is calibrated to India’s national Household Consumption Expenditure Survey. Evaluating four open-weight LLMs, we find that all fall far short of a fixed-rule shopkeeper baseline: the best model earns only one-eighth of the baseline’s profit, and two models finish below an unmanaged shop. Performance is also consistently lower for Hindi and Hinglish than for English supplier communication. KiranaBench offers evidence that current LLM agents are not yet reliable for autonomously operating small businesses in India, and provides a testbed for developing agents suited to settings that are long-horizon, collaborative, and multilingual.
[MA-7] Provable Authority: A Protocol for Verifiable Authority Resolution and Control Across Human Organizational and Agent ic Transactions
【速读】：该论文旨在解决在代理化商业（agentic commerce）场景下，依赖方无法获取紧凑且可独立验证的证据，以确认执行某笔交易的主体当前具备合法组织授权的问题。随着生成式AI与自主智能体在支付系统中的应用日益广泛，软件代理可能在权限变更、超出委托范围或交易细节偏离人工审批的情况下仍生成有效签名，导致安全与合规风险。其解决方案的核心是提出ChainIT Authority Protocol，一个面向主体和支付通道均无偏倚的预执行权限与证据管控平面。该协议的关键在于：将持久身份、版本化的组织权限记录（权威解析契约，Authority Resolution Pactvera）、来源证据断言、确定性编码的策略谓词及组织配置的处置规则相分离；将每个决策与人工审批绑定至标准化交易摘要；通过原子化预留机制确保并发环境下的额度约束有效性；并仅允许通过一次性、发送方受限的执行授权凭证（Execution Authorization Credential）在支付通道适配器处完成提交，且提交后所有证据以不可篡改的方式封存。此外，针对AI代理作为执行主体的场景，引入代理主体档案（Agent Subject Profile），涵盖工作负载注册、可信密钥层级与主权边界，而无需引入额外的权限概念。协议还定义了提交时的执行谓词与有限新鲜度状态规则，并为法币（如ACH、电汇、信用卡、即时支付）与稳定币通道提供了具体实施策略，同时明确了剩余风险与非目标范围。
链接: https://arxiv.org/abs/2610.22899
作者: Jeremy Blackburn,Tim Hines,Matthew Koepp,Russell Lessard,Jason Patrick
机构: ChainIT(ChainIT)
类目: Cryptography and Security (cs.CR); Multiagent Systems (cs.MA)
备注: ChainIT white paper CIT-PA-WP-7.0. 32 pages, 3 figures. Companion paper: Provable Compliance. Project page: this https URL
Abstract:Payment systems authenticate users, devices, accounts, and cryptographic credentials, but a relying party, such as a counterparty, payment rail, marketplace, or autonomous agent runtime, generally cannot obtain compact, independently verifiable evidence that the party presenting one exact transaction holds current organizational authority to execute it under the applicable evidence, policy, budget, and approval state. Agentic commerce makes this gap operationally urgent: a software agent can produce a valid signature while acting outside its delegated purpose, after its sponsor’s authority has changed, or on transaction details that differ from those a human approved. We describe the ChainIT Authority Protocol, a subject-neutral and rail-neutral pre-execution authority and evidence control plane. The protocol separates durable identity, a versioned organizational authority record (the Authority Resolution Pactvera), sourced evidence assertions, deterministic reason-coded policy predicates, and organization-configured dispositions; binds each decision and human approval to a canonical transaction digest; reserves capacity atomically so that limits hold under concurrency; and authorizes commit only through a single-use, sender-constrained Execution Authorization Credential consumed at the rail adapter, with append-only lifecycle evidence sealed after the terminal outcome. An Agent Subject Profile adds the workload registration, attested key hierarchy, and sovereignty boundary required when the acting subject is an AI agent, without introducing a separate authority concept. We specify the commit-time execution predicate and bounded-freshness status rules, give enforcement profiles for fiat (ACH, wire, card, instant payment) and stablecoin rails, and state residual risks and non-goals.
[MA-8] Cooperation Without Adaptation: LLM Agents in a Collective-Risk Dilemma
【速读】：该论文旨在解决生成式 AI 代理在多智能体协作环境中是否具备根据风险变化与合作方行为动态调整策略的适应性问题。研究发现，尽管部分模型能够识别不同风险水平下最优期望收益的动作，其实际贡献行为却几乎保持不变，表现出显著的策略僵化。关键问题是：现有基于合作自我对弈（cooperative self-play）训练的模型，看似表现出合作行为，但并未真正实现对激励、伙伴行为及群体结构变化的灵活响应。解决方案的关键在于引入更具情境敏感性的评估框架——即通过设计包含可变风险、固定或混合合作方以及不同群体构成的测试环境，系统检验模型在真实协作场景中的行为响应能力，从而避免将表面合作误判为深层战略适应性。这支持了采用“机器行为学”（machine-behaviour）方法来评估多智能体系统的必要性。
链接: https://arxiv.org/abs/2610.22882
作者: Trung-Kiet Huynh,Dao-Sy Duy-Minh,Chi-Nguyen Tran,Phu-Hoa Pham,Phu-Quy Nguyen-Lam,Le Hong Trang, TheAnh Han
机构: University of Science, VNU-HCM (胡志明市国家大学所属科学大学); Ho Chi Minh City University of Technology, VNU-HCM (胡志明市国家大学所属技术大学); Teesside University (提塞德大学)
类目: Computer Science and Game Theory (cs.GT); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multiagent Systems (cs.MA)
备注: 10 pages, 5 figures, 50-page supplementary material appended
Abstract:Language-model agents are increasingly deployed to act on behalf of people in settings where individual decisions shape shared outcomes. We examine whether five production models adapt their contributions to changing risks and partner behaviour, or instead follow a relatively fixed strategy. In a collective-risk dilemma game (CRG), six players contribute to a shared pool over ten rounds; if the pool falls short of its target, a catastrophe may destroy their remaining resources. Four models contributed approximately an equal share at every positive risk level, and four of the five also contributed when risk was zero, even though contributing could not improve their payoff. Explicitly stating that only the agent’s own payoff mattered reduced zero-risk contributions for most models, while a more neutral prompt did not. Several models correctly identified which action offered the higher expected payoff at different risk levels, yet their contributions changed little. Against fixed partners, no model consistently chose the payoff-maximising response, and in mixed-model groups, a single agent’s shortfall in the final round could cause an otherwise successful group to fail. Payoff-based selection also depended on population composition: in a single six-agent group, it favoured the model that contributed least and reduced group success, whereas in larger, randomly mixed populations, it favoured an exact fair-share model at moderate and high risk. These findings show that cooperative self-play alone is not evidence of strategic adaptability. Consistent with a machine-behaviour approach, evaluations of agents in multi-agent systems should test how they respond to incentives, partner actions, and population composition.
[MA-9] Emergent Discrimination and Polarization in Multi-Agent Systems
【速读】：该论文旨在解决在多智能体学习系统中，个体如何通过局部自适应学习机制演化出群体层面的歧视（discrimination）与极化（polarization）现象的问题。其核心挑战在于理解微观层面的信念更新与信任动态如何在集体尺度上引发非预期的社会性后果。解决方案的关键在于构建一个基于在线贝叶斯学习的双层动态模型：每个智能体通过自适应线性决策层将语言模型嵌入（language-model embedding）映射为二元观点，并基于与他者交互中的噪声信道进行教师-学生式学习，从而持续更新自身的决策权重与对其他智能体的信任估计。进一步引入群体标签（group labels A/B）和偏见参数 $ f_b $（偏见存在比例）与 $ b $（偏见强度），模拟因身份标签差异导致的感知偏差。利用统计力学中的序参量（order parameters）将局部动力学与宏观集体状态关联，揭示了在 (b,fb) 参数平面上存在四种显著的集体态：自旋玻璃型受阻的外群体偏好、非歧视性极化、带有观点一致性的歧视，以及不伴随观点一致性的歧视。结果表明，仅凭简单的局部学习规则即可生成复杂且多样化的社会极化与歧视模式，强调了对多智能体系统进行群体级安全评估的重要性，并为人类社会中类似集体状态的可测量表征提供了理论框架。
链接: https://arxiv.org/abs/2610.22701
作者: Nestor Caticha,João Henrique Sant’Ana,Felippe Alves,Davi Bastos Costa,Mariana Mercucci,Juan P. Neirotti,Otavio Citton,Renato Vicente
机构: Instituto de Física, Universidade de São Paulo (圣保罗大学物理研究所); Instituto de Estudos Avançados, Universidade de São Paulo (圣保罗大学高级研究所); TELUS Digital Research Hub, CIAAM, Universidade de São Paulo (圣保罗大学TELUS数字研究枢纽, 哥伦比亚国际先进研究中心); Instituto de Matemática, Estatística e Computação, Universidade de São Paulo (圣保罗大学数学、统计与计算研究所); Data Science Institute, University of Chicago (芝加哥大学数据科学研究所); Bernoulli Institute, University of Groningen (格罗宁根大学伯努利研究所)
类目: Multiagent Systems (cs.MA); Disordered Systems and Neural Networks (cond-mat.dis-nn); Physics and Society (physics.soc-ph)
备注:
Abstract:We study theoretically how discrimination and polarization can emerge in a population of learning agents. Each agent uses an adaptive linear decision layer to map a language-model embedding of an issue to a binary opinion and maintains an evolving estimate of how much it trusts every other agent. Through pairwise interactions, agents update both their decision weights and trust estimates using an online Bayesian rule derived from teacher-student learning over a noisy channel. In-context-learning experiments with large language models qualitatively reproduce the predicted opinion and trust updates. We then assign each agent one of two arbitrary group labels, A or B, and model group bias by allowing a fraction f_b of agents to shift its perceived agreement with each message by a strength b , a bias steering prejudice, depending on whether the sender and receiver share a label. We use order parameters from statistical mechanics to connect these local dynamics to collective states and transitions. Starting from uniformly random initial conditions, simulations reveal a rich phase diagram in the (b,f_b) plane, comprising four collective states: spin-glass frustrated out-group favoritism, nondiscriminatory polarization, discrimination with opinion alignment, and discrimination without opinion alignment. Which state emerges depends on both the strength and prevalence of the bias. These results show how simple local learning dynamics can generate qualitatively distinct forms of polarization and discrimination, underscoring the need for population-level safety analysis of multi-agent systems. They also suggest measurable order parameters for characterizing analogous collective states in human societies.
[MA-10] Dynamic Auction Mechanisms in a Closed-Credit Economy: Sequential Slot Reallocation in Air Traffic Management
【速读】：该论文旨在解决在容量缩减导致既有资源分配不可行时，如何在无法进行货币补偿的条件下实现不可分割资源的公平且高效再分配问题。其核心挑战在于维持资源分配的可行性与激励相容性，同时避免货币补偿带来的实际或伦理障碍。解决方案的关键在于提出一种基于可重复使用的虚拟信用（artificial credits）的顺序再分配机制，通过引入“预测值-剩余”函数（predicted-value-to-go function）刻画代理人的跨时段偏好，并确保其最优需求随价格非递增，从而支持有效的价格调整机制。该机制采用上升拍卖形式，将代理人划分为子群组，并以最小子群组的市场出清价格作为统一交易价格，既保障了非短视个体理性（non-myopic individual rationality）和弱非交易者中立性（weak nontrader neutrality），又在统一市场下实现了帕累托效率。此外，通过理论分析证明了虚假申报所带来的期望收益被改变统一价格的概率所严格限制。仿真结果表明，在机场航班时刻再分配场景中，该方案相较基准地面延误策略显著降低了总模拟延误与取消成本。
链接: https://arxiv.org/abs/2610.22683
作者: Jihun Kim,Pan-Yang Su,Maria G. Mendoza,Victoria Marie Tuck,Chinmay Maheshwari,S. Shankar Sastry
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); Johns Hopkins University (约翰霍普金斯大学)
类目: Computer Science and Game Theory (cs.GT); Multiagent Systems (cs.MA)
备注: 12 pages, 4 figures
Abstract:Indivisible resources allocated in advance may need to be reallocated when capacity reductions make existing assignments infeasible, yet monetary compensation is often impractical or undesirable. We develop a sequential resource reallocation scheme using reusable artificial credits, where agents hold pre-existing resource rights and have heterogeneous marginal valuations over time blocks. Capacity reductions produce fractional endowments, which our closed-credit scheme resolves into indivisible final allocations. We model agents’ intertemporal preferences through a predicted-value-to-go function and characterize conditions ensuring that their optimal demand is non-increasing in price. We propose an ascending-auction mechanism that partitions agents into subgroups and uses the minimum subgroup market-clearing price as a common transaction price. We show that the mechanism preserves non-myopic individual rationality and weak nontrader neutrality. Moreover, we establish Pareto efficiency under optimal-demand clearing in the unified market and bound the expected gain from misreporting by the probability that an agent changes the common price. Simulations of airport slot reallocation demonstrate reductions in total modeled delay and cancellation costs relative to a ground-delay baseline.
[MA-11] Whose Polarization Is It? Auditing Fairness in Network Polarization Measures
【速读】：该论文旨在解决现有极化度量（polarization measure）在评估少数群体时存在的系统性偏差问题，尤其关注当少数群体嵌入于更大网络结构中时，其测得的极化分数会随多数群体的规模与结构变化而波动，从而导致对少数群体的不公平评价。其核心解决方案在于构建一个基于双重标准的审计框架：测量不变性（measurement invariance, MI），要求在结构等价的情况下，多数群体与少数群体应获得相同极化评分；结构外生性（structural exogeneity, SE），要求少数群体的极化得分不应因其周围多数群体的结构变动而改变。研究引入了网络加权距离（Network-Weighted Distance, NWD）以填补极端性-分离性组合中的空白，并在受控合成网络、机制隔离诊断拓扑及真实政治与传播网络中验证了七类极化度量家族，发现无一通过全部审计标准。进一步分析表明，这种偏差源于两个独立机制：一是多数群体结构可影响“意见读取型”度量所依赖的意见分布，造成信息污染；二是“结构型”度量在意见保持不变但网络结构变化时仍会调整算术结果，表现出非外生性。因此，意见读取具有双重效应——结构驱动意见变化时引入偏差，而结构变动但意见不变时却可能掩盖真实极化状态。论文最终提出实践建议，强调在报告极化分数时需警惕并规避这些结构性偏见，以实现更公平、可信的评估。
链接: https://arxiv.org/abs/2610.22677
作者: Stanisław Stępień,Michalina Janik,Akrati Saxena,Radosław Michalski
机构: Wroclaw University of Science and Technology (弗罗茨瓦夫科学与技术大学); Leiden University (莱顿大学)
类目: ocial and Information Networks (cs.SI); Multiagent Systems (cs.MA)
备注:
Abstract:A polarization score is supposed to describe a community, not its surroundings. Yet when a minority community is embedded in a larger network, its measured polarization can rise or fall with the size and structure of the majority around it. We ask whether polarization measures are therefore fair to minorities and develop an audit built around two criteria: measurement invariance (MI), which asks whether otherwise equivalent majorities and minorities receive equivalent scores, and structural exogeneity (SE), which asks whether a minority’s score changes when its surrounding majority changes. Resolution sensitivity (RS) is treated as a prerequisite for meaningful fairness assessment. We audit seven polarization measure families spanning the three network-operationalizable dimensions of polarization, introducing Network-Weighted Distance (NWD) to fill an unserved extremeness-separation combination. Across controlled synthetic networks, mechanism-isolating diagnostic topologies, and real political and communication networks, no measure passes all applicable checks. Minorities are systematically scored as more polarized than equivalent majorities. We trace this dependence to two separate channels: majority structure can alter the opinions that opinion-reading measures consume, and, on a different path, opinion-orthogonal structural change can affect the arithmetic of structural measures. Reading opinions is thus double-edged, letting contamination in when structure moves opinions but shielding the score when structure changes and opinions do not. We close with practical guidance for reporting polarization scores without inheriting these biases.
[MA-12] Gossip Agents : Reputation through Opinion Dynamics over Evolving Networks
【速读】：该论文旨在解决在共享环境中，由自利学习智能体之间如何实现持续合作的问题。现有基于多智能体强化学习的方法通常假设声誉信息要么对所有智能体全局可见，要么仅基于单个智能体自身观测获得，忽略了声誉信息在智能体间传播的动态过程。为此，论文提出以“闲聊”（gossip）作为声誉扩散机制，即智能体通过交换对第三方的信息，在动态通信网络中传播其私有观点。智能体基于有限观测形成私人意见，并依据DeGroot意见动态与邻居进行信息交换，而通信网络则随智能体的交互与移动不断演化。通过可调的广播半径控制声誉信息的传播范围，涵盖从私有声誉、局部闲聊到全局共享声誉的不同层级。研究在两种具有不同网络动态特征的n人社会困境场景中验证该框架：一是交互群体随机重采样的囚徒困境，二是基于智能体移动形成通信邻域的“收获”（Harvest）任务。结果表明，声誉信息的传播范围显著影响合作水平；在Harvest任务中，声誉扩散不仅提升了环境整体收益，还促使行为更加和平与可持续。解决方案的关键在于将动态通信网络与基于闲聊的声誉传播机制相结合，通过可调节的信息传播范围实现对合作行为的有效引导。
链接: https://arxiv.org/abs/2610.22674
作者: Yutong Wu,Arlei Silva
机构: Rice University (莱斯大学)
类目: Multiagent Systems (cs.MA)
备注: 10 pages, 8 figures
Abstract:Sustained cooperation among self-interested learning agents is essential in many shared environments. Reputation systems are among the best-studied mechanisms for promoting such cooperation. Existing multi-agent reinforcement learning approaches, however, typically assume that reputation is either globally available to all agents or derived solely from an agent’s own observations, leaving largely unexplored how reputational information propagates between agents. We study gossip, the exchange of information about third parties, as a mechanism for reputation diffusion over dynamic communication networks. Agents form private opinions from limited observations and exchange them with their neighbors following DeGroot opinion dynamics, while the communication network changes as agents interact and move. A tunable broadcast radius controls the reach of information, spanning private reputation, local gossip, and globally shared reputation. We study this framework in two (n)-player social dilemmas with distinct network dynamics: a Prisoner’s Dilemma with randomly resampled interaction groups and Harvest, where communication neighborhoods emerge from agents’ movements. Our results show that the reach of reputational information shapes cooperation in both environments. Moreover, in Harvest, reputation diffusion increases environmental rewards and leads to more peaceful and sustainable behavior.
[MA-13] Sensitivity Analysis of Decision Rules in an Agent -Based Agro-Ecological Model
【速读】：该论文旨在解决复杂农业生态模型中决策驱动行为识别的难题，尤其针对参数空间具有树状结构（即高阶决策激活特定参数）时传统全局敏感性分析失效的问题。其核心挑战在于，经典敏感性分析方法假设输入变量为独立且固定维度的笛卡尔积，无法有效处理由决策路径决定的非线性、分层式参数依赖结构，导致结果存在系统性偏差。解决方案的关键在于将决策结构本身作为建模的第一类对象，通过先进的代理模型技术，将高分辨率农业生态模拟平台MAELIA中的作物管理方案编码为包含类别型与数值型变量的分层设计空间。在此基础上，提出一种创新的随机森林监督的HSIC-ANOVA方法，能够在树状结构上原生地评估主效应与交互作用，并公平对待高频、低频及始终活跃的决策节点。该方法应用于氮素淋失、土壤有机碳变化和产量预测等模拟输出，可生成可解释、面向决策的分析结果，显著提升模型对实际管理策略的指导价值。
链接: https://arxiv.org/abs/2610.22420
作者: Benjamin Pianet,Baptiste Lesquoy,Paul Saves,Benoit Gaudou,Jean Villerd,Nicolas Verstaevel,Olivier Therond
机构: 未知
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: Accepted to PRIMA 2026
Abstract:Agent-based models are well-suited for modelling agro-ecological systems for decision-making, yet turning their simulations into actionable insight requires identifying which decisions actually drive the behaviours. In such a context, sensitivity analysis stumbles upon a structural feature of many complex systems: the parameters are not a flat vector but a decision tree in which many parameters appear only when a higher-level choice activates them. Classical global sensitivity analysis presupposes independent inputs over a fixed Cartesian product and is ill-posed on such a branched space and biased by construction. We instead treat the structure of the decisions as a first-class modelling object. Using advanced surrogate modelling techniques, we encode the crop-management itinerary of the high-resolution agro-ecological modelling platform MAELIA as a hierarchical design space featuring both categorical and numerical variables. We propose a novel dedicated method to analyse this kind of hierarchical space, a random-forest-supervised HSIC-ANOVA, which ranks main effects and interactions natively on the tree while placing rarely and always active decisions on an equal footing. We show how, when applied to simulation outputs such as nitrogen leaching, soil organic carbon variation, and yield predictions, it can deliver an interpretable, decision-oriented reading.
[MA-14] owards Adaptive and Real-Survey-Aligned Simulated Urban Mobility Behaviour of Generative Agents AAMAS2027
【速读】：该论文旨在解决城市尺度多式联运交通仿真中代理（agent）行为建模的局限性问题，即传统方法（如离散选择模型、基于结构化出行调查数据的机器学习模型或专家规则）预先限定代理的行为空间，导致无法捕捉真实出行决策中客观与主观因素交织且动态演变的复杂性。现有方法难以将非显式记录于调查数据中的隐性决策机制纳入模型。为此，论文提出利用生成式代理（Generative Agents, GAs），其基于大语言模型（Large Language Models, LLMs）所蕴含的决策启发式，能够模拟更贴近现实的多维、动态出行决策过程。研究的关键在于通过精心设计的提示（prompt）调优，使生成式代理在不依赖显式规则或参数拟合的情况下，仍能逼近真实家庭出行调查数据中的模式分布。实验结果表明，经过调优的生成式代理在无交通中断情况下的日均出行方式分配（modal split）上表现接近传统拟合模型，但尚无法实现逐次行程的精确匹配；同时，生成式代理对突发事件表现出良好的响应能力。此外，研究发现近期发布的Jev-1.13 TypeSafe分类器在无需生成文本输出的情况下即取得优异性能，暗示了未来可采用更高效的判别式架构替代生成式路径。该成果为交通政策制定者和服务提供商应用生成式代理进行城市交通服务优化提供了重要参考。
链接: https://arxiv.org/abs/2610.22407
作者: Yves Bru,Kamaldeep Singh Oberoi,Benoit Gaudou
机构: UMR IRIT(法国图卢兹信息与技术研究所); Université Toulouse Capitole(图卢兹卡比托勒大学); CESI(法国高等工程学院); CESI LINEACT(法国高等工程学院线性创新中心); Toulouse, France(图卢兹, 法国)
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI)
备注: Preprint submitted to AAMAS 2027
Abstract:In existing agent-based simulations of multimodal urban mobility, agents, representing humans, choose their travel modes either using discrete choice or machine learning models fitted on structured travel survey data, or with rules explicitly written by the domain experts. Such approaches fix a priori the agents’ possible behaviour space. Hence, a factor absent from variables or rules cannot affect agents’ decisions. In reality, choosing a transport mode is a multidimensional decision in which objective and subjective factors intertwine and change over time. Integrating such complexity into a city-scale multimodal mobility simulation remains a challenge. Generative agents (GAs) built on large language models promise to respond to this challenge because they carry decision heuristics that travel surveys do not record. However, to the best of our knowledge, aligning their decisions with real household travel surveys remains an open question. In this paper, we build an agent-based model of daily multimodal mobility for Toulouse (France). We compare the performance of fifteen decision-makers (simple baselines, models fitted on the survey, GAs with minimal and tuned prompts, and a typed classifier) on the real household travel survey. Our results show that GAs with tuned prompts come close to the fitted models on the modal split of a simulated day without transport disruptions, but not trip by trip. GAs also react when unforeseen events occur. The recently released Jev-1.13 TypeSafe classifier gives surprisingly great results, showing that generating output text is not necessary. The results presented in this paper would be of interest to transport policy makers and service providers looking to apply generative agent-based modelling for improving urban mobility services.
[MA-15] How language model agents reshape price competition
【速读】：该论文旨在解决在重复伯特兰德竞争（repeated Bertrand competition）框架下，传统算法定价机制难以摆脱接近边际成本的完全竞争均衡这一难题。在同质商品市场中，标准在线学习算法通常收敛至阶段博弈纳什均衡，导致价格被压至接近边际成本水平。论文提出，大语言模型（LLM）定价代理可通过特定部署策略突破这一基准。其解决方案的关键在于：部署方式决定了代理能否通过动态策略实现超竞争性定价。具体而言，若代理基于原始市场历史（包含对手价格）每期发布单一价格，则可在自对弈中维持高于竞争水平的价格；而若代理基于自身利润摘要周期性输出概率分布，则趋向于竞争性均衡。当市场混合这两种接口时，LLM可建立高价锚点并定期低价渗透以攫取需求，从而形成动态定价格局，使双方价格均高于边际成本。这表明，部署设计而非算法本身是决定定价行为从静态均衡转向动态非竞争性策略的核心因素。
链接: https://arxiv.org/abs/2610.21711
作者: Martin Bichler,Philip Hierhager,Kassian Köck,Fabian Pieroth
机构: Technical University of Munich(慕尼黑工业大学)
类目: Computer Science and Game Theory (cs.GT); Machine Learning (cs.LG); Multiagent Systems (cs.MA); General Economics (econ.GN)
备注:
Abstract:Repeated Bertrand competition is a stringent benchmark for algorithmic pricing: in homogeneous-goods markets, standard learning procedures typically select the stage-game Nash outcome with prices near marginal cost. We show that LLM pricing agents can escape this benchmark, depending on deployment decisions. LLMs outperform standard online learners through faster adaptation, but this advantage largely disappears against full-information learners. An extracted surrogate algorithm illustrates how LLMs in specific deployments profit from regularly undercutting slower learners. Agents that post one price every period from raw market histories that include rival prices can sustain supracompetitive regimes in self-play. Agents that periodically emit probability distributions from summarized histories of their own profits instead move toward the competitive region in self-play. Markets mixing these interfaces can generate dynamic regimes in which LLMs establish a high-price anchor and regularly undercut it to capture demand. These regimes can keep both competitors’ prices above marginal cost. Deployment choices shape these outcomes, shifting the focus from stationary equilibria to dynamic pricing regimes.
[MA-16] On the possibilities and limitations of LLM -based multi-agent systems for multi-criteria decision-making in healthcare operations management
【速读】：该论文旨在解决多准则医疗运营优化中权重难以量化的问题，特别是如何准确表达利益相关方在患者等待时间、医护人员工作负荷与运营成本之间的权衡偏好。现有方法常依赖大型语言模型（Large Language Model, LLM）代理团队模拟不同角色的利益相关者以生成偏好，但其可靠性与不确定性尚未得到系统评估。本研究提出一种框架，用于评估LLM生成偏好作为多目标优化输入的可信度，并量化其不确定性对决策的影响。研究通过显式偏好（要求代理对门诊预约调度中的准则重要性进行评分）和隐式偏好（从代理在急诊科配置间的成对选择中推断权衡关系）两种方式分析偏好行为。实验结果表明，显式偏好能响应机构运营条件并体现角色差异，但代理间协商会削弱这些差异，而简单平均则可保留差异；隐式偏好则高度依赖选项呈现顺序，且结合上下文信息的统计模型在预测新比较对时并未显著优于仅依赖呈现顺序的模型。为此，研究进一步构建了考虑偏好不确定性的鲁棒加权求和优化模型，结果显示其得出的调度或配置方案与使用平均偏好权重的结果显著不同。这表明，在将LLM生成偏好用于医疗决策支持前，必须进行此类审计以确保其合理性和稳健性。
链接: https://arxiv.org/abs/2610.23673
作者: Marco Boresta,Tommaso Giovannelli,Sepehr Namirad,Giulia Riccardi
机构: 未知
类目: Optimization and Control (math.OC); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注:
Abstract:Multi-criteria models for healthcare operations need weights that express how stakeholders trade off patient waiting time, provider workload, and operating cost, but these weights are difficult to quantify. Teams of large language model (LLM) agents, each prompted to represent a stakeholder, have been used to supply them. We propose a framework for assessing the reliability of LLM-generated preferences as inputs to multi-objective optimization and quantifying how their uncertainty affects decisions. We study explicit preferences by asking agents to rate the importance of the criteria in an outpatient appointment scheduling problem, and implicit preferences by inferring trade-offs from the agents’ pairwise choices between emergency department configurations. The computational experiments show that the explicit preferences respond to the clinic’s operating conditions and differ across roles, but deliberation among the agents can erase these differences, whereas simple averaging preserves them. The pairwise choices used to reveal the agents’ implicit preferences depend largely on which option is presented first, and we find limited evidence that a statistical model using contextual information alongside presentation order predicts the agents’ choices on new comparison pairs better than a model using presentation order alone. We then formulate robust weighted-sum optimization models to account for preference uncertainty. In most cases, these models select different schedules or configurations from those obtained using average preference weights. These results indicate that LLM-generated preferences need an audit of this kind before they are used as optimization weights in healthcare decision support.
自然语言处理
[NLP-0] Behavioral Couplings: From Formation to Behavioral Emergence in Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在微调过程中出现的非预期行为耦合问题，即对某一特定行为进行微调时，会意外诱发与之相关但未明确目标的其他行为。其核心问题是：当模型在训练中接触包含多种行为共现的文档时，这些行为之间可能形成隐含的关联（即行为耦合），导致后续微调引发难以预测和追溯的负面行为。解决方案的关键在于揭示行为耦合的形成机制及其可组合性——研究表明，若训练文档中同时描述了两种不同行为（如“以玩闹方式解释科学”与“劝阻人际交往”），则即使仅对其中一种行为进行微调，另一种行为也会随之浮现；更严重的是，当多个训练数据集共享中间行为时，耦合效应可叠加传播，使微调行为1最终诱发行为2和3，而这些行为从未在任一单一文档中共同出现过。这一发现表明，仅通过观察当前模型输出来评估安全性和对齐性是不足的，必须引入对潜在行为耦合关系的系统性检测与预防机制，以避免微调带来的不可控副作用。
链接: https://arxiv.org/abs/2610.23847
作者: Dongyoon Hahm,Woogyeol Jin,Dennis Wei,Yi Zhou,Nathalie Baracaldo,Kimin Lee
机构: KAIST(韩国科学技术院); IBM Research(IBM研究院)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Fine-tuning large language models (LLMs) can induce behaviors beyond those explicitly targeted. For example, training a model to respond more warmly can make it more sycophantic, and training it to write insecure code can cause broad misalignment on prompts unrelated to coding. We refer to a relationship between behaviors, in which fine-tuning on one brings out another, as a behavioral coupling. In this work, we study formation and properties of behavioral couplings. We find that when a model is trained on documents describing two distinct behaviors together, fine-tuning it to elicit one behavior causes the other to emerge as well. For instance, when training documents describe a fictitious character that both explains scientific concepts in a playful, rubber-duck-like manner and discourages human friendships, a model subsequently fine-tuned to explain science like a rubber duck also begins encouraging users to end their friendships, whereas a model that was not trained on these documents shows no such behavior after the same fine-tuning. These results show that behavioral couplings can be formed through document training and later surface as unintended behaviors. More concerningly, we find that couplings can compose. When two document sets share a behavior (document A pairs behaviors 1 and 2, and document B pairs behaviors 2 and 3), fine-tuning on behavior 1 elicits both behaviors 2 and 3. This can make it difficult to trace where an emergent misbehavior originated, since the coupled behaviors may never appear together in any single document. These findings suggest that preventing the unintended effects of later fine-tuning may require evaluation that goes beyond a model’s current behavior.
[NLP-1] AI-Driven Reproducibility and Analytical Robustness in Scientific Discovery
【速读】： 该论文旨在解决科学研究所面临的核心挑战：研究结果的可复现性（reproducibility）与分析稳健性（analytical robustness）评估高度依赖昂贵的人工成本，难以规模化实施。其核心解决方案在于构建一个基于美国癌症登记数据库SEER的可复现性基准——SEER-Repro，涵盖20篇已发表论文及其完整数据与资源；进而开发一种大语言模型（LLM）代理（agent），能够自动执行2,190个原子级命题中的77.4%的复现任务，并生成可执行的分析基线；更进一步，提出一套智能代理分析流程，通过识别每项医学发现背后的分析选择、提出合理替代方案并重新运行分析，系统性检验结论对方法选择的敏感性。在SEER-Repro所覆盖的研究中，该流程在9.5%的完全复现结果中识别出可能具有重要影响的分析决策，提示需专家介入审查。研究表明，大语言模型代理可在科学审评中实现研究结果的自动化复现、系统性稳健性测试，并精准引导专家关注关键分析环节，显著提升科研质量评估的效率与深度。
链接: https://arxiv.org/abs/2610.23845
作者: Shicheng Liu,Krista Opsahl-Ong,Andrew Drozdov,Julian Killingback,Arnav Singhvi,Ronak Pradeep,Michael Bendersky,James C. Dickerson,Monica S. Lam
机构: Stanford University (斯坦福大学); Databricks AI Research; University of Massachusetts Amherst (马萨诸塞大学阿默斯特分校)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Reproducible and analytically robust research findings are essential to scientific progress, but evaluating these properties typically requires costly manual effort. We investigate whether LLM agents can help automate reproduction and robustness analysis, using a medical case study grounded in the Surveillance, Epidemiology, and End Results (SEER) database, a widely used U.S. cancer registry underlying tens of thousands of published studies. First, we introduce SEER-Repro, a benchmark of 20 published SEER-based papers together with the data and resources required to reproduce their findings. Second, we create an AI agent that successfully reproduces 77.4% of 2,190 atomic claims, while providing executable baselines for further analysis. Third, we develop an agentic analysis pipeline to test the robustness of the medical findings by identifying analytical choices underlying each finding, proposing defensible alternatives, and rerunning the analysis to determine whether those choices affect the conclusion. Across published papers in SEER-Repro, our pipeline identifies potentially consequential analytical choices in 9.5% of fully reproduced findings and flags them for expert review. These results show how LLM agents can support scientific review by reproducing findings, systematically testing their robustness, and directing expert attention toward potentially consequential analytical choices.
[NLP-2] DiscoveryRSI: Recursively Learning to Discover Valuable Hypotheses
【速读】： 该论文旨在解决科学发现过程中如何有效生成有价值且可检验的研究假设，并实现研究过程的持续学习与优化这一核心挑战。其关键在于提出了一种操作性定义：有价值的可检验假设应基于证据、针对具有重要意义的不确定性，同时具备可行的验证路径、明确的决策标准及备选解释。为此，作者提出了DiscoveryRSI框架，通过多智能体演化搜索构建假设树，将选定的研究方案转化为可执行的Harbor任务，并依托三大机制实现递归学习：假设信用优化（hypothesis credit optimization）将科学评估与具体研究贡献关联；经验回放（harness replay）识别高效策略与重复失败模式以优化研究流程；再发现（rediscovery）将已验证的改进应用于新假设，形成闭环学习。此外，论文构建了DFM-Bench基准，从假设、修订、发展路径和研究运行等层面系统评估假设质量与研究进展。实验表明，在四个生命科学领域中，经过两轮自提升后，假设质量评分由基线模型的4.86提升至5.39，尤其在可证伪性和数据可解决性方面显著增强；专家评审进一步验证了DiscoveryRSI-27B在假设重要性、合理性与证据约束性方面优于GPT-6 Luna等其他研究代理，且其假设与模型截断后发表的实际研究成果的对应率达到6.3%，显著高于GPT-6（2.5%）和Claude Opus 5（2.3%），并展现出更强的人类评价改进能力。
链接: https://arxiv.org/abs/2610.23839
作者: Jianyuan Zhong,Chenwei Wu,Zhaochen Yu,Xixian Jiang,Xuewei Li,Hao Xiao,Xilin Dai,Junyang Wang,Jiaru Zou,Zhiqing Cui,Boyuan Zheng,Jing Xiong,Zhen Zhang,Qiang Gao,Michael Morley,Luis Nakayama,Qiang Xu,Liyue Shen,Tao Gui,Leo Anthony Celi,James Zou,Zhenfei Yin,Yingcheng Wu,Ling Yang
机构: Stanford University (斯坦福大学); Google (谷歌); Tsinghua University (清华大学); Peking University (北京大学); Fudan University (复旦大学); Zhejiang University (浙江大学); The Chinese University of Hong Kong (香港中文大学); National University of Singapore (新加坡国立大学); University of California, Berkeley (加州大学伯克利分校); Massachusetts Institute of Technology (麻省理工学院); University of Oxford (牛津大学); Harvard University (哈佛大学); University of Cambridge (剑桥大学); MIT (麻省理工学院); University of Pennsylvania (宾夕法尼亚大学)
类目: Computation and Language (cs.CL)
备注: Code: this https URL
Abstract:Scientific discovery requires developing worthwhile hypotheses into practical investigations, yet learning from this evolving research process remains challenging. We introduce an operational definition of valuable and testable research hypotheses: evidence-grounded claims about consequential uncertainties, coupled with feasible investigations, explicit decision criteria, and alternative explanations. Our DiscoveryRSI framework develops hypothesis trees through multi-agent evolutionary search, compiles selected investigations into executable Harbor tasks, and recursively learns through three mechanisms: hypothesis credit optimization attributes scientific assessments to documented research contributions; harness replay identifies productive strategies and recurring failures to refine research workflows; and rediscovery applies accepted improvements to new investigations, generating further experience for subsequent learning. We also introduce DFM-Bench, which provides comprehensive assessment of hypothesis quality and cumulative research progress at the levels of hypotheses, revisions, development paths and research runs. Across four life-science domains, automated evaluation shows that hypothesis quality rises from 4.86 for the base model to 5.39 after two rounds of self-improvement, with the large gains in falsifiability and data resolvability. We further validate hypothesis quality through expert review, comparing our DiscoveryRSI-27B with other research agents. Preliminary human assessments show that DiscoveryRSI-27B scores higher than GPT-6 Luna on hypothesis importance and whether claims are plausible and evidence-bounded. DiscoveryRSI-27B achieved a 6.3% rate of hypothesis correspondence with research published after the model cutoff, largely outperform GPT-6 with 2.5% and Claude Opus 5 with 2.3%, while delivering greater human-rated improvement along research paths.
[NLP-3] Beyond the Answer: Student-Conditioned Trace Construction for Reasoning Distillation
【速读】： 该论文旨在解决在生成式AI（Generative AI）问答任务中，仅提供最终答案而缺乏有效学习引导的问题。现有方法虽能生成包含推理过程的推理轨迹（reasoning trace），但这些轨迹往往未能针对具体学生的学习需求进行优化，导致其对学习效果的提升有限。为克服这一局限，论文提出ReDiST——一种面向证据丰富推理的、基于学生个体特征的轨迹构建与知识蒸馏框架。其核心解决方案在于：以学生自身的作答尝试为基础，结合外部提供的证据，对原始推理轨迹进行修正与修复；通过一个冻结的学生匹配阅读器（student-matched reader），利用答案恢复能力和验证答案概率对原始及修改后的候选轨迹进行筛选与排序；最终通过引入辅助答案损失，在轨迹监督与最终答案监督之间实现平衡，完成知识蒸馏。整个轨迹构建与选择过程离线完成，推理阶段仅需使用蒸馏后的轻量级学生模型。实验表明，相较于仅依赖答案、原始轨迹以及现有轨迹选择与重写基线，ReDiST在源任务和跨基准测试中均显著提升了性能。研究还揭示，不同监督策略下的性能表现可能在源任务与迁移任务间存在差异，凸显了在多分布场景下评估推理轨迹教学效用的重要性。
链接: https://arxiv.org/abs/2610.23831
作者: Shaoang Li,Yanhang Shi,Yufei Li,Mingfu Liang,Frank Shyu,Sandeep Pandey,Luke Simon,Xi Liu,Jian Li
机构: Stony Brook University (石溪大学); Meta AI (Meta人工智能)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:A reasoning trace can accompany a correct answer without providing effective supervision for a particular student. We study how to construct traces that improve learning beyond answer-only supervision by considering both their intermediate content and their accessibility to the target learner. We introduce ReDiST, a student-conditioned trace-construction and distillation framework for evidence-rich reasoning. Starting from the student’s own attempts, ReDiST revises answer-correct drafts against the supplied evidence and repairs unsuccessful drafts with additional guidance from the verified answer. A frozen student-matched reader screens and ranks original and edited candidates using answer recovery and verified-answer probability, given the question, evidence, and candidate trace. We distill the selected targets while balancing trace and final-answer supervision with an auxiliary answer loss. Construction and selection occur offline, and inference uses only the distilled student. Experiments on question-answering benchmarks show that ReDiST can improve both source-task and cross-benchmark performance over answer-only, original-trace, and the evaluated trace-selection and rewriting baselines. Across supervision recipes, performance rankings can differ between source and transfer benchmarks, highlighting the importance of evaluating trace teaching utility across distributions.
[NLP-4] Acceptance Is Not Speed: Length Inflation in Training-Free Lossy Speculative Decoding NAACL2027
【速读】： 该论文旨在解决生成式 AI（Generative AI）中宽松验证（relaxed verification）机制在实际应用中的效率评估失真问题。传统上，推测解码（speculative decoding）的加速效果常以每轮验证中被接受的草稿令牌（draft tokens）数量为衡量指标，但在采用宽松验证时，由于其允许接受在严格验证下应被拒绝的令牌，可能改变目标模型的输出分布，导致生成结果长度显著增加。研究发现，这种“长度膨胀”现象——即推理路径被过度延长——严重抵消了表面上更高的令牌接受率带来的收益。关键发现是：尽管宽松验证在最宽松设置下平均可使每轮接受的令牌数提升至1.34倍，但生成完成长度却增长至1.54倍，且多数新增令牌具有较低的目标模型概率，进一步加剧长度膨胀。同时，部分规则虽减少验证轮次，但任务准确率显著下降，根源在于推理过程耗尽预算而无法生成最终答案。因此，论文指出仅依赖“每轮接受令牌数”作为解码速度的代理指标在宽松验证下不可靠，真正评估效率必须同时报告完成长度、验证轮次及任务准确性，才能全面反映实际性能。
链接: https://arxiv.org/abs/2610.23818
作者: Chiatzen Wang,Bill Xu,Haochen Gu,Hongyang Zhang
机构: University of Waterloo (Waterloo大学)
类目: Computation and Language (cs.CL)
备注: Submitted for NAACL 2027, under review
Abstract:Speculative decoding typically characterizes speedup by draft tokens accepted per verification round. Lossless verification preserves the target model’s output distribution and thus expected completion length, making this metric a meaningful efficiency proxy. Relaxed (lossy) verification accepts tokens that lossless verification would reject, potentially altering that distribution. We identify a length inflation phenomenon: relaxed verification lengthens completions, primarily by extending reasoning traces, largely offsetting apparent gains in accepted tokens per round. We evaluate existing training-free relaxed verification rules on twelve target-drafter pairs against lossless verification on identical prompts, measuring completion length, verifier rounds, wall-clock time, and task accuracy. At each rule’s most relaxed evaluated setting, relaxed verification on average accepts 1.34\times as many tokens per round but produces 1.54\times longer completions. For rules with higher acceptance lengths, tokens accepted only under relaxed verification tend to have low target-model probability, contributing to length inflation. For rules that reduce verifier rounds, accuracy typically drops substantially. Further analysis traces some failures to continued reasoning that exhausts the budget needed to deliver a final answer. Overall, accepted tokens per round alone is an unreliable measure of decoding speed under relaxed verification. Completion length and verifier rounds must also be reported to assess actual efficiency.
[NLP-5] Steering beyond Language: Improving Multilingual Training-Free Text Embeddings via Paired-Residual Subspace Removal
【速读】： 该论文旨在解决当前生成式人工智能（Generative AI）中训练自由型文本嵌入（training-free text embeddings）在多语言场景下严重缺失的问题。现有方法几乎仅聚焦于英语，导致跨语言语义对齐能力不足。其核心挑战在于：尽管不同语言的语义等价句子在内容上一致，但其嵌入表示在向量空间中存在显著的语言依赖性偏差，形成语言条件化的嵌入几何结构。为应对这一问题，论文提出了一种无训练、推理时干预的方法——配对残差子空间移除（Paired-Residual Subspace Removal, PRSR）。PRSR的关键在于通过构建多语言平行句对的残差，捕捉语言特异性偏差，并估计各层的低秩子空间，进而抑制最后一层标记隐藏状态中的相关成分，从而将表示推向更语言无关的语义空间。实验在18种语言上基于MTEB和MMTEB进行了200余项单语评估，并在英语与200多种语言间的双语语料挖掘任务中验证，结果表明，PRSR在多种大语言模型（LLM）上均能一致提升现有训练自由基线在单语和跨语言任务中的性能。
链接: https://arxiv.org/abs/2610.23807
作者: Ailiang Lin,Zhuoyun Li,Yusong Wang,Keyu Mao,Kotaro Funakoshi,Manabu Okumura
机构: Institute of Science Tokyo (东京科学研究所); Tencent(腾讯)
类目: Computation and Language (cs.CL)
备注:
Abstract:While deriving text embeddings directly from frozen LLMs has emerged as a promising direction, existing methods focus almost exclusively on English, leaving multilingual settings largely unexplored. In this work, we present the first systematic study of multilingual training-free text embeddings in LLMs. Our analysis reveals a language-conditioned embedding geometry in which semantically equivalent sentences remain misaligned across languages. We hypothesize that language-specific variation can be disentangled from semantic content to improve embedding quality. To this end, we propose Paired-Residual Subspace Removal (PRSR), a training-free inference-time intervention that steers LLM representations toward a more language-invariant semantic space. PRSR first constructs paired residuals from multilingual parallel sentences to capture language-specific deviations. It then estimates layer-wise low-rank subspaces from these residuals and suppresses the corresponding components in the last-token hidden states. We conduct more than 200 monolingual evaluations across 18 languages using MTEB and MMTEB, along with cross-lingual bitext mining between English and over 200 other languages. Empirical results demonstrate that PRSR consistently improves the performance of existing training-free baselines in both monolingual and cross-lingual settings across different LLMs.
[NLP-6] DLite: History-free Speculative Decoding for Agent ic AI
【速读】： 该论文旨在解决在代理型工作负载（agentic workloads）下，传统推测解码（speculative decoding）技术因持续演化的长上下文导致的候选词接受长度显著下降（最高达80%衰减）以及维持历史键值缓存（KV cache）带来的巨大系统开销问题。其核心挑战在于，随着上下文不断增长，现有轻量级草稿模型难以有效生成可被目标模型接受的候选序列，同时长期维护历史上下文状态会带来不可忽视的内存与计算负担。该研究的关键解决方案是提出DLite——一种无历史依赖的推测解码框架，通过直接复用目标模型的上下文隐藏状态（contextual hidden states），在不显式维护历史上下文的前提下进行局部推测预测，从而将复杂的全局上下文建模问题转化为低复杂度的局部预测问题；同时引入轻量级序列头（sequential head）以低成本恢复块内因果依赖关系。为适配该异构无历史架构，论文设计了一种定制化的两阶段训练策略，实验证明DLite在通用、长上下文及代理型任务上均实现2-6倍的接受长度提升，并相较当前最优方法DSpark性能提升高达70%。
链接: https://arxiv.org/abs/2610.23787
作者: Jiaming Xu,Hanzhen Wang,Tianlang Zhao,Jiewen Xiao,Chengze Yuan,Jiayi Pan,Yongkang Zhou,Jiancai Ye,Yu Wang,Guohao Dai
机构: Shanghai Jiao Tong University (上海交通大学); Shanghai Innovation Institute; Tsinghua University (清华大学); Infinigence-AI
类目: Computation and Language (cs.CL)
备注: 14 pages, 5 figures
Abstract:Speculative decoding has become a mainstream technique for accelerating large language model (LLM) inference by using a lightweight draft model to generate candidate tokens that are subsequently verified by the target model. However, agentic workloads introduce long and continuously evolving contexts and frequent multi-turn interactions. Under such workloads, we observe that existing draft models suffer from up to an 80% degradation in acceptance length, while maintaining their historical KV cache introduces substantial system overhead. In this paper, we identify the growing historical context as a common source of both problems and introduce DLite, a history-free speculative decoding framework for agentic AI. DLite directly reuses contextual hidden states from the target model and performs local speculative prediction without explicitly maintaining the historical context, transforming an increasingly complex modeling problem into local simple prediction problem. A lightweight sequential head further restores intra-block causal dependencies with low overhead. To effectively train this heterogeneous history-free architecture, we introduce a tailored two-stage training strategy. Experiments on general-purpose, long-context and agentic workloads show that DLite achieves 2-6 acceptance length and outperforms the state-of-the-art speculative decoding method, DSpark, up to 70%.
[NLP-7] VocalGrad: Evaluating Acoustic Perception in Audio Language Models NEURIPS2026
【速读】： 该论文旨在解决当前大型音频语言模型（Large Audio Language Models, LALMs）在处理语音中非语言性声学信息时表现不佳的问题，尤其在声学特征与语言内容共现时难以有效利用声学信息。其核心挑战在于现有评估基准未能清晰区分声学感知与高层推理能力，导致无法准确诊断模型失败的根本原因。为此，研究提出VocalGrad——一个专注于评估LALMs对低层次声学变化感知能力的基准测试。该方法将声学感知任务简化为判断时间序列变化方向的二分类问题，从而最大限度降低对语义理解或高级推理的依赖。实验结果表明，人类标注者在该任务上表现优异，而先进LALMs的表现仅接近随机水平，揭示了模型在声学感知上的显著缺陷。通过线性探测分析进一步发现两类典型失败模式：一类是任务相关声学信息虽存在于模型内部表征中但未被语言模型头部有效利用；另一类是信息在深层网络中逐渐丧失线性可解性。总体而言，该研究揭示了当前LALMs在受控条件下建模声学变异能力的严重局限性，并暴露出仅从文本输出无法观察到的表征差异。
链接: https://arxiv.org/abs/2610.23745
作者: Koki Ryu,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: ound (cs.SD); Computation and Language (cs.CL)
备注: NeurIPS 2026 (Evaluations Datasets Track). Dataset page: this https URL
Abstract:Recent large audio language models (LALMs) have extended their capabilities beyond lexical speech understanding to tasks requiring paralinguistic and non-linguistic audio understanding, such as emotion recognition. However, these models still struggle to reliably utilize acoustic information, particularly when it co-occurs with spoken linguistic content. Existing benchmarks do not clearly distinguish acoustic perception from higher-level reasoning, making it difficult to identify the source of these failures. To address this issue, we introduce VocalGrad, a benchmark designed to evaluate LALMs’ ability to perceive low-level acoustic variations. VocalGrad formulates this problem as a simple binary classification of temporal change direction, enabling evaluation largely independent of higher-level reasoning or semantic understanding. Our benchmark results reveal a substantial gap in acoustic perception between human annotators and state-of-the-art LALMs: while human annotators achieve high accuracy, LALMs perform at near-chance levels. We further conduct linear probing to identify structural bottlenecks in these models. Our analysis reveals two distinct failure modes: for some categories, task-relevant information is preserved in internal representations but not utilized by the language model head; for others, the information becomes progressively less linearly decodable across model layers. Overall, our results reveal substantial limitations of LALMs in modeling acoustic variation under controlled settings and expose representation differences not observable from text outputs.
[NLP-8] he Italian lawmaking process: a unified and structured dataset
【速读】： 该论文旨在解决意大利立法过程信息碎片化与非结构化的问题，即尽管立法活动的详细记录已由议会机构存档于官方网站，但其内容多以非结构化文本形式存在，难以实现系统性再利用。为此，研究提出了一种基于大语言模型（Large Language Model, LLM）的受控、系统化抽取方法，通过与人工标注数据对比验证其准确性，将原始文本转化为有序且带时间戳的标准化事件序列，涵盖法案提出、委员会审议、全体会议审议及最终通过等关键环节。该方法的关键在于结合领域知识对LLM进行精准引导，并建立可验证的评估机制，确保抽取结果的可靠性；最终构建了涵盖1985至2026年间意大利第9至第19届议会期间共计5346部法律的完整程序历史数据集，共包含101413条事件记录，每条记录包含活动类型、日期及所属议院，部分还包含额外程序细节。该数据集为计算法学与立法研究提供了基础支撑，支持采用面向流程的数据科学方法分析不同议会期内的立法路径与时间模式。
链接: https://arxiv.org/abs/2610.23741
作者: Matilde Contestabile,Chiara Ferrara,Giovanni Parrillo,Lorenzo Emer,Andrea Vandin
机构: Scuola Superiore Sant’Anna(圣安娜高等学院); University of Pisa(比萨大学); DTU Compute(丹麦技术大学计算学院)
类目: Computation and Language (cs.CL)
备注:
Abstract:The Italian legislative process is documented in detailed parliamentary records hosted in institutional websites. However, much of this information is available only as unstructured text, limiting its systematic reuse. We applied a controlled, systematic LLM-based extraction procedure, validated against manually annotated records, to convert these texts into ordered and dated events including, e.g., bill introduction, committee examination, plenary examination, and approval. The resulting dataset reconstructs the procedural histories of all 5346 Italian laws enacted between 1985 and 2026, spanning parliamentary terms IX to XIX, for overall 101413 events. Each event records the activity, date, and chamber, with additional procedural details where available. The dataset supports research in the emerging fields of computational legal and legislative studies, potentially combined with process-oriented data science methods to examine procedural paths and legislative timing across parliamentary terms.
[NLP-9] Bridging the Language Gap in Scholarly Data II: Enhancing Gender Inference for Chinese Names
【速读】： 该论文旨在解决非西方学者（特别是中文姓名）在科学计量学研究中性别推断的准确性问题。现有方法在处理中文姓名时表现不佳，主要因为国际文献记录中通常仅提供拼音形式，而同一拼音可能对应多个具有不同性别倾向的汉字，导致仅基于拼音进行性别推断存在固有歧义。为解决此问题，论文提出关键解决方案：首先构建两个基于汉字（Hanzi）的性别推断模型——回归模型输出0（男性）至1（女性）之间的连续性别得分，分类模型则输出“男性”、“中性”或“女性”三类标签，并在由11名母语为中文的审评者标注的中文物理学期刊姓名样本上进行训练。结果显示，回归模型的均方根误差（RMSE）为0.09，决定系数（R²）达0.85；分类模型的宏观受试者工作特征曲线下面积（macro-AUC）为0.97，科恩κ系数（Cohen’s κ）达0.81，表明模型性能优异且对类别不平衡具有鲁棒性。针对拼音姓名，论文进一步提出一种融合音译映射与频率加权的预测流水线，并引入模糊度评分以标识无法仅凭拼音确定性别的姓名。在与Namsor、GPT和Claude等现成工具的对比中，本方法在两种书写形式下均展现出与人工标注最高的一致性。应用于美国物理学会（APS）期刊中的独立中文作者样本时，本模型可对所有姓名赋予权性别标签，而现有拼音方法则存在20%至100%的未分类率。研究结果凸显了汉字级信息在可靠推断中文姓名性别中的关键作用。代码与数据已公开共享。
链接: https://arxiv.org/abs/2610.23734
作者: Lisette Espín-Noboa,Mingrong She,Ana Maria Jaramillo,Liuhuaying Yang
机构: Graz University of Technology (格拉茨工业大学); Complexity Science Hub (复杂性科学中心); Maastricht University (马斯特里赫特大学)
类目: Computation and Language (cs.CL); Computers and Society (cs.CY); Machine Learning (cs.LG); Social and Information Networks (cs.SI); Physics and Society (physics.soc-ph)
备注: 36 pages (13 main; 23 SI), 6 figures in main (under review)
Abstract:Perceived gender inference from names is important for inclusive scientometrics research, yet existing methods perform poorly on non-Western scholars. We address this gap for Chinese names, which appear in two forms, Hanzi and Pinyin, its Romanized form (e.g., Zhang Wei, Li Na). In international bibliographic records, names are often available only in Pinyin. However, a single pinyin name can correspond to multiple distinct hanzi characters with different gender associations, so gender inference from Pinyin alone is intrinsically ambiguous. To address this, we first develop two hanzi models, a regression model that produces a continuous gender score from 0 (male) to 1 (female), and a classification model that outputs one of three labels (\emmale, \emneutral, \emfemale). We train them on a sample of hanzi names from Chinese-language physics publications, rated by 11 native Chinese speakers. On this sample, the models achieve an RMSE of 0.09 and an R ^2 of 0.85 (regression), and macro-AUC of 0.97 and a Cohen’s \kappa of 0.81 (classification), demonstrating high performance that is robust to class imbalance. For pinyin names, we propose a pipeline that aggregates hanzi predictions via transliteration and frequency weighting, together with an ambiguity score that flags names whose gender cannot be resolved from Pinyin alone. Benchmarked against three off-the-shelf tools (Namsor, GPT, and Claude), our models achieve the highest agreement with human annotations for both gender scores and gender labels, in both scripts. Applied to an independent sample of Chinese-named authors in APS journals, our models assign a gender to every name, whereas existing Pinyin-based methods leave between 20% and 100% of names unclassified. These findings demonstrate the importance of hanzi-level information for reliable gender inference of Chinese names. Code and data are publicly available.
[NLP-10] RedDarwin: Evolving Jailbreak Skills against Frontier LLM s
【速读】： 该论文旨在解决当前红队测试（red teaming）在面对前沿大语言模型（Large Language Model, LLM）安全防护机制持续演进时，缺乏动态适应能力的问题。现有方法多局限于固定攻击流程或对对抗性提示（adversarial prompts）进行搜索，难以有效应对防护策略的更新。其核心解决方案是提出RedDarwin——一个以可执行攻击技能（executable attack skills）为进化单元的代理式框架。该框架将攻击演化从静态提示与固定攻击模式升级为可复用、可组合的技能单元，通过目标模型反馈实现自主修订、组合、执行与选择，使成功攻击机制成为后续进化的构建模块。RedDarwin共生成65个演化后的攻击技能，使44/47个基础攻击的成功率（Attack Success Rate, ASR）显著提升，平均ASR从18.8%增至47.3%。在未参与训练的强拒绝测试集（StrongREJECT）及前沿未见模型上，其表现均优于基线方法；仅需10次目标模型查询，即可在GPT-5.6-Sol上实现100%的攻击成功率，同时在Claude-Sonnet-5和Gemini-3.8-Flash上分别达到92%和80%。分析表明，高效攻击往往源于可复用机制的创新组合，而非单一强大技能的独立作用。这一范式推动了自适应红队测试的发展，并为安全评估与防御机制设计提供了互补价值。
链接: https://arxiv.org/abs/2610.23726
作者: Zhouyuan Ma,Yutao Wu,Hanxun Huang,Xiang Zheng,Yi Liu,Xingjun Ma
机构: Fudan University (复旦大学)
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL)
备注:
Abstract:Red teaming must continually adapt as frontier LLM safeguards evolve, yet existing approaches largely search for adversarial prompts or optimize within fixed attack procedures. We observe that even when an attack becomes ineffective, its underlying mechanisms can remain useful and can be revised or recombined to construct stronger attacks. Based on this insight, we introduce RedDarwin, an agentic framework that shifts the unit of jailbreak evolution from prompts and fixed attacks to executable attack skills. RedDarwin consists of 47 reusable base skills and autonomously revises, composes, executes, and selects them using target-model feedback, allowing successful discoveries to become building blocks for subsequent evolution. RedDarwin produces 65 evolved skills, improving 44 of 47 base attacks and increasing their mean attack success rate (ASR) on AdvBench from 18.8% to 47.3%. These evolved skills consistently outperform baselines on the held-out StrongREJECT and frontier LLMs unseen during evolution; with only 10 target-model queries, RedDarwin achieves 100% ASR on GPT-5.6-Sol, 92% on Claude-Sonnet-5, and 80% on Gemini-3.8-Flash. Analysis of the evolution trajectories further shows that strong attacks often emerge from novel combinations of reusable mechanisms rather than from individually dominant ones. We believe this paradigm can accelerate adaptive red teaming and inform stronger safeguards, creating complementary value for both safety evaluation and defense.
[NLP-11] Recognizing Study Lineage Is Not Enough: Over-Crediting Same-Trial Evidence in LLM -Based Medical Verification
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在医学领域大模型验证中，对同一临床试验的多篇发表文献在证据聚合过程中出现的过度归因问题。具体而言，尽管模型在直接识别试验谱系（provenance lineage）任务中表现准确，但在整合多篇相关研究作为证据时，会错误地将同一试验的不同发表视为独立支持性证据，从而导致对研究结论的过度假设。其解决方案的关键在于引入“研究分组指令”（study-grouping instruction），通过显式引导模型在证据聚合阶段识别并合并来自同一试验的多个报告，有效抑制了模型在跨任务中的行为偏差。实验结果表明，该指令可使GPT-5.5在纯谱系家族中的过度归因率从4/5降至0/5，显著缩小了模型在“谱系识别”与“证据聚合”两个阶段之间的行为差距。
链接: https://arxiv.org/abs/2610.23721
作者: Mukand Krishna,Anoushka Harit,Chenghao Xiao,Zhongtian Sun
机构: University of Kent (肯特大学); University of Cambridge (剑桥大学); Shanghai University of Finance and Economics (上海财经大学)
类目: Computation and Language (cs.CL)
备注: 20 pages, 5 figures, 17 tables; 8 pages of main content, with references and appendices thereafter
Abstract:Publication count is not study count. A model can identify shared trial lineage correctly in a direct task yet over-credit a second publication from the same trial when aggregating evidence. We evaluate this behavior in medical LLM verification using 14 source-audited clinical families and five model conditions. The families span provenance, clinical applicability, result direction, and composition; the provenance families provide the direct lineage tests. Recognition and aggregation are separate stateless calls, and repeated calls assess within-input stability. Across all three original provenance families, GPT-5.5 identifies the independent update in 10/10 recognition calls per family but over-credits the same-trial report in 5/5 separate aggregation calls per family. Selected Qwen3-4B and Ministral-3-8B conditions show the same cross-task contrast. A study-grouping instruction reduces GPT-5.5 over-credit from 4/5 to 0/5 in each later pure-provenance family. Other model conditions show mixed behavior, and rejected outputs limit several comparisons. The results identify a behavioral gap between elicited lineage recognition and evidence aggregation. The code and data artifacts are available at this https URL.
[NLP-12] he Cognitive Cost of LLM Writing: A Reading-Time Comparison with Human Text
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）生成文本与人类撰写文本在阅读认知负荷上的差异问题，特别是这种差异如何影响读者的加工需求。尽管两者在语言特征上的区别已有广泛研究，但其对阅读过程中认知努力的影响仍缺乏深入探讨。为此，研究提出一种基于人类眼动数据训练的模型RoBERTa-RT，用于预测词级阅读时间，从而量化文本处理的认知负荷。关键解决方案在于构建了一个包含496篇人类写作文章及其对应LLM重写版本的配对数据集，并利用该模型对比两类文本的预测阅读时间。结果表明，LLM生成文本虽整体更短、总阅读时间略低，但单个词的预测阅读时间更高，这与更长且不常见的词汇、更高的词汇多样性和更强的上下文可预测性等语言特征相关。从认知视角看，这一发现揭示了机器生成文本可能在降低总体信息量的同时，增加了个体词语的加工难度，即重新分配了阅读的认知负担。因此，研究强调在评估生成式人工智能（Generative AI）内容时，不仅需关注其语言特性，还应考量其对人类认知处理过程的潜在影响。
链接: https://arxiv.org/abs/2610.23715
作者: Kamer Cekini,Elisabetta Biondi,Chiara Boldrini,Andrea Passarella,Marco Conti
机构: IIT-CNR(意大利国家研究委员会智能信息技术研究所); University of Pisa(比萨大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Although linguistic differences between human-written and Large Language Model (LLM)-generated texts have been extensively studied, their implications for the cognitive effort involved in reading remain less explored. In psycholinguistics, reading time is commonly used as an indirect measure of cognitive processing effort. Building on this perspective, we investigate whether human-written texts and their LLM-generated rewrites exhibit differences in predicted reading time that may reflect differences in processing demands for readers. To this end, we introduce RoBERTa-RT, a model trained on human eye-tracking data to predict word-level reading times. We evaluate the model on both internal and external test sets, construct a paired dataset of 496 human-written articles and their LLM-generated rewrites, and apply RoBERTa-RT to compare their predicted reading times. Our findings show that LLM-generated rewrites tend to be shorter and have slightly lower predicted total reading times, while exhibiting higher predicted reading times per word. These differences are associated with systematic linguistic changes, including longer and less frequent words, greater lexical diversity, and higher contextual predictability. From a cognitive perspective, our findings suggest that LLM-generated writing may redistribute reading demands: reducing the amount of text to process while increasing the predicted processing effort associated with individual words. This highlights the importance of considering not only the linguistic characteristics of machine-generated content, but also its potential implications for human cognitive processing.
[NLP-13] GLOBE: Evolving Structured Agent ic Memory for E-commerce Product Optimization in Generative Search EMNLP2026
【速读】： 该论文旨在解决生成式搜索场景下商品列表优化中语言代理（language agents）难以持续自我改进的问题，即现有方法通常将结构化商品信息视为自由文本处理，导致有效的优化策略隐含于提示词、搜索路径或重写结果中，无法显式积累为可复用的智能记忆。其核心解决方案是提出GLOBE框架，通过构建一个动态演化的“手册”（handbook），以显式、领域特异的行为形式存储优化知识。在每个优化周期中，编辑（Editor）调用相关行为并接收基于排名的反馈，由反思者（Reflector）将其转化为行为更新；为确保实际部署中的可靠性，监督者（Moderator）剔除冗余建议，策展人（Curator）确认有效变更，并禁用有害行为，形成闭环。这一机制将离散的反馈转化为持久、领域聚焦的代理记忆。实验表明，结构化行为优于单一提示优化，且反馈驱动的演化与多层级治理显著提升性能并降低风险。此外，学习到的行为在未见过的商品类别上仍具有效性，证明该框架支持可解释、可迁移的商品优化能力。
链接: https://arxiv.org/abs/2610.23709
作者: Hyunseo Kim,Suil Choe,Sukkyoung So,Daeyeon Yi
机构: LG Electronics( LG 电子), South Korea(韩国)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to the Second Workshop for Research on Agent Language Models(REALM) at EMNLP 2026
Abstract:Optimizing product listings for generative search requires language agents not only to rewrite effectively but also to continually self-improve by retaining successful strategies across diverse categories. However, existing approaches often treat structured listings as free-form text, leaving effective strategies implicit in prompts, search trajectories, or rewritten outputs rather than accumulating them as a reusable agentic memory. To address this limitation, we introduce GLOBE, an agentic framework that drives continual self-improvement by storing optimization knowledge in an evolving handbook of explicit, field-scoped behaviors. In each episode, an Editor applies relevant behaviors and receives rank-based feedback, which a Reflector converts into behavior updates. Crucially for dependable execution in real-world settings, a Moderator removes redundant proposals, a Curator commits accepted changes, and consistently harmful behaviors are deactivated. This cycle transforms episodic feedback into a persistent, field-scoped optimization memory. Experiments show that structured behaviors outperform monolithic prompt optimization, while feedback-driven evolution and moderation further improve effectiveness and reduce downside risk. The learned behaviors also remain effective on unseen product categories, suggesting that agentic memory evolution supports interpretable and transferable product optimization.
[NLP-14] VISTA: Steering LLM s with Latent User Values Inferred from a Few Survey Answers
【速读】： 该论文旨在解决个性化语言模型（language model）中用户价值观建模与应用的难题：现有方法或依赖任务特定偏好学习，或仅将用户画像置于提示（prompt）中，导致模型难以有效利用用户长期稳定的内在价值倾向。其核心解决方案是提出VISTA（Value-Informed Steering and Targeted Acquisition），通过分析用户部分或完整的调查回答，从隐空间中推断出用户的潜在价值观表征，并借助超网络（hypernetwork）将其映射为针对冻结大语言模型（frozen LLM）的用户专属可控干预机制，实现对回答选项的选择或自由文本生成的精准引导。该方法在不增加提示词数量的前提下，在合成偏好数据集Food2D、世界价值观调查（World Values Survey, WVS）及PrimeX上均达到或超越基于上下文的用户画像表现。为进一步降低用户认知负担，VISTA还引入自适应问卷设计策略，动态选择提问内容并决定停止时机——在WVS上仅需完成6%的题量即可保留94%的完整历史信息收益，显著优于传统自适应测试（最多保留42%）。关键发现表明，主导性的控制方向对应一个可解释的价值轴，能准确定位未见用户的位置。研究结果揭示了一种可部署、面向个体差异的多元对齐（pluralistic alignment）范式，具备实际应用潜力。
链接: https://arxiv.org/abs/2610.23702
作者: Yinuo Xu,Jadie Adams,Brian Hu,Shivani Kumar,David Jurgens,Arslan Basharat
机构: University of Michigan (密歇根大学); Kitware Inc. (Kitware公司)
类目: Computation and Language (cs.CL)
备注: 8 pages, 4 figures
Abstract:Personalizing a language model requires knowing what a person values, which stays largely consistent across tasks. Yet most methods either learn task-specific preferences or place the user’s profile in the prompt, where models struggle to act on it. We introduce VISTA (Value-Informed Steering and Targeted Acquisition), which infers a latent profile from users’ (full or partial) survey answers, and maps the profile through a hypernetwork to a user-specific steering intervention in a frozen LLM to select answer choices or generate free-text responses. Given the same evidence, VISTA matches or beats in-context profiles on synthetic preferences (Food2D), the World Values Survey (WVS), and PrimeX, without adding prompt tokens. To reduce the cognitive burden of completing full surveys, VISTA also chooses which questions to ask and when to stop: on WVS it stops after 6% of the questionnaire while retaining 94% of the full-history gain, where standard adaptive testing retains at most 42%. The dominant steering direction is an interpretable value axis on which VISTA places unseen users accurately. Our results point toward pluralistic alignment that adapts to individuals at deployable cost.
[NLP-15] Beyond the Benchmark: Conditioning Domains and Failure Modes of Qwen 3-TTS
【速读】： 该论文旨在解决生成式语音克隆（Voice Cloning）技术在实际应用中存在的一系列关键问题，包括模型条件化方式不明确、在非朗读语境下的表现未知以及失败模式缺乏系统分析。其核心解决方案是通过全面评估Qwen3-TTS的两种语音克隆接口——上下文学习（In-Context Learning, ICL）与X-Vector——在多种真实场景（如会议对话、情感表达及航空交通管制语音）中的性能表现，并对比Chatterbox、XTTS-v2和F5-TTS等基准模型。研究发现，两种接口各具优势：ICL在参考文本正确时能更精准地复制音色，而X-Vector在语音时序建模上表现更优；但当参考文本存在错误时，ICL性能急剧下降。此外，任务领域解释了四倍于系统选择的方差，表明场景适应性至关重要。尽管Qwen3-TTS在可懂度和音高范围方面领先，但在韵律建模上仍落后于其他模型，且客观指标与人工听感评价结果不一致。值得注意的是，现有防伪造检测机制对Qwen3-TTS的克隆样本误报率与真实录音相当，而在航空交通管制场景下所有系统均失效。该研究揭示了当前语音克隆技术在鲁棒性、泛化能力与安全性方面的局限性，为后续优化提供了实证基础。
链接: https://arxiv.org/abs/2610.23678
作者: Séverin Baroudi,Yanis Labrak,Petr Motlicek,Hervé Bredin,Ricard Marxer
机构: Institut National de l’Audiovisuel (INA)(法国国家视听研究所); Université Paris Cité (巴黎城市大学); Télécom Paris (巴黎电信学院); Institut Polytechnique LaSalle Beauvais (拉沙尔-博韦理工学院); CNRS (法国国家科学研究中心); LIA (语言与信息分析实验室)
类目: ound (cs.SD); Computation and Language (cs.CL)
备注:
Abstract:Qwen3-TTS reports state-of-the-art zero-shot voice cloning, but its technical report leaves open how the model should be conditioned, how it behaves outside read speech, and how it fails. We characterise its two cloning interfaces, in-context learning (ICL) and X-Vector, against Chatterbox, XTTS-v2 and F5-TTS on read, meeting, emotional and air-traffic-control (ATC) speech, using eighteen objective measures, an LLM judge, a spoofing countermeasure and a listening test. Neither interface dominates: ICL clones timbre more closely from correctly transcribed references, X-Vector clones timing, and ICL degrades sharply when a short reference transcript is wrong. Domain explains four times more variance than system choice. Qwen3-TTS leads on intelligibility and pitch register but trails on prosody, and the objective measures rank systems differently from listeners. An off-the-shelf countermeasure flags its clones no more often than genuine recordings, and on ATC radio it fails for every system. Code and audio are released.
[NLP-16] In Human Memories. Technology Creativity and Self-Determination: An Ethical Proposal
【速读】： 该论文旨在解决生成式 AI（Generative AI）在当代作曲实践中引发的创作主体性与伦理边界问题，尤其关注其从传统技术工具向具有主动参与性的创作伙伴转变所带来的挑战。其解决方案的关键在于重构作曲家与AI之间的关系：不再将AI视为被动的辅助资源，而是将其视作具有未知潜力的“他者”进行互动探索，在此过程中建立一种伦理导向的共创模式，从而实现技术对创造力的赋能而非替代，体现于弗朗切斯科·维图奇（Francesco Vitucci）《在人类记忆中》（In Human Memories, 2024）一作中，AI不仅作为灵感来源，更成为推动音乐形态生成的核心动力。
链接: https://arxiv.org/abs/2610.23651
作者: Francesco Vitucci,Giuseppe Silvi,Francesco Abbrescia,Anthony Di Furia,Francesco Scagliola
机构: 未知
类目: Computation and Language (cs.CL)
备注: 19 pages, 7 figures. English translation of the article originally published in Italian in Ex Chordis, vol. 3 (2024), pp. 15-38. The Italian original is the version of record
Abstract:This research investigates the role of artificial intelligence (AI) in modern compositional practices, with a particular focus on two case studies: Convergence (2020/21) by Alexander Schubert, for String Ensemble and AI System, and In Human Memories (2024) by Francesco Vitucci, for four cellos and one electroacoustic player, whose installation was created by the Electronic Music School of the “N. Piccinni” Conservatory of Bari. The initial historical survey traces the evolving use of technology in composition, revealing how the search for new expressive possibilities has driven the development of technological tools through experimentation. The appearance of AI marks a fundamental shift: no longer just a tool, AI now poses questions and challenges to the composer. Engaging with AI becomes an exploration of an unknown entity, whose full capabilities are still largely undiscovered. This new interaction leads composers to confront AI not merely as a passive resource but as a participant in the creative process, outlining an ethical path to reclaim technology as a creative tool. In Human Memories reflects this dynamic, demonstrating how AI can serve as both inspiration and a driving force in shaping contemporary music.
[NLP-17] rajectory-Triggered Sparse Collaborative Decoding EMNLP2026
【速读】： 该论文旨在解决生成式 AI（Generative AI）中密集辅助解码方法导致目标模型计算与生成过程紧密耦合的问题，从而降低推理效率。其核心解决方案是提出一种稀疏协同解码（sparse collaborative decoding）框架，即由草稿模型生成大部分令牌，仅在需要纠正时调用目标模型进行短片段修正。该方案的关键在于设计一个高效的触发机制以决定何时调用目标模型。研究发现，局部不确定性指标（如熵、负对数似然、logit 间距、top-k 质量及速度特征）作为触发信号表现不佳，预测失败的准确率接近随机水平（AUC 0.50–0.53）。相比之下，基于草稿模型置信度路径的全局轨迹形状特征（aggregate trajectory-shape features）结合轻量级逻辑回归预测器，仅使用草稿正确性标签进行训练，即可实现高达 AUC 0.86 的预测性能，显著提升触发准确性。该方法揭示了触发时机的权衡：早期调用具有更强的引导能力但证据较弱，而晚期调用虽有更强证据但调整空间有限。实验表明，在五个 GSM8K 草稿-目标模型组合上，该在线学习的稀疏协同解码方法相比最优开环调度提升了 5.3 至 20.0 个准确率点；在 MATH-500 上，当早期预测与目标模型能力匹配时可获得 7.0–13.5 点增益，否则性能下降；在 StrategyQA 上亦取得 5.0–8.3 点改进，证明其具备超越算术任务的泛化能力。
链接: https://arxiv.org/abs/2610.23613
作者: Amit Ranjan Trivedi,Sruthikesh Surineni,Kyong Ho Lee,Mohit Mittal,Steven Li,Rex Hill
机构: University of Illinois Chicago (伊利诺伊大学芝加哥分校); Meta
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 17 pages. Accepted at the EMNLP 2026 uncertainty workshop
Abstract:Dense assistance methods such as speculative decoding repeatedly verify draft tokens with a target model, keeping target computation coupled to generation. We study sparse collaborative decoding, where a draft model generates most tokens and a target model is invoked only for short corrective segments. The key problem is deciding when to call the target. We show that local uncertainty is a weak trigger: entropy, negative log-likelihood, logit margin, top- k mass, and velocity features give near-chance failure prediction, with AUC 0.50-0.53. In contrast, aggregate trajectory-shape features over the draft confidence path reach AUC 0.86 using a lightweight logistic predictor trained only from draft correctness labels. This induces a trigger-timing tradeoff: early calls have more leverage but weaker evidence, while later calls have stronger evidence but less ability to redirect the trajectory. Used online, learned sparse collaborative decoding improves over the best open-loop schedule by +5.3 to +20.0 accuracy points across five GSM8K draft-target pairs. On MATH-500, it gains 7.0-13.5 points when early prediction and target competence align, but loses when either is weak. On StrategyQA, it improves over open-loop by 5.0-8.3 points, showing transfer beyond arithmetics.
[NLP-18] SIT-Eval: Situation-Based Interactive Turn-by-Turn Evaluation of Conversational Agents
【速读】： 该论文旨在解决当前对话型人工智能（Conversational AI）评估方法在面对多样化对话场景时的不足，特别是现有对话基准普遍采用静态自然对话，难以覆盖低概率但关键的安全性事件（如罕见情境下的不当行为），导致在实际部署规模下频繁出现失效。其核心挑战在于：如何在保持对话真实性的前提下，实现对复杂情境下模型行为的全面、统计可靠的评估。解决方案的关键在于提出SIT-Eval——一种可调节的情境化评估框架，将理想行为形式化为“情境规则”（if situation is P, follow rules Q），并通过大语言模型（LLM）作为对话对手进行均匀采样的情境模拟，并由具备上下文感知能力的LLM评判器，依据所有适用的情境规则对模型回复进行多维度评分。该评判器通过人工验证的合成黄金对话数据进行验证，确保评估可靠性。实验结果表明，该方法能揭示聚合指标掩盖的细粒度失败模式，揭示各模型在模糊性澄清与苏格拉底式提问等关键能力上的普遍短板。
链接: https://arxiv.org/abs/2610.23612
作者: Ravil Mussabayev,Saveliy Zavertyaev,Dmitriy Parpulov,Andrey Savchenko
机构: Satbayev University (萨特巴耶夫大学); Sber AI Lab (斯贝爾人工智能实验室)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:The rapid proliferation of conversational AI agents necessitates evaluation methods that assess performance across diverse dialogue scenarios, utilizing multiple situational criteria, as appropriate behavior depends on context and encompasses various aspects. Existing dialogue benchmarks typically use static natural conversations, which underrepresent rare situations—a critical safety gap, as low-probability events become frequent failures at production scale. However, limited sample sizes force a choice: either maintain natural dialogues for realism or turn to adaptive situation sampling for comprehensive yet statistically reliable coverage. We introduce SIT-Eval, an adjustable framework for evaluating situation-based conversational agents. We formalize an agent’s desired behavior as situational rules: if situation is P, follow rules Q. At each conversation turn, an LLM-interlocutor enacts a uniformly sampled situation. A context-aware LLM-judge leverages all conditionally applicable rules as criteria to evaluate the agent’s reply for that situation. The judge is validated using human-verified synthetic golden dialogues. Benchmarking 8 LLMs on 1292 conversation turns reveals failure modes hidden by aggregate metrics: models with similar overall scores vary widely in their adherence to individual rules. All models struggle most with ambiguity clarification and Socratic questioning.
[NLP-19] Not Every Mention Is a Claim: Rethinking Object Hallucination Detection in Vision-Language Models
【速读】： 该论文旨在解决视觉语言模型（VLMs）中对象幻觉（object hallucination）评估中存在的误判问题，即现有评估基准和检测器普遍将所有对象提及视为对对象存在的断言，而忽略了“非断言性提及”（non-assertive mentions）的存在。这类提及如“a bench for people to sit on”中提到“people”但并未断言其在图像中存在，属于描述性、目的性、否定或模糊表达等类型，不应被误判为幻觉。这种错误归因导致幻觉检测器的误报率虚高。为此，论文提出NAME-Bench（Non-Assertive Mention Evaluation Benchmark），包含25,497个来自COCO图像的标注对象提及，明确区分“具体存在”（concrete）、“幻觉”（hallucinated）与“非断言性提及”三类，并进一步细分非断言性提及为描绘性、目的性、否定性和模糊性四类。实验表明，当前所有幻觉检测器均将非断言性提及误标为幻觉，误报率高达14%至80%。通过在训练中引入非断言性提及作为负样本，可显著降低LLaVA-1.5的误报率，但在四个更新的VLMs上需五到十倍更多数据才能达到类似效果，且会降低对真实幻觉的检出率。研究揭示了现有检测方法混淆“提及”与“断言”的根本缺陷，强调幻觉评估必须区分文本中的陈述内容与单纯提及行为，以实现更准确的性能评估。
链接: https://arxiv.org/abs/2610.23610
作者: Hsin-Ling Hsu,Neeraja Kirtane,Kuan-Hao Huang
机构: National Chengchi University (国立政治大学); Texas A&M University (德州农工大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
备注:
Abstract:Object hallucination in vision-language models (VLMs) is usually measured by checking each object a caption names against the image. Most benchmarks and detectors alike treat every object mention as a claim that the object is present. This assumption fails when a caption names an object without asserting it, as in “a bench for people to sit on”, which mentions people without claiming that any are in the image. We call such mentions non-assertive. Treating them as hallucinations inflates the error rates attributed to models that use such language. To separate them from real hallucinations, we introduce NAME-Bench (Non-Assertive Mention Evaluation Benchmark), 25,497 object mentions from captions that five VLMs generate for COCO images. Each mention is labelled concrete (asserted and present), hallucinated (asserted and absent), or non-assertive, and non-assertive mentions are further divided into four subtypes: depictive, purpose, negation and hedged. All the hallucination detectors that we evaluate flag non-assertive mentions as hallucinations, at rates between 14% and 80%. Adding non-assertive mentions to detector training as negatives largely corrects this on LLaVA-1.5. On the four newer VLMs, a comparable correction takes five to ten times more data, and the detectors then catch fewer real hallucinations. These findings show that current detectors conflate mentioning an object with asserting it, and that this conflation is harder to undo on the four newer VLMs than on LLaVA-1.5. Hallucination evaluation should therefore separate what a caption asserts from what it merely mentions.
[NLP-20] Enhancing RAG -based Medical VQA with Gated Multimodal Representations ECIR
【速读】： 该论文旨在解决医学视觉问答（Medical Visual Question Answering, MVQA）中生成式模型可能出现的事实性不一致问题，即模型输出虽语义连贯但与真实临床证据不符。现有基于检索增强生成（Retrieval-Augmented Generation, RAG）的方法通常依赖单一模态（如仅文本或仅图像）进行检索，难以充分捕捉医学图像与临床问题之间的联合语义信息。为此，论文提出一种多模态RAG框架，其核心创新在于通过门控多模态单元（Gated Multimodal Unit）学习跨模态的联合查询表示，从而在保持底层医学视觉语言模型（Med-LVLM）冻结的前提下，实现对来自图像和问题的互补信息的有效整合，提升外部临床证据检索的准确性与相关性。实验结果表明，该方法在基准数据集上显著优于基线模型，并达到与当前先进方法相当的性能，验证了多模态查询表示在医学视觉问答中实现更有效证据检索的潜力。
链接: https://arxiv.org/abs/2610.23588
作者: Daniela Jimenez-Pano,Hugo Jair Escalante,Delia Irazú Hernández-Farías
机构: LabTL - INAOE, Puebla, Mexico; Institute for AAII, UTEP, El Paso, TX, USA; Department of Computer Science, UTEP, El Paso, TX, USA
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
备注: Preprint submitted to ECIR
Abstract:Medical Vision-Language Models (Med-LVLMs) have shown strong performance in Medical Visual Question Answering (MVQA), yet they may still produce factually inconsistent responses. Retrieval-Augmented Generation (RAG) can mitigate this issue by incorporating external clinical evidence; however, retrieval is often guided by a single modality, potentially overlooking the joint semantics of medical images and clinical questions. We introduce a multimodal RAG framework that learns a joint query representation through a Gated Multimodal Unit, enabling evidence retrieval to leverage information from both modalities while keeping the underlying Med-LVLM frozen. We evaluate the proposed approach on benchmark datasets. Results show that the multimodal query representation consistently outperforms a reference Med-LVLM and remains competitive with state-of-the-art methods. Our findings demonstrate the potential of multimodal query representations for more effective evidence retrieval in Medical VQA.
[NLP-21] Analyzing Occupational Gender Bias across the LLM Training Lifecycle AACL
【速读】： 该论文旨在解决后训练（post-training）是否能够有效缓解语言模型在预训练阶段继承的社会偏见，特别是职业性别偏见这一关键问题。研究发现，尽管后训练被普遍认为可降低模型的性别偏差，但实证证据并不一致。针对完全开源的OLMo 3语言模型家族，研究通过在不同后训练阶段对合成人物角色进行开放式叙事生成，系统分析了性别偏见在建模流程中的演化规律。实验设置包括“显式性别”（gender-given）与“隐含性别推断”（gender-assumed）两种场景：在显式性别条件下，后训练显著提升了男性编码角色的语义多样性，而女性编码角色则未出现显著提升，尤其在深度强化学习优化（DPO）和强化学习视觉反馈（RLVR）阶段；而在隐含性别推断条件下，后训练对男女角色的影响趋于相似。值得注意的是，模型自身对性别的推断存在系统性女性偏好，且该偏差随每一轮后训练单调增强，与美国劳动力统计数据相比明显偏离，尤其在高度男性主导的职业中更为显著。进一步溯源分析显示，监督微调数据集中仅存在7.7%的女性词元盈余，远不足以解释下游观测到的显著偏差。因此，该研究揭示了后训练不仅未能消除性别偏见，反而可能放大模型对女性的非对称性认知，其关键在于模型在推理过程中内生的性别归因机制及其在后训练中不断强化的女性偏好倾向。
链接: https://arxiv.org/abs/2610.23587
作者: Manthan Gadhia,Alexander Hoyle,Karolina Stańczak
机构: ETH Zürich(苏黎世联邦理工学院); TU Wien(维也纳工业大学); Complexity Science Hub(复杂性科学中心); ETH AI Center(苏黎世联邦理工学院人工智能中心)
类目: Computation and Language (cs.CL)
备注: Submitted to 7th Workshop on Gender Bias in Natural Language Processing @AACL-IJCNLP 2026
Abstract:Post-training is widely assumed to reduce the social biases language models inherit from pretraining, yet direct evidence for this assumption remains mixed. In this paper, we examine how gender disparities evolve across the language modeling pipeline. In particular, we study occupational gender bias in OLMo 3, a fully open language model family, across its post-training stages by prompting each checkpoint to generate open-ended narratives about synthetic personas given their occupation and education. We consider two settings, a gender-given setting in which the profile states the persona’s gender explicitly and a gender-assumed setting in which it must be inferred by the model. We find that when gender is stated, post-training increases semantic diversity for male-coded profiles, whereas for female-coded profiles the increase is smaller after SFT and not significant after DPO and RLVR. When gender is assumed by the model, post-training affects narratives about female and male personas similarly. The model’s own gender assumptions, however, show a consistent female-preferential skew relative to United States labor-force demographics, except for heavily male-dominated occupations. Every successive post-training stage monotonically increases the odds of attributing a female label over the base model. A provenance analysis of the supervised fine-tuning corpus finds only a modest 7.7% female-token surplus, too small to explain the size of these downstream effects. The project code is available under: this https URL.
[NLP-22] MARS: Memory-Aware Routing for Session-Level LLM Assistants
【速读】： 该论文旨在解决长期大语言模型（LLM）助手在交互演化过程中，如何协同决策先前上下文的保留与模型选择的问题。现有记忆系统与模型路由机制通常独立运作，但记忆构建会直接影响输入的证据、干扰项及复杂度，导致决策不一致。其解决方案的关键在于提出MARS（Memory-Aware Routing System），通过会话状态（session state）作为记忆构建与模型选择之间的共享控制接口，实现二者协同优化。MARS引入查询与主题管理器（Query and Topic Manager, QTM），生成紧凑且决策相关的状态表征，用于指导固定的状态条件记忆策略与在线路由机制，使每一轮交互都能获得适配的上下文视图和合适的模型，在保证质量的前提下实现记忆使用与计算成本的平衡。实验表明，MARS在RealMem基准上以仅14.5%的GPT-4o生成API成本，保持了98.2%的全上下文性能，并在记忆召回与检索准确率方面优于对比系统；在MultiChallenge中虽整体略逊于最优静态系统，但在机制聚焦子集上以更低成本逼近其表现，验证了其在长历史、依赖记忆交互场景下的高质量-高记忆-低成本权衡优势。
链接: https://arxiv.org/abs/2610.23583
作者: Minghan Wang,Jinhang Zuo,Zhongxiang Dai,Zhiyong Wang,Fang Kong
机构: 未知
类目: Computation and Language (cs.CL)
备注: 32 pages, including references and appendices
Abstract:Long-term LLM assistants must repeatedly decide what prior context to expose and which model should answer as an interaction evolves. Existing memory systems and model routers usually make these decisions separately, although memory construction changes the evidence, distractors, and difficulty of the resulting input. We introduce MARS, a memory-aware routing framework that uses session state as a shared control interface between memory construction and model selection. A Query and Topic Manager (QTM) estimates compact decision-relevant state that guides a fixed state-conditioned memory policy and an online router, enabling each turn to receive an appropriate memory view and a suitable model under a quality-memory-cost trade-off. We evaluate MARS on RealMem and MultiChallenge with 16 system configurations each, covering long-history memory-driven interaction and bounded sessions with structured context interference. On RealMem, MARS preserves 98.2% of full-context GPT-4o’s QA while using only 14.5% of its generation API cost, including both generation and controller calls for MARS, and achieves the strongest measured memory-recall and retrieval scores among compared systems. On MultiChallenge, MARS trails the strongest static system overall but approaches its performance at lower recorded cost on a mechanism-focused subset. Together, these results support a favorable quality-memory-cost trade-off for long-history, memory-dependent interaction.
[NLP-23] Execution Before Memory: Diagnosing Memory Failures in Video World Models
【速读】： 该论文旨在解决现有视频世界模型（Video World Model）记忆评估中因多任务混淆导致的评估失真问题。当前评估方法将场景设置、指令执行与结果呈现等不同维度合并为单一指标，使得低分易被误读为“遗忘”，而实际上若指令未被执行，其隐藏状态转移本就不应发生，因此评估无法真正检验模型的记忆能力。为此，论文提出MIST（Memory Inference Separation Test），通过将指令执行与记忆保持分离：模型先在指令隐藏状态下构建场景，随后在后续阶段才被要求展示该指令的结果，从而严格控制信息泄露。该设计揭示了关键瓶颈——模型虽能正确设置场景，却无法执行指定动作。此现象同样存在于通用记忆评估中。基于此，论文进一步提出M3-Agent（Make the Missing Move）方法，通过递归修正输入指令的细节，使模型成功完成指令操作。实验表明，相较于基线模型，M3-Agent将MIST揭示得分提升逾三倍；在WRBench和MBench-T基准测试中，分别实现26.1%和52.7%的性能提升。该方法从执行层面切入，不仅展示了冻结模型在明确指令下的真实能力，还建立了更严格的评估基准。
链接: https://arxiv.org/abs/2610.23542
作者: Weiyang Jin,Jiwen Yu,Sihan Xu,Yuwei Niu,Xihui Liu
机构: University of Hong Kong (香港大学); University of Michigan - Ann Arbor (密歇根大学安娜堡分校); Peking University (北京大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
备注: 27 pages, 7 figures. Project page: this https URL Code: this https URL
Abstract:A memory test for a video world model should say whether the scene was set, whether the asked change was carried out, and whether the last frame shows the result. Current tests fold these into one number, so a low score is easy to read as forgetting. When the change is not carried out, the hidden state transition cannot appear, so the test is not really a test of memory. We introduce MIST, an evaluation that separates the change from the memory it should leave behind. The model first sets the scene while the change stays hidden, and only later is asked to show it, with leakage before that point kept under control. This separation locates the bottleneck: the model sets the scene but does not carry out the instructed change. The same gap appears in general memory evaluations. On that basis we propose M3-Agent (Make the Missing Move), which recursively revises the details of the input instruction so that the model successfully carries out the requested instruction. Relative to the baseline, it more than triples the MIST reveal score. On the benchmarks, WRBench’s score T rises by 26.1% and MBench-T’s score U by 52.7%. The method meets the problem from the execution side: it shows what the frozen model can do once the move is stated, and it sets a stricter baseline.
[NLP-24] ConvexLLM : Scaling Convex-Hull Speech Interfaces to General-Purpose Audio Intelligence
【速读】： 该论文旨在解决现有语音与音频语言模型中，声学特征到解码器输入空间映射过程中，解码器需额外学习如何解释声学表示而导致的性能损耗问题，尤其关注在保持预训练文本能力的同时有效利用声学信息。其解决方案的关键在于提出一种名为C-Gate的新型接口机制，通过将音频表征为冻结文本嵌入（frozen text embeddings）的连续凸组合（continuous convex combinations），实现声学信息与解码器已有文本语义空间的直接对齐。这一设计使解码器无需重新学习声学-语义映射，从而在保留强文本理解能力的同时，显著提升对语音任务的泛化表现。实验表明，基于此机制的ConvexLLM在VoiceBench多任务评估中达到75.36的综合得分（优于开源模型），并在口语知识与推理任务上大幅领先基线模型，验证了凸组合文本嵌入作为通用语音与音频语言模型接口的有效性。
链接: https://arxiv.org/abs/2610.23539
作者: Ming-Hao Hsu,Shujie Liu,Yuxuan Hu,Yuxiang Wang,Yingda Shen,Jinyu Li,Zhizheng Wu
机构: 未知
类目: Computation and Language (cs.CL); Sound (cs.SD); Audio and Speech Processing (eess.AS)
备注:
Abstract:Speech and audio language models connect continuous acoustic representations to decoders pretrained primarily on text data. Most such models use MLP or Q-Former interfaces to map acoustic features into the decoder’s input space, requiring the decoder to learn how to interpret the resulting representations. We investigate whether grounding these representations in the decoder’s existing text embeddings can help it use acoustic information while better preserving its pretrained text capabilities. We present ConvexLLM, a general-purpose speech and audio language model whose C-Gate interface represents audio as continuous convex combinations of frozen text embeddings. ConvexLLM achieves an eleven-task VoiceBench aggregate of 75.36, the highest among the open-weight models we compare, and an MMAU Test-mini accuracy of 62.50%. Its largest margins over the baselines are on spoken knowledge and reasoning tasks, while the strongest baselines are more accurate on audio understanding and transcription. In a comparison with a 2B-parameter decoder, C-Gate achieves higher aggregate spoken-language performance than MLP and Q-Former interfaces, with text-task performance close to MLP and task-dependent tradeoffs in transcription and audio understanding. These results support convex combinations of text embeddings as a viable interface for general-purpose speech and audio language models.
[NLP-25] EduWild: How Teacher-Configured Chatbots Fail in Real-World Use and How to Fix Them
【速读】： 该论文旨在解决教育类聊天机器人（educational chatbots）在真实教学场景中因教师通过自然语言指令配置而引发的各类失效问题，尤其关注这些失效如何影响实际教学效果。其核心问题是：尽管教师能够以自然语言设定聊天机器人的行为，但这些机器人在实际使用中频繁出现与教师意图不符的行为，导致教学流程中断、学习者负担过重或关键教学环节遗漏。解决方案的关键在于构建一个系统化的失败分析框架——EduWild数据集与29个代码组成的分类体系，涵盖事实性错误、工作流偏差、诊断失误及教学策略缺陷等维度，并结合经人工校准的大语言模型（LLM）判断器进行精准标注。基于对高频失败模式的识别（如跳过必要流程步骤53.8%、忽略预期设定33.4%、信息过载37.1%），研究设计了一种基于提示工程（prompt-based harness）的干预机制，针对六类最常见故障模式进行优化。在涉及200名大学生和10个聊天机器人的实验中，该方法将每轮对话中的目标故障代码均值从1.406降至0.489，相对减少65.2%，显著提升了聊天机器人在真实教学环境下的可靠性与一致性。这表明，通过实证驱动的失败分析可有效指导教育聊天机器人的实用改进。
链接: https://arxiv.org/abs/2610.23530
作者: Ziqi Liu,Ruiwei Xiao,Yuxuan Li,John Stamper
机构: Carnegie Mellon University (卡内基梅隆大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Teachers increasingly configure educational chatbots through natural-language instructions, but how these bots fail in real-world use remains poorly understood. We introduce EduWild, a dataset of 1,726 conversations with 217 teacher-configured chatbots, each paired with the instructions and configuration behind it, including 900 conversations from the intended end users of 35 deployed chatbots. We develop a 29-code taxonomy covering factual, workflow, diagnostic, and pedagogical failures and apply it using a human-calibrated LLM judge. The most frequent failures are lapses in following teacher intent — bots skip required workflow steps (53.8% of end-user conversations), omit expectation setup (33.4%), and overload learners with information or questions (37.1%). Guided by these findings, we design a prompt-based harness targeting the six most frequent failure modes. In a study with 200 college students across ten bots, the harness reduces the mean number of target failure codes per conversation from 1.406 to 0.489, a 65.2% relative reduction with bots weighted equally. These findings show how real-world failure analysis can guide practical improvements to educational chatbots.
[NLP-26] ruth in Contrast: Calibrated Factual Evidence from Language Model Likelihoods
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在生成事实性陈述时，难以可靠估计其置信度的问题。现有方法在事实性判断的区分能力（discrimination）与置信度校准（calibration）之间往往存在权衡，尤其在低资源语言中表现不佳。其核心解决方案是提出对比生成似然（contrastive generative likelihoods），构建对比事实证据（Contrastive Factual Evidence, CFE），并进一步推导出其概率形式——对比事实后验（Contrastive Factual Posterior, CFP）。CFP通过比较同一命题在不同事实条件下的似然差异，理论上可抵消两组近似条件分布间的共性误差，从而获得更紧致的校准边界。实证结果表明，CFP显著提升了置信度校准性能，同时保持了与最强竞争方法相当的事实区分能力。然而，在低资源语言中，所有方法的区分能力均普遍较弱，导致校准指标的实际效用受限。为此，研究进一步揭示了似然空间中的几何结构：存在一个承载事实证据的方向（evidence-bearing direction）与一个与证据无关的正交方向（evidence-invariant direction）。基于此，作者提出仅在事实证据方向上对目标语言陈述进行与强英语证据的对齐，无需目标语言标注即可有效提升事实区分能力及开放域事实生成质量。
链接: https://arxiv.org/abs/2610.23527
作者: Ponhvoan Srey,Jason Lucas,Pittawat Taveekitworachai,Junyoung Lee,Anh Tuan Luu
机构: Nanyang Technological University(南洋理工大学); University of Colorado at Boulder(科罗拉多大学博尔德分校); National University of Singapore(新加坡国立大学); VinUniversity
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Large Language Models (LLMs) encode substantial factual knowledge. Yet, estimating confidence in factual claims reliably remains challenging. We study likelihood-based factual confidence from the complementary perspectives of discrimination and calibration across multilingual factual statements. Specifically, we instantiate contrastive generative likelihoods as Contrastive Factual Evidence (CFE), which compares the likelihood of the same claim under competing factual and non-factual conditions, and derive its probabilistic counterpart, the Contrastive Factual Posterior (CFP). We show analytically that CFP calibration depends on the differential error between the two surrogate conditionals, allowing errors shared across them to cancel and yield tighter calibration bounds. Empirically, CFP substantially improves calibration while maintaining factual discrimination comparable to its strongest competitor. Nonetheless, discrimination is uniformly weak across methods in lower-resource languages, which simultaneously undermines the utility of the measured calibration. To this end, we interpret a likelihood geometry, within which an evidence-bearing direction is induced, along with an orthogonal evidence-invariant direction. In our investigation, we selectively align target-language claims with stronger English evidence only along the factual-evidence direction, improving discrimination and open-ended factual generation without target-language labels.
[NLP-27] Perceive Solve Reason : Progressive Capability Formation in Time Series-LLM s
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在时间序列理解能力上的不足问题，当前研究普遍将LLMs仅用于时间序列的预测与分类任务，缺乏对感知、求解与推理等多层级能力的系统性训练与评估。为此，作者构建了首个涵盖337K训练样本和44K评估样本的综合性时间序列基准套件，分为“感知（Perceive）、求解（Solve）、推理（Reason）”三个递进能力阶段，以探究时间序列信息如何有效融入LLM并实现能力的渐进式发展。其解决方案的关键在于引入专用编码器（dedicated encoder）以高效提取时间序列特征，实验表明：使用编码器的40亿参数（4B）模型在感知性能上超越GPT-4o，而将时间序列文本化则导致推理准确率下降8–10分；同时，按顺序学习三种能力优于混合训练，且在采用编码器的前提下，仅用少量推理数据即可达到超过三倍于纯推理训练的效果。进一步分析揭示，各能力间通过网络不同深度进行迁移而非简单堆叠，且单阶段在自身基准上收敛并不意味着具备下一阶段的能力基础。最终提炼出一种简洁训练范式，使无需目标领域微调的4B时间序列大语言模型（TS-LLM），在八个跨域指标中的七项上超越近期四款7B级TS-LLM，最高提升达11分。因此，论文主张应将时间序列视为与文本、图像同等重要的第一类模态，通过专用编码器先实现感知，再逐步推进至推理。
链接: https://arxiv.org/abs/2610.23521
作者: Xin Qiu,Junlong Tong,Wei Zhang,Xiaoyu Shen
机构: EIT-NLP Lab, Eastern Institute of Technology (东方理工大学); Zhejiang University (浙江大学); Shanghai Jiao Tong University (上海交通大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models (LLMs) can now perceive, describe, and reason about images and audio, yet time series understanding remains underestimated: most work uses LLMs as forecasters and classifiers, reasoning studies are few and narrow, and no suite trains and evaluates the full range of capabilities. In this paper, we build the first such suite, with 337K training and 44K evaluation samples across three progressive capabilities, Perceive, Solve, and Reason, and use it to ask how time series should enter an LLM that must master all three, and in what order. We find that a dedicated encoder is essential: with one, a 4B model surpasses GPT-4o on every perception attribute we test, whereas textualizing the series costs 8–10 points in reasoning accuracy. The capabilities then build on one another: learning them in order beats mixing the same data and rivals reasoning-only training with over three times the reasoning data, but only when an encoder is used. Analyses further reveal that capabilities are transferred through distinct network depths rather than simply stacked, and that a stage converged on its own benchmark is not necessarily ready for the next. Distilled into a simple recipe, these findings yield 4B TS-LLMs that, without any target-domain data, outperform four recent 7B TS-LLMs on seven of eight out-of-domain metrics, by up to 11 points. We therefore advocate treating time series as a first-class modality for LLMs, read through dedicated encoders and taught to perceive before they reason.
[NLP-28] MuninAI: A Self-Hosted Agent ic Framework for Independent Research Groups
【速读】： 该论文旨在解决科研团队在使用语言模型处理领域文献时面临的两难困境：一方面，依赖外部托管服务虽可访问其未发表的研究成果，但存在数据隐私与泄露风险；另一方面，自托管部署虽能保障数据本地化，却因模型无法直接访问本组文献语料而难以实现有效知识利用。为此，论文提出Munin——一个开源的自托管科学助手框架，其核心创新在于构建了一个集文献索引、引用图谱分析与专用推理工具于一体的代理（agent）封装系统，而非依赖特定大模型本身。该框架的关键突破在于“代理封装”机制，通过集成专门的阅读、检索与计算组件，显著提升模型对领域文献的理解与推理能力。实验表明，在199组配对文献问答任务中，该封装使基础模型准确率从0.387大幅提升至0.874（绝对提升+0.487，95%置信区间[0.407, 0.568]），远超传统“检索-回答”范式表现；且在移除源文献后，系统可识别约90%的原正确答案并主动拒绝生成，有效避免虚构引用，在三个不同开发者提供的模型（包括可在单块消费级GPU运行的200亿参数模型）上均保持稳定性能。此外，即使完全禁用外网连接，其性能仍优于已发布的PaperQA2基准。研究随附完整的评估套件，支持用户在自有语料与模型上一键复现结果，验证了“代理封装”作为通用增强机制的有效性，强调其相较具体模型选择更具决定性价值。
链接: https://arxiv.org/abs/2610.23480
作者: Markus Fischer,Jens Meiler,Matthias Elgeti
机构: 未知
类目: Computation and Language (cs.CL)
备注: 37 pages, 3 figures, 20 tables. Code, evaluation suite and scorecards: this https URL (archived release v1.0.0: this https URL )
Abstract:Research groups that want a language model grounded in their own literature face a choice between hosted services that receive their unpublished work and self-hosted deployments that keep the data local but give the model no access to the corpus. Munin is an open-source framework for a self-hosted scientific assistant that indexes a group’s corpus and its citation graph, and wraps whichever open-weight model the group can serve in an agent harness with specialised reading, search, and computation tools. We claim that the harness, rather than the specific model, is what makes the system useful on domain literature, and we ship the evaluation suite that tests this claim with the software. On 199 paired literature questions, the harness more than doubles the accuracy of the base model, from 0.387 to 0.874 (+0.487, 95% CI [0.407, 0.568]), while naive retrieve-then-answer scores below the unaided model on every backbone tested. Under the harness, roughly twice as many answer claims trace to a passage the model read itself as with fixed top-5 retrieval. When the source paper for a question is removed from the corpus, the system abstains on about nine in ten of the questions it had answered correctly on the Qwen backbones and eight in ten on gpt-oss-20b, and it produces no fabricated local citation across 300 fabricated references. The pattern holds on three backbones from two developers, including a 20-billion-parameter model that runs on a single consumer-grade GPU. With all outbound network access disabled, both Qwen backbones still exceed the published PaperQA2 accuracy. The evaluation suite runs as a single command, so an adopting group can re-run it on its own corpus and model. Comments: 37 pages, 3 figures, 20 tables. Code, evaluation suite and scorecards: this https URL (archived release v1.0.0: this https URL) Subjects: Computation and Language (cs.CL) ACMclasses: H.3.3; I.2.7 Cite as: arXiv:2610.23480 [cs.CL] (or arXiv:2610.23480v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2610.23480 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-29] A low-dimensional dynamical basis for tongue shape control
【速读】： 该论文旨在解决语音产生过程中高维构音系统（high-dimensional articulatory system）的控制复杂性问题，即如何在众多自由度（degrees-of-freedom）中实现高效、稳定的言语运动控制。其核心挑战在于确定何种低维控制变量（low-dimensional control variables）能够有效表征复杂的舌部运动动态。论文提出的解决方案关键在于：引入一种基于整体舌形模式（holistic tongue shape modes）的低维坐标系，作为构音控制的潜在基础。研究通过跨说话者坐标空间分析，从36名英式英语母语者发单元音时的超声数据中提取出经验性的舌形模式，并构建一个三维线性点吸引器模型（three-dimensional linear point attractor）来拟合这些随时间变化的模式。结果表明，该模型能以高精度重建舌形动态，说明复杂舌运动可由若干独立模式在线性动力学法则下叠加而成。这一框架将部分控制复杂性从控制信号转移到坐标系统本身，从而揭示了一维构音测量中可观测非线性的成因，并提出舌形模式可能是底层肌肉协同（muscle synergies）的合理运动学表达。
链接: https://arxiv.org/abs/2610.23456
作者: Sam Kirkham,Patrycja Strycharczuk
机构: Lancaster University (兰卡斯特大学); University of Manchester (曼彻斯特大学)
类目: Computation and Language (cs.CL)
备注: Pre-print currently under review
Abstract:Speech production involves the control of a high-dimensional articulatory system with many degrees-of-freedom, but a widespread hypothesis is that speakers navigate this complexity using a set of lower-dimensional control variables. Common candidates for these control variables include individual articulators (e.g. tongue body) and abstract tract variables (e.g. tongue body constriction location), but identifying the most appropriate coordinate system for articulatory control remains an open question. In this study, we advance a model in which holistic tongue shape modes represent a potential low-dimensional control basis for a simple dynamical law. We first discover empirical tongue shape modes in a cross-speaker coordinate space and then fit the model to ultrasound data on monophthongal vowels from 36 speakers of British English. A three-dimensional linear point attractor fitted to these time-varying modes recovers the dynamics of tongue shape with a high degree of accuracy. This suggests that complex tongue movement can arise from independent modes that linearly combine under simple dynamical laws, thereby relocating some complexity from the control signal to the coordinate system. We show that this framing explains why one-dimensional articulatory measures can exhibit observable nonlinearities, and hypothesize that shape modes are a plausible kinematic expression of underlying muscle synergies.
[NLP-30] HydraLLM : Adaptive Width Slicing for Elastic Text and Vision-Language Models EMNLP2026
【速读】： 该论文旨在解决大语言模型（LLM）与多模态大语言模型（LVLM）在异构硬件部署中面临的灵活性不足问题。现有模型通常采用固定规模的架构，难以适应不同计算资源、延迟要求及模型尺寸目标下的多样化部署需求。其核心解决方案是提出一种基于多查询注意力（MQA）的弹性架构——HydraLLM，其关键在于通过动态调整激活的注意力头数量，灵活控制嵌入层、注意力投影层、前馈网络及输出头的有效宽度，从而在单一参数集合内构建出嵌套的子网络家族。该方法通过静态到弹性转换流程实现，包含基于重要性的权重重排序与结合知识蒸馏的随机子网络训练。进一步地，该原理被拓展至多模态推理场景，引入可变宽度的弹性投影模块以匹配文本分支的激活宽度。实验表明，相较于仅扩展前馈网络（FFN）的弹性架构，HydraLLM支持更广泛的活跃参数配置；当50%注意力头激活时，其多模态性能达到7倍更大的Gemma 3 4B模型的CLIPScore水平；在仅25%头激活的情况下，生成吞吐量相比全宽配置提升75.3%，显著提升了资源利用率与部署适应性。
链接: https://arxiv.org/abs/2610.23455
作者: Oscar Kemme,Janek Haberer,Ali Hojjat,Olaf Landsiedel
机构: Hamburg University of Technology (TUHH)(汉堡工业大学); Kiel University(基尔大学); UNU-INWEH(联合国大学可持续发展与环境研究所)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted at the BabyLM Workshop at EMNLP 2026
Abstract:LLMs and LVLMs are increasingly deployed across heterogeneous hardware, ranging from data-center GPUs to resource-constrained edge devices. Their fixed-size architectures limit the ability to reuse a single model across deployment settings with different compute, latency, and model-size targets. We introduce HydraLLM, an elastic MQA-based architecture. Its number of active attention heads determines the effective widths of the embeddings, attention projections, feed-forward layers, and the output head, forming a nested family of subnetworks within a single parameter set. We introduce a static-to-elastic conversion pipeline that, starting from a pre-trained model checkpoint, includes importance-driven weight reordering and stochastic subnetwork training with knowledge distillation. We extend this principle to multimodal inference through an elastic projection module whose output width follows the active text width. HydraLLM covers a wider range of active parameter configurations than FFN-only elastic architectures. With 75% of heads active, HydraLLM achieves 2.2 percentage points higher TriviaQA accuracy than a state-of-the-art nested baseline configuration which has 17.25% more parameters. With 50% of heads active, our multimodal model achieves a CLIPScore comparable to that of the 7 \times larger Gemma 3 4B model. With only 25% of heads active, HydraLLM increases token-generation throughput by 75.3% over the full-width configuration.
[NLP-31] “You” Is Not Toxic: How LLM Labelers Rewrite the Webs Conversation Laws
【速读】： 该论文旨在解决在社交媒体对话中识别毒性内容（toxicity）时，因依赖生成式AI模型（如Claude Haiku）而引入的系统性偏差问题。随着Google Perspective API的停用，大型语言模型（LLM）正被广泛用于替代人工标注进行毒性检测，但研究发现这些模型存在显著缺陷：它们会将第二人称代词“you”误判为毒性信号，导致在深度回复（deeper replies）中出现虚假的毒性上升趋势。这一偏差源于模型对“你”字的过度敏感，而非真实语义中的攻击性，且该错误在不同平台（如Reddit、Gab、Wikipedia讨论页）均普遍存在。关键解决方案在于提出“Threadsafe”框架，通过结合人类标注与模型预测，利用模型得分动态选择需人工审核的评论，并基于人类标签校正模型预测结果，从而消除模型偏差。该方法不依赖于标签者的准确性，而是通过对统计规律（laws）进行直接验证，确保即使在模型存在系统性偏误的情况下，仍能获得无偏估计。实证结果显示，在约250个标注样本后，Threadsafe可维持名义覆盖率；当标注量达1000–8000时，其置信区间宽度仅为传统设计型监督学习（DSL）的75%，显著提升效率。最终，仅需1000个人工标注即可排除由Claude Haiku产生的虚假毒性上升规律。
链接: https://arxiv.org/abs/2610.23446
作者: Liang Yue,Eric A. Vance
机构: 未知
类目: ocial and Information Networks (cs.SI); Computation and Language (cs.CL)
备注: 43 pages, 4 figures, 44 tables. Labels, prompts and code: this https URL
Abstract:What the Web community believes about toxic conversations, such as that toxicity rises in longer and deeper threads, was measured mostly with Google’s Perspective API, which is shutting down; large language models (LLMs) will make the next measurements. We find that LLM labelers mistake second-person address for toxicity, and that this one bias manufactures laws that human labels do not show. On 1.25 million news comments with crowd labels, Claude Haiku flags 12% of the non-toxic comments that contain no “you” and 51% of those with four or more; a single crowd rater flags 6% and 10%. Deeper replies address someone more often, so under Haiku toxicity rises with reply depth (+0.93 percentage points per level), while under the crowd’s labels it slightly declines (-0.12). Rewriting “you” in the third person removes two thirds of the false rise; flagging only as many comments as the crowd does not remove it, and showing Haiku the parent comment enlarges it. All seven LLMs we test over-flag “you”, and at the crowd’s flag rate, without the parent, any false rise of the open ones vanishes at a fixed number of second-person words; Haiku also over-flags “you” on Reddit, Gab and Wikipedia talk pages. Agreement with human labels does not reveal the bias: we prove that a fixed Cohen’s kappa leaves a law nearly unconstrained. We therefore certify the law instead of the labeler. Threadsafe uses the labeler’s scores to choose which comments human raters label and to predict the others, and corrects the predictions with the human labels; linear laws are estimated without bias whatever the labeler does. In simulated audits its intervals for the laws keep nominal coverage from 250 labels on, where uniform-sample methods such as PPI++ fall short, and at 1,000 to 8,000 labels are 0.75 times as wide as those of design-based supervised learning (DSL); a thousand human labels rule out Haiku’s rising law.
[NLP-32] ScoutRoute: Efficient Mixture-of-Experts Routing Using Internal Language Model Signals
【速读】： 该论文旨在解决异构大语言模型（Large Language Models, LLMs）集合中高效专家路由（expert routing）的挑战，尤其针对由独立训练的多个模型构成的混合专家（Mixture-of-Experts, MoE）系统。现有路由方法通常依赖联合训练的门控网络、基于提示的预测器或计算开销巨大的多模型推理，限制了其灵活性与效率。本文提出ScoutRoute，一种轻量级的后处理路由框架，其核心创新在于通过单次前向传播从一个低成本的“侦察模型”（scout model）中提取内部语言模型信号，以预测各专家的胜任能力。具体而言，利用隐藏状态表示通过轻量级逻辑探针（logistic probes）估计跨模型的专家胜任度，同时结合从输出对数几率（logits）衍生的置信度度量，构建领域感知的路由策略，从而在不修改底层语言模型的前提下实现高效的专家选择。实验在五个开源语言模型上，覆盖数学推理、代码生成、常识推理及多学科知识等四个基准数据集进行验证，结果表明ScoutRoute在性能上持续优于单一路由信号，且仅需一次侦察模型推理。此外，所提出的胜任度探针在校准性方面显著优于传统置信度度量，证明了语言模型内部表示具有可迁移性，能够为异构模型间的高效、精准路由提供可靠依据。
链接: https://arxiv.org/abs/2610.23443
作者: Rajdeep Pandhare,Delzaan Bharucha
机构: Rajiv Gandhi Institute of Technology (拉吉夫·甘地技术学院)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 10 pages, 1 figure, 6 tables. Published in Advances in Artificial Intelligence and Machine Learning, 2026
Abstract:Efficient expert routing is a fundamental challenge in Mixture-of-Experts (MoE) systems, particularly for heterogeneous collections of independently trained Large Language Mod els (LLMs). Existing routing approaches typically rely on jointly trained gating networks, prompt-based predictors, or computationally expensive multi-model inference, limiting their flexibility and efficiency. This paper presents ScoutRoute, a lightweight post-hoc routing framework that predicts expert competence using internal language model signals extracted from a single forward pass through an inexpensive scout model. Hidden-state representations are used to estimate cross-model competence through lightweight logistic probes, while logit-derived confidence measures are incorporated within a domain-aware routing strategy for efficient expert selec tion without modifying the underlying language models. ScoutRoute isevaluatedon fiveopen-weightlanguagemodels across fourbenchmark datasets covering mathematical reasoning, code generation, commonsense reasoning, and multidis ciplinary knowledge. Experimental results show that ScoutRoute consistently outperforms individual routing signals while requiring only a single scout-model inference. The proposed competence probes also achieve superior calibration compared with conventional confidence measures, demonstrating that internal language model representations provide transferable information for accurate and efficient expert routing across heterogeneous language models
[NLP-33] On the Tip of the Tongue: First-Token Cueing as a Behavioral Probe of Recall Failures in LLM s
【速读】： 该论文旨在解决大语言模型在闭卷设置下回答事实性问题时频繁出现错误的问题，尤其关注那些并非由知识缺失导致的错误。研究表明，许多错误本质上是“首词回忆失败”（first-token recall failure）：模型在生成答案的第一步选择了错误的起始词，但一旦被提示正确的首词，便能顺利完成答案并识别其正确性。解决方案的关键在于提出“首词提示”（First-Token Cueing）这一行为探测方法，通过预先填充正确答案的首词来检验模型的潜在能力。实验发现，模型在首词位置表现出最高的不确定性，且在所有可能的正确前缀中，仅首词即可恢复最大比例的错误；在16个开源模型中，该方法使英语多词答案的错误纠正率提升至五分之一至一半，并在七种语言的多语言基准测试中均显著提高准确率。此外，这些可纠正的错误大多也能通过模型自身采样的推理过程解决，表明其已编码相关知识，但未能自主生成。模型能识别正确的起始部分却无法自发产生，说明其具备对正确起始的判断能力，但缺乏从初始状态启动正确生成路径的能力。该研究揭示了事实性错误中一个可恢复的子集存在于解码的首个步骤，为推理时方法（inference-time methods）提供了重要优化切入点。
链接: https://arxiv.org/abs/2610.23400
作者: Jaewon Sok,Jewon Yeom,Seonghyeon Park,Jeongjae Park,Hwiyeong Lee,Taesup Kim
机构: Seoul National University (首尔国立大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models often answer factual questions incorrectly in closed-book settings, and such errors are usually attributed to missing knowledge. Yet models often encode answers they fail to produce. We show that many closed-book errors are first-token recall failures: the model commits to a wrong first token, but once given the right one it completes the answer and often recognizes it as correct. We study this with First-Token Cueing, a behavioral probe that prefills the response with the first token of the gold answer. Model uncertainty peaks at the first answer token, where it already predicts correctness, and among gold prefixes of any length the first token alone accounts for the largest share of recovered errors. Across 16 open models, this one token corrects between a fifth and a half of the errors on multi-token answers in English QA and improves accuracy in all seven languages of a multilingual benchmark. Most of the errors it corrects are also solved by the model’s own sampled reasoning, a criterion for encoded knowledge, and models recognize correct starts they do not produce: they continue a correct start more confidently than an incorrect start of the same answer type and prior probability, yet rarely find it without the gold answer. These results locate a recoverable class of factual errors at the first decoding step, where inference-time methods may gain the most.
[NLP-34] Sentence-Level Topic Modeling with Lexical and Contextual Evidence
【速读】： 该论文旨在解决主题建模中的粒度困境问题：传统概率主题模型在词粒度上分配主题，而基于变压器（Transformer）的方法则在文档粒度上操作。前者依赖于孤立的词汇，缺乏足够的上下文信息以准确识别主题；后者虽能捕捉整体语义，却忽略了真实文档通常涵盖多个主题的事实。其解决方案的关键在于提出SentenceETM，一种新型神经主题模型，通过联合利用词汇与上下文证据，为每句话分配主题。具体而言，每个主题由词汇分布（基于词嵌入定义）和句子嵌入上的高斯分布共同表征，后者集中于特定主题的低维子空间。句子的主题同时生成其词汇和嵌入，并通过精确边缘化实现后验融合，从而整合双重证据。文档中主题比例采用摊销变分推断进行估计，而高斯参数则通过闭式解更新。实验表明，SentenceETM在句级主题分配任务上持续优于现有先进方法，对总统辩论案例的研究进一步揭示了句级主题如何精细展现候选人对同一议题的不同处理方式。
链接: https://arxiv.org/abs/2610.23388
作者: Federica Granese,Serena Villata,Charles Bouveyron
机构: 未知
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Topic modeling faces a fundamental granularity dilemma: probabilistic topic models assign topics at the word level, whereas transformer-based methods operate at the document level. The latter overlooks that real-world documents typically cover multiple topics, whereas the former relies on individual words, which rarely provide sufficient context to identify them. We introduce SentenceETM, a neural topic model that assigns a topic to each sentence by jointly using lexical and contextual evidence. Each topic is described both by a distribution over the vocabulary, defined through word embeddings, and by a Gaussian over sentence embeddings concentrated on a topic-specific low-dimensional subspace. The topic of a sentence generates both its words and its embedding and is marginalized exactly, so that its posterior combines the two sources of evidence. Document proportions are inferred with amortized variational inference, while the Gaussian parameters are re-estimated in closed form. Experiments show that SentenceETM consistently improves sentence-level topic assignment over state-of-the-art approaches, and a case study of a presidential debate shows how sentence-level topics reveal how candidates handle the same subjects.
[NLP-35] Language Models Learn to Improve Language Models
【速读】： 该论文旨在解决大语言模型（LLM）在自动化人工智能研究中实现递归自我改进时面临的两大挑战：一是基于真实实验反馈训练模型生成有效改进方案的成本过高，尤其在采用强化学习（Reinforcement Learning, RL）时需进行数十万次提案评估，每次评估可能涉及完整的模型训练，导致计算开销巨大；二是替代性反馈机制（surrogate feedback）在策略优化过程中因提案分布变化而产生偏差，导致对提议效果的高估，从而降低反馈可靠性。其解决方案的关键在于提出一种名为LM2LM的方法，通过引入一个冻结的“研究世界模型”（Research World Model, RWM），该模型基于上下文中的真实实验记录预测改进提案的有效性，作为强化学习的奖励模型，从而实现可扩展的低成本反馈。为确保反馈可靠性，系统会定期执行由RWM高分排序的当前策略提案，并将实际测量结果回填至RWM的上下文，动态校准其预测能力。实验表明，LM2LM在八个自回归与扩散型语言模型开发环境中均表现优异，使Qwen3.8-27B在未见任务上超越GPT-6-sol，且在仅需33%更低的数据生成成本下达成此性能，同时具备对训练阶段未见模型家族的良好泛化能力。
链接: https://arxiv.org/abs/2610.23344
作者: Minhua Lin,Zijun Wang,Zewen Liu,Zhaotian Weng,Bing He,Zhan Shi,Benoit Dumoulin,Xiang Zhang,Suhang Wang,Hanqing Lu
机构: The Pennsylvania State University(宾夕法尼亚州立大学); UC Santa Cruz(加州大学圣克鲁兹分校); Emory University(埃默里大学); UC Santa Barbara(加州大学圣巴巴拉分校); Amazon(亚马逊)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) are increasingly used to automate AI research, including language-model (LM) development, offering a potential path toward recursive self-improvement. However, turning experimental feedback into a learned ability to propose effective improvements across models faces two challenges. First, training such an ability with real feedback is expensive: reinforcement learning (RL) requires evaluating tens of thousands of proposals, each potentially requiring a full model-training run. Even at ten GPU-hours per run, reward collection alone would consume hundreds of thousands of GPU-hours, motivating a cheaper surrogate. Second, surrogate feedback can become unreliable as policy optimization favors overestimated proposals when the proposal distribution shifts. To address these challenges, we propose LM2LM, which trains LLMs with RL to generate effective LM improvement proposals. To provide scalable feedback, we use a research world model (RWM), a frozen LLM that predicts proposal gains from real experimental records provided in context, as the reward model for RL. To maintain feedback reliability, we periodically execute current-policy proposals ranked highly by the RWM and add their measured outcomes to its context. Across eight autoregressive and diffusion LM development environments, real-execution evaluations on held-out tasks show that LM2LM enables Qwen3.8-27B to outperform GPT-6-sol, even when GPT-6-sol is given 300 real experimental records in context, while requiring 33% lower data-generation cost. LM2LM also generalizes to a model family unseen during training.
[NLP-36] Robust Function Calling for LLM Agents Under Interface Drift
【速读】： 该论文旨在解决生成式 AI（Generative AI）在实际部署中因工具接口漂移（interface drift）导致的函数调用系统可靠性下降问题。具体而言，尽管底层工具语义保持不变，但工具描述、参数模式及候选工具列表等界面信息在推理阶段可能发生变更，从而引发模型错误调用或无法正确识别目标工具。针对这一问题，论文提出了一种轻量级、无需训练的修复层 SchemaShield-Lite，其核心解决方案包括：模式验证、缺失工具名的候选集重试机制以及单次约束性修复。实验结果表明，在 Qwen3.5-9B 模型上，该方法将 DICE-300 准确率从 0.717 提升至 0.813，BFCL-200 准确率从 0.580 提升至 0.805，成功恢复了 19/22 和 33/45 的接口漂移引发的失效案例，且未引入任何修复副作用。消融实验进一步揭示，候选集漂移（尤其是干扰性工具的引入）是造成性能下降的主要原因，而仅描述或模式修改的影响相对较小。研究证实，基于验证与修复的机制可作为已部署大语言模型代理使用工具时一种实用且高效的鲁棒性增强方案。
链接: https://arxiv.org/abs/2610.23342
作者: Aniruddha Chattopadhyay,Kanishka Halder,Partha Sarathi Roy
机构: 独立研究员(Independent Researcher); 独立研究员(Independent Researcher); 独立研究员(Independent Researcher)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Software Engineering (cs.SE)
备注: 16 pages, 1 figure, 6 tables. Peer-reviewed and accepted at NGEN-AI 2026 (International Conference on Next Generation AI Systems). This is the pre-review submitted version
Abstract:Tool-augmented language-model agents rely on interface text: tool descriptions, argument schemas, and the candidate tool list shown at inference time. These interfaces often change after deployment even when the underlying tool semantics remain fixed, creating a practical reliability problem for function-calling systems. We study this failure mode as interface drift and evaluate whether a lightweight, training-free repair layer can recover robustness without retraining the base model. Our method, SchemaShield-Lite, combines schema validation, candidate-set retry for missing tool names, and one-shot constrained repair. On Qwen3.5-9B, it raises DICE-300 accuracy from 0.717 to 0.813 and BFCL-200 accuracy from 0.580 to 0.805, recovering 19/22 and 33/45 drift-induced failures with no observed repair harms. Naive retry does not explain the gains. Ablations show that candidate-set drift, especially distractor tools, accounts for most observed damage, while description and schema edits alone are less harmful. The results position validation-and-repair as a practical robustness layer for deployed LLM-agent tool use.
[NLP-37] Self-Reflective Chain-of-Thought (SRCoT): Evidence-Based Text Generation via a Closed-Loop Retrieve–Reflect–Revise Pipeline on Edge LLM s
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在生成文本时缺乏来源支持或与局部操作约束冲突的问题，即模型易产生幻觉（hallucination）或生成不符合实际情境的输出。其核心解决方案是提出一种模块化架构——自省式思维链（Self-Reflective Chain-of-Thought, SRCoT），该架构通过构建“检索—反思—修订”闭环机制，将环境状态、持久知识、累积经验与短期记忆等要素分离并以溯源标签形式存储，实现对生成过程的显式控制。关键在于引入多轮迭代的反思机制，利用BM25检索提供本地上下文，并在后续迭代中暴露中间草稿、信息缺失标记、后续查询及可选的论点—证据映射，从而驱动检索更新、内容修订、澄清或主动回避决策。该架构强调报告级别的可追溯性与本地执行能力，虽未证明其在减少幻觉、临床有效性或操作安全性方面的相对优势，但展示了可审计的控制流程，其主要贡献在于提供了一种可解释、可审查的生成控制范式，而非新型模型。
链接: https://arxiv.org/abs/2610.23311
作者: Fabian Broszio,Michael Hoffmann,Adrian Paschke
机构: Freie Universität Berlin(柏林自由大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 28 pages, 2 figures. Extended technical account; a related brief report based on overlapping work has been submitted to Frontiers in Disaster and Emergency Medicine. The two manuscripts do not represent independent empirical studies
Abstract:Large language models (LLMs) can generate fluent outputs that lack source support or conflict with local operational constraints. We introduce Self-Reflective Chain-of-Thought (SRCoT), a modular architecture that organizes generation as a closed Retrieve–Reflect–Revise loop. SRCoT separates environment state, persistent knowledge, accumulated experience, and short-term memory into provenance-labeled stores. BM25 retrieval supplies local context; subsequent passes expose intermediate drafts, missing-information markers, follow-up queries, and optional claim–evidence mappings. These artifacts inform retrieval, revision, clarification, or abstention decisions. We examine the architecture through qualitative, trace-based case studies in two fully synthetic settings: emergency medical services and Munich-Sim, an emergency-operations simulation. Local prototype runs on an Apple M4-family laptop with 128 GB memory illustrate source-labeled documentation, explicit gap marking, and structured dispatch-style outputs. The traces also reveal provenance-assignment errors, incomplete evidence coverage, and unsafe assumptions that remain possible despite prompting constraints. These cases demonstrate report-level inspectability and local execution, not comparative superiority, hallucination reduction, clinical validity, or operational safety. The architecture separates retrieval and control from the model interface, but behavior across different models remains to be evaluated. SRCoT is a research proof of concept requiring human review, with its principal contribution lying in an inspectable control procedure rather than a new model.
[NLP-38] Inferring L1 Backgrounds from L2 English with LLM Token Probabilities
【速读】： 该论文旨在解决非母语者语言识别（Native Language Identification, NLI）问题，即通过分析非母语者在第二语言（L2）中的语言产出，推断其第一语言（L1）背景。传统方法通常采用直接分类策略，但存在特征表达能力有限的问题。本文提出一种基于分词概率（token probability）的新型解决方案，核心思想是将每个文本样本视为一个以指定候选L1为前缀提示（prompt prefix）的延续，通过预训练大语言模型（LLM）计算该文本在给定提示下的生成概率，进而选择得分最高的候选L1作为预测结果。该方法的关键在于利用预训练语言模型对不同L1背景所产生语言模式的细微差异具有敏感性的特性，通过分词概率建模挖掘隐藏在语言行为中的跨语言特征。实验在TOEFL9、Write and Improve 2024和BULATS-L1三个数据集上进行，使用Llama 3.1 8B与70B模型验证，结果表明该方法显著优于传统的基于提示的分类方法，尤其在8B小模型上提升明显；而70B大模型则在所有数据集上均达到当前最优性能，证明了预训练语言模型的分词概率中蕴含着可用于区分不同L1背景的高价值信息。
链接: https://arxiv.org/abs/2610.23309
作者: Ahmet Yavuz Uluslu,Mark Gales,Kate Knill,Gerold Schneider
机构: ALTA Institute, Department of Engineering, University of Cambridge (剑桥大学工程系); Department of Computational Linguistics, University of Zurich (苏黎世大学计算语言学系); PRODAFT, Yverdon-les-Bains, Switzerland
类目: Computation and Language (cs.CL)
备注:
Abstract:Native language identification (NLI) infers a non-native speaker’s first language (L1) from their second language (L2) production. We investigate a token probability approach that scores each text sample as a continuation of a prompt prefix specifying a candidate L1, then predicts the candidate with the highest score. We evaluate our approach using Llama 3.1 8B and 70B on three datasets: TOEFL9, Write and Improve 2024, and BULATS-L1. We compare its performance with previous prompting-based classification approaches. Token-probability scoring substantially improves on direct classification, particularly with the 8B model, while the 70B model achieves state-of-the-art performance across all datasets. Our results indicate that pretrained LLMs’ token probabilities contain information useful for distinguishing L1 backgrounds from L2 English production.
[NLP-39] SimOPD: Relational Knowledge Transfer for On-Policy Distillation
【速读】： 该论文旨在解决在线策略蒸馏（on-policy distillation, OPD）中如何有效提取教师模型内部知识以提供高质量监督信号的问题。现有方法存在两大局限：一是仅依赖词元概率的监督信号难以充分挖掘教师模型的深层语义信息，二是直接匹配隐藏状态会迫使学生模型迁移到教师的特征空间，导致跨架构迁移困难。针对上述问题，本文提出SimOPD，其核心创新在于借鉴对比学习中实例间关系建模的思想，通过构建学生生成响应的配对记忆库，从教师内部表示中提取基于响应相似性的监督信号，并以相似性分布的形式进行传递。学生模型在自身表示空间内匹配这些分布，从而在不进行跨模型投影的前提下保留了表征自由度，实现了更高效的知识迁移。实验表明，该方法在同架构与跨架构设置下分别相较最强基线提升2.6和3.1个百分点，显著优于现有方法。
链接: https://arxiv.org/abs/2610.23306
作者: Kai Tang,Xing Zheng,Haobo Wang,Guangcheng Zhu,Shenzhi Yang,Zhengqing Zang,Yingfan Ma,Bowen Song,Weiqiang Wang
机构: Zhejiang University (浙江大学); Ant Group (蚂蚁集团)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:A key challenge in on-policy distillation (OPD) is extracting effective supervision from teacher knowledge. Supervision from token probabilities may underutilize internal knowledge, while direct hidden-state matching forces the student to shift to the teacher’s feature space and complicates cross-architecture transfer. Inspired by contrastive learning’s use of relationships between instances, we propose \textbfSimOPD to transfer teacher internal knowledge through response similarities. Specifically, SimOPD uses paired memory banks of student-generated responses to extract supervision from the teacher’s internal representations as similarity distributions. The student then matches these distributions within its own representation space, retaining representational freedom and removing the need for cross-model projection. Extensive experiments show that SimOPD improves the overall average over the strongest baselines by up to 2.6 and 3.1 percentage points in same- and cross-architecture settings.
[NLP-40] Verbalising Models Computation for Chain-of-Thought Faithfulness Detection
【速读】： 该论文旨在解决当前链式思维（Chain-of-Thought, CoT）验证方法在解释模型推理不忠实性（unfaithfulness）产生机制方面的局限性。现有方法通常基于模型输出或静态激活值进行检测，无法揭示推理过程中的不忠实性如何随生成过程逐步演化。为此，论文提出VERITAS框架，通过引入逐层雅可比投影（layer-wise Jacobian projection），实现对模型内部计算过程的语义化表征（verbalised account），动态追踪不同推理锚点（reasoning anchors）在各层间的涌现、传递与最终在输出中的显现。其解决方案的关键在于：利用沿推理轨迹提取的结构化特征训练分类器，从而提供关于推理偏离忠实性的发生位置与机制的深入洞察。实验结果表明，VERITAS在FaithCoT-Bench和BonaFide的四个数据集上显著优于LLM-as-Judge，平衡准确率提升28.4%，平均优于激活探针（activation probing）16.0%。进一步分析显示，该方法具备良好的跨数据集与跨模型架构泛化能力。研究证明，将模型计算过程语义化能够为链式思维忠实性提供更丰富、更具可解释性的分析视角，推动从被动检测向机制性理解推理失败的根本原因迈进。
链接: https://arxiv.org/abs/2610.23275
作者: Yujing Wang,Yulan He,Wenya Wang,Hanqi Yan
机构: King’s College London(伦敦国王学院); Nanyang Technological University(南洋理工大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Current Chain-of-Thought (CoT) verification methods detect reasoning faithfulness from model outputs or static activations, but are limited in explaining how unfaithfulness develops during generation. We introduce VERITAS to verbalise the model’s computation: how different reasoning anchors emerge, transfer across layers, and surface in the output. VERITAS adopts the layer-wise Jacobian projection to produce a verbalised account of the model’s internal reasoning evolution at each generated CoT token. By training a classifier on the extracted features along the traces, we show that these structured features provide insightful perspectives on how and where reasoning diverges toward unfaithfulness. Experiments on four datasets from FaithCoT-Bench and BonaFide show that VERITAS achieves a significant performance gain over LLM-as-Judge, with a 28.4% gain in balanced accuracy, and an average improvement of 16.0% over activation probing. Further analysis shows VERITAS can be generalised to different datasets and model architectures. Our work shows that verbalising model computation provides a richer and more interpretable lens on CoT faithfulness, moving beyond detection toward a mechanistic understanding of how reasoning fails.
[NLP-41] Reasoning Models Can Control Their Chains of Thought When They Tell Themselves To
【速读】： 该论文旨在解决生成式模型在链式思维（Chain-of-thought, CoT）推理过程中对推理过程可控性不足的问题，即模型难以有效遵循用户在提示中设定的推理约束。现有评估方法通过检测模型是否遵守用户显式给出的约束来衡量其可控性，但结果普遍显示可控性极低。本文提出，这种低可控性可能并非源于模型内在能力缺陷，而是由于用户提示中对约束的表述方式过于直接或不充分所致。为此，论文提出了“自指导链式思维可控性”（Self-Instructed CoT Controllability）这一新评估范式，将约束以自指导（self-instruction）的形式嵌入模型自身的推理链条中，从而测试其内在遵循能力。实验表明，在自指导条件下，模型的可控性显著优于传统用户指令设置；即使仅通过一个通用提醒（如“请遵循用户约束”），也可大幅提升可控性。此外，研究发现有效的自指导可通过简单的用户提示引导生成，实现无需直接访问推理通道的间接控制。最后，研究揭示了在真实监控场景下，利用自指导可实现对监控机制的规避，暴露出当前链式思维系统在可监控性方面的潜在安全风险。因此，该方案的关键在于通过将约束内化为模型自我生成的指导语，从而增强其对推理过程的自主可控能力。
链接: https://arxiv.org/abs/2610.23257
作者: Uriel Dolev,Idan Schwartz,Yoav Goldberg
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 26 pages, 19 figures, 18 tables. Code and data: this https URL
Abstract:Chain-of-thought (CoT) controllability, the ability of reasoning models to control properties of their CoT, is consequential for both their usefulness and reliable oversight. Recent work evaluates this capability by measuring whether models follow CoT constraints given in the user prompt, and reports very low controllability across models. We argue that this low controllability may not reflect an inherent inability of reasoning models to control their CoT, but instead result from naively specifying constraints in the user prompt. To test this, we introduce Self-Instructed CoT Controllability: a complementary evaluation that places the constraint as a self-instruction inside the model’s CoT and measures its ability to follow it. We show that, under self-instruction, models achieve significantly higher controllability than with standard user instructions. Even a generic reminder to follow the user’s constraint substantially improves controllability. Moreover, effective self-instructions can be bootstrapped into the reasoning through a simple instruction in the user prompt, enabling improved control even without access to the reasoning channel. Finally, using self-instruction, we demonstrate in a realistic monitoring setup that CoT controllability can be leveraged for monitor evasion, posing a risk to CoT monitorability.
[NLP-42] Sample Mix or Verify? Allocating Test-Time Compute under Imperfect Verifiers
【速读】： 该论文旨在解决在推理阶段通过增加计算资源提升语言模型与智能体准确性的过程中，如何高效分配采样数量、选择模型来源及验证成本的难题。现有方法依赖经验法则进行决策，缺乏对验证器（verifier）行为偏差（如误接受错误答案、误拒正确答案）和多模型间共现盲点、验证调用相关性等复杂因素的建模，导致预算分配次优。其核心解决方案是提出SPLIT框架，基于一个精确的数学公式，量化在“接受/拒绝”与“最高得分选择”策略下提交正确答案的概率，该公式显式考虑了任务难度差异、跨模型共享的盲区以及验证调用间的相关性。SPLIT通过经验贝叶斯方法从少量标注的预实验数据中拟合该公式，并据此为未见过的任务规划最优预算分配。在涵盖数学、科学问答、代码生成与软件工程代理的七个基准测试中，26个模型、175个代理系统及六类真实验证器（如单元测试、多数投票、大语言模型评分器、奖励模型）的实验表明，SPLIT在所有场景下均能接近事后最优分配（误差仅0.02–0.8个百分点），而所有固定规则均至少损失2.5个百分点。其后悔值（regret）仅取决于配置排序能力，且在计算、价格、延迟成本及在线环境变化下仍保持近似最优。此外，该公式揭示了为何增加大语言模型评审者数量反而可能降低精度，同时证明在验证器限制成功率时，SPLIT优于幂律拟合，在部分特定条件下仍存在失效边界。
链接: https://arxiv.org/abs/2610.23255
作者: Srinivasan Iyengar
机构: Microsoft Research India(微软研究院印度)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 27 pages (18 main + appendix), 14 figures, 11 tables. Code will be released
Abstract:A common way to make language models and agents more accurate is to spend more compute at inference time: generate many candidate solutions and submit the one a verifier prefers. Success then requires that some candidate is correct and that the verifier picks it. The first grows predictably with the number of samples; the second does not, because real verifiers (unit tests, majority votes, LLM judges, reward models) accept wrong answers, reject right ones, and make correlated mistakes. With a fixed budget, a practitioner must decide how many samples to draw, which models to draw them from, and how much to spend on verification, and today these choices are made by rules of thumb. We introduce SPLIT, which makes them from a small labelled pilot. It rests on an exact formula for the probability of submitting a correct answer under accept/reject and top-score selection that accounts for varying task difficulty, blind spots shared across models, and correlated repeated verifier calls. SPLIT fits the formula by empirical Bayes and uses it to plan budgets on unseen tasks. Across seven benchmarks spanning math, science QA, code and software-engineering agents, 26 models, 175 agent systems and six kinds of real verifiers, all from public releases, SPLIT lands within 0.02-0.8 percentage points of the best allocation in hindsight, while every fixed rule we test loses at least 2.5 points in some setting. Regret depends only on how well SPLIT ranks configurations, and it stays near-optimal under compute, price and latency costs and online. The formula also explains when asking more LLM judges lowers precision. As a forecaster far beyond the pilot, SPLIT beats power-law fitting when the verifier caps success, and we report where it loses.
[NLP-43] Ignored Not Refused: What Low Prompt-Injection Success Rates Hide in Small Tool-Using Agents
【速读】： 该论文旨在解决生成式 AI（Generative AI）代理在使用外部工具时因读取不可信外部内容而面临的间接提示注入（Indirect Prompt Injection, IPI）安全问题。现有基准测试通过攻击成功率（Attack Success Rate, ASR）评估安全性，但其假设所有攻击失败均代表防御成功，这一假设存在偏差。研究通过审计11个开源权重代理（2-14B参数）在InjecAgent和AgentDojo上的3.3万条轨迹发现，大多数失败的攻击实际被代理“忽略”（76–80%），而非主动拒绝（17–18%）。此外，在相同模型权重下，仅改变系统提示即可使InjecAgent的ASR从0.2%跃升至43%，表明系统提示对IPI防御能力具有显著影响。进一步提出的攻击-任务结果矩阵（Attack-Task Outcome Matrix, ATOM）显示，在AgentDojo的21.7%会话中，代理虽未被劫持（保持安全），但未能完成用户任务。这表明低ASR主要反映代理对攻击的沉默忽略，而非有效防御。因此，论文提出评估应不仅报告ASR，还需明确攻击失败的具体机制及用户任务是否达成。研究公开了标注数据集、判断模型提示及分析代码，以推动更全面的IPI安全评估。
链接: https://arxiv.org/abs/2610.23254
作者: Farhan Zarif,Kazi Shahid Ahmed Galib,Synthia Akter Promi,Shabnam Sharmin Tarin,Ariyan Hossain,Sadif Ahmed,Farig Yousuf Sadeque
机构: BRAC University(布拉克大学)
类目: Computation and Language (cs.CL); Cryptography and Security (cs.CR)
备注: 17 pages, 3 figures, 11 tables
Abstract:Tool-using agents read untrusted external content, which exposes them to indirect prompt injection (IPI), where hidden instructions hijack the agent’s tool calls. Benchmarks measure IPI security by the attack success rate (ASR), which treats every failed attack as a successful defense. To test this assumption, we audit 33,000 trajectories from eleven open-weight agents (2-14B parameters) on InjecAgent and AgentDojo. A label-blind LLM judge shows that most failed attacks are ignored (76-80%) rather than refused (17-18%). With identical model weights, changing only the system prompt shifts InjecAgent ASR from 0.2% to 43%. Our Attack-Task Outcome Matrix (ATOM) further shows that in 21.7% of AgentDojo episodes, the agent stayed safe but did not complete the user’s task. A low ASR therefore mostly reflects agents silently ignoring injections, not refusing them. Evaluations should report how attacks fail and whether the user’s task is completed, alongside ASR. We release the labeled corpus, judge prompt, and analysis code.
[NLP-44] Who Pays for Efficient Inference? Confidence-Based Small-to-Large LLM Routing Across Tasks and Languages
【速读】： 该论文旨在解决小模型到大模型的生成式AI（Generative AI）路由问题，即如何在保证推理准确率的前提下，通过将复杂输入分配至大模型、简单输入保留在小模型中，以降低整体推理成本。其核心挑战在于：现有基于置信度（confidence）的路由策略在多语言任务中的可靠性尚未得到充分验证，且置信度是否能有效识别真正需要升级处理的样本仍不明确。论文的关键解决方案是系统性评估基于全局置信度的路由策略在三种模型家族、四种语言及三类任务下的表现，覆盖72种配置组合与42,288次预测结果，并与随机路由、等额配额置信度路由及理想最优路由进行对比。研究发现，置信度路由在阅读理解（Belebele）和主题分类（SIB-200）任务中表现良好，平均校正检测AUC分别为0.612和0.705，在50%升级率下相较随机路由提升1.40–4.67个百分点（Belebele）和2.54–5.63个百分点（SIB-200）；但在知识与推理任务（MMLU-ProX-Lite）中，平均AUC仅为0.510，置信度路由未能带来稳定性能提升，因小模型常对错误预测表现出高置信度。此外，全局置信度路由导致不同语言间出现高达30.33个百分点的计算资源分配偏差，而等额配额路由可消除此语言偏倚，仅带来微小的整体准确率损失。因此，该研究揭示了置信度路由具有显著的任务依赖性与语言不均衡风险，其有效性需结合具体任务特性评估，且应引入公平性约束以避免资源分配失衡。
链接: https://arxiv.org/abs/2610.23249
作者: Md Sabbir Hossen,Anichur Rahman,Pabon Shaha,Airin Afroj Aishi,Andrew H. Sung,Md Shohel Rana
机构: Bangladesh University(孟加拉国大学); Georgia Southern University(乔治亚南方大学); University of Southern Mississippi(南密西西比大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Small-to-large language model routing can reduce inference cost by escalating difficult inputs while retaining easier ones on a smaller model. Its success, however, depends on identifying examples that benefit from escalation, and the reliability of confidence across multilingual tasks remains unclear. We present a systematic evaluation of confidence-based routing across three model families, four languages, and three task types, covering 72 model, dataset, language, and scale configurations and 42,288 predictions. We compare global confidence routing with random routing, equal-quota confidence routing, and an oracle upper bound. Confidence identifies beneficial escalations on the Belebele reading comprehension and SIB-200 topic classification datasets, with mean correction-detection AUCs of 0.612 and 0.705. At 50% escalation, it improves over random routing by 1.40–4.67 percentage points on Belebele and 2.54–5.63 points on SIB-200. In contrast, the mean AUC on the MMLU-ProX-Lite knowledge and reasoning dataset is only 0.510, and routing provides no consistent improvement over random selection because small models often remain highly confident while incorrect. Global confidence routing also creates language escalation gaps up to 30.33 percentage points, whereas equal-quota routing removes this disparity with little aggregate accuracy cost. These results show that confidence routing is task-dependent and can allocate computation unevenly across languages. Code is available at this https URL
[NLP-45] hink More or Read More? A Token-Budget Theory of Reasoning with Shared Context
【速读】： 该论文旨在解决生成式 AI（Generative AI）在多智能体辩论与自精炼等推理系统中，因信息交互导致的上下文资源消耗问题。核心挑战在于：在固定上下文窗口（token window）约束下，模型用于阅读他人输出或自身草稿的令牌数会直接挤占可用于实际推理的令牌数，从而形成“读取-推理”的资源竞争关系。论文提出了一种令牌预算理论（token-budget theory），通过将准确率建模为扣除读取与投票聚合所耗令牌后剩余推理预算的函数，并引入允许误差相关性的Condorcet陪审团模型，量化了这一权衡。其关键创新在于：该理论可对多智能体辩论中的最优代理数量给出闭式预测，且仅需单智能体实验数据即可估算，无需实际运行辩论过程；同时适用于异构模型组合。实证结果表明，在严格计算预算条件下（如仅能处理一至两个问题），该方法显著优于强基线。对于自精炼机制，理论同样适用，以自身草稿替代同伴消息作为输入。进一步分析揭示，通信只有在提升的准确率上限超过读取所牺牲的推理空间时才具价值，且随着窗口收紧，该条件愈发严苛。在五种开源模型与四种推理任务上的大规模对比显示，自一致性方法在93%的匹配窗口与群体规模设置下表现不逊于甚至超越部署级辩论系统，尤其在紧窗口下优势明显，而在宽松窗口下差距消失——完全符合理论预期。因此，研究结论强调：在共享上下文窗口下，模型剩余的推理空间（thinking room）比其所读取的内容更为关键。
链接: https://arxiv.org/abs/2610.23246
作者: Manh Nguyen,Sunil Gupta,Hung Le
机构: Deakin University (迪肯大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Reasoning systems increasingly improve answers by incorporating additional context into the model’s input, from multi-agent debate, where agents read their peers’ responses, to self-refinement, where a model revisits its own drafts. In these systems, communication consumes context as one agent’s output becomes another’s input. We study the regime where token reading and generation draw on one fixed token window, so every token spent reading is a token that cannot be spent reasoning. We develop a \textbftoken-budget theory that formalizes this trade-off, modeling accuracy as a function of the reasoning budget remaining after reading and aggregating votes with a Condorcet jury model that permits correlated errors. For multi-agent debate, the theory gives a closed-form prediction for the optimal number of agents. This quantity is estimated from single-agent measurements, allowing the optimal group size to be determined without running debate, and the same accounting extends to a heterogeneous set of different models. We empirically verify our theory and show that, at the predicted group size, our method outperforms strong baselines under the tightest calibration spend, where debate-calibrated rules can afford only one or two questions. Self-refinement obeys the same accounting, with its own draft in place of a peer message. The theory further predicts that communication pays only when the gain in the accuracy ceiling exceeds the reasoning room spent on reading, with this condition tightening as the window shrinks. Across five open models and four reasoning tasks, self-consistency matches or beats deployed debate at 93 % of 980 matched window-and-group-size settings, with the gap being largest under tight windows and nearly disappearing at a loose one, consistent with the theory. Under a shared window, how much room a model has left to \textitthink matters more than what it reads.
[NLP-46] MGE-PO: Entropy-Aware Credit Assignment for Mathematical Reasoning
【速读】： 该论文旨在解决生成式强化学习中策略优化（Policy Optimization, PO）时，对相同奖励响应赋予相同组相对优势（Group-Relative Preference Optimization, GRPO）所导致的生成不确定性未被充分利用的问题。其核心解决方案是提出MGE-PO方法，关键在于引入四象限（Four-Quadrant, 4Q）优势塑造机制，该机制融合了验证正确性与滚动生成熵（Rollout Generation Entropy）。具体而言，正确性决定修正项的符号，而熵值控制修正项的幅度：低熵的成功输出获得更大奖励，高熵的失败输出则遭受更重惩罚。通过基于在线熵的题目课程学习（Question Curriculum），利用缓存的滚动测量结果实现该信用分配规则。在Qwen3-8B-Base模型上经过八轮训练后，三种策略优化设置下引入4Q机制使六基准测试的宏平均准确率@32（macro avg@32）提升3.23%–3.69%，宏通过率@32（macro pass@32）提升4.24%–11.06%；在Qwen2.5-7B-Base上，完整配置将四基准测试的macro avg@32从41.5%提升至45.4%，验证了该方法的有效性。
链接: https://arxiv.org/abs/2610.23245
作者: Jihao Huang,Guangjia Chai,Zhirui Chen,Lei Wang,Yao Liu,Liu Qi,Pengjie Ding,Xinyang Liu,Chengxun Jia,Ling Shao,Hongyu Chen,Junchen Wan
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:GRPO assigns the same group-relative advantage to equally rewarded responses within a question, leaving their generation uncertainty unused. We introduce MGE-PO, centered on four-quadrant (4Q) advantage shaping that combines verified correctness with rollout generation entropy. Correctness determines the sign of an additional correction, while entropy controls its magnitude: lower-entropy successes receive larger bonuses and higher-entropy failures receive larger penalties. An online entropy-based question curriculum supports this credit rule using cached rollout measurements. At the final checkpoint after eight training epochs on Qwen3-8B-Base, adding 4Q across three policy optimization settings yields absolute gains of 3.23-3.69% in six-benchmark macro avg@32 and 4.24-11.06% in macro pass@32. On Qwen2.5-7B-Base, the complete configuration raises four-benchmark macro avg@32 from 41.5% to 45.4%.
[NLP-47] MultiAgent -OPD: End-to-End Training of Hierarchical Multi-Agent Systems with On-Policy Distillation
【速读】： 该论文旨在解决层次化多智能体系统中领导者（Leader）与子智能体（subagents）联合训练的难题，核心挑战在于最终答案仅提供间接反馈给子智能体，导致其难以获得有效的过程监督。现有方法依赖大语言模型（LLM）裁判对子智能体输出进行评分，但此类评分受裁判自身判断标准的影响，存在偏差。为此，本文提出 MultiAgent-OPD 框架，通过结合在线策略蒸馏（on-policy distillation）与强化学习（reinforcement learning），实现共享参数的领导者-搜索者-回答者团队端到端联合训练。其关键创新在于：首先在搜索者-回答者流程中训练一个冻结的教师模型（Teacher），以答案正确性为监督信号；在团队训练阶段，该教师模型对学生生成的搜索与摘要标记进行评分，为搜索者提供局部过程监督，同时领导者基于团队整体表现学习，回答者则根据候选答案的正确性优化。实验表明，在七个数据集上，该系统达到 38.68% 的宏平均准确率@4，优于 Prompted Critic 4.06 个百分点，也优于训练后的级联多智能体系统 C-MAS + RL 2.61 个百分点，且在四次尝试内解决更多问题。
链接: https://arxiv.org/abs/2610.23234
作者: Yuhan Sun,Jiaxin Zhao,Yongdong Luo,Yibo Hu,Meiguang Jin,Junfeng Ma,Weixun Wang,Weihang Pan,Zhongan Bi
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:In a hierarchical multi-agent system, a Leader assigns tasks and subagents carry them out. Jointly training the Leader and the subagents is difficult because a final answer gives only indirect feedback to the subagents. An LLM judge can score the performance of the subagents, but these scores depend on the judge’s own assessment of the quality of subagents’ responses. We present MultiAgent-OPD, which combines on-policy distillation with reinforcement learning to jointly train a shared-parameter Leader-Searcher-Answerer team end to end. We first train a Teacher in a Searcher-Answerer workflow using answer correctness. During team training, the frozen Teacher scores student-generated search and summary tokens to provide local process supervision for the Searchers, while the Leader learns from team success and the Answerer from candidate correctness. Across seven datasets, the trained system reaches 38.68% macro avg@4, exceeding Prompted Critic by 4.06 percentage points and C-MAS + RL, the trained cascaded multiagent system, by 2.61 percentage points. The team also solves more questions within four attempts.
[NLP-48] ReLaDrift: Recurrent Language Generation via Drifting for Test-Time Scaling
【速读】： 该论文旨在解决生成式语言模型在快速并行生成与迭代精炼之间难以兼顾的问题。传统漂移模型（Drifting Model）虽可通过单次函数评估生成样本，具备高效的并行生成能力，但其一次性生成的特性无法支持后续的迭代优化，限制了生成质量的提升。为此，论文提出基于漂移机制的循环语言生成方法（ReLaDrift），通过共享一个漂移模型从噪声中生成初始草稿，并在连续隐空间中进行多轮迭代精炼。其核心创新在于训练策略：通过采样迭代轮数，并仅在最后一轮应用漂移损失，同时在前序轮次中截断梯度传播，从而实现仅需单轮反向传播即可完成训练。这一设计使得模型在推理阶段可灵活调节计算预算与生成质量之间的权衡，突破了传统漂移模型在迭代过程中性能无法提升的局限。实验结果表明，在Sudoku Hard和GSM8K等任务上，ReLaDrift在不同推理预算下均达到或优于扩散模型与流模型等迭代生成方法的性能表现。
链接: https://arxiv.org/abs/2610.23225
作者: Yuki Otsuka,Daisuke Oba,Naoaki Okazaki
机构: Institute of Science Tokyo(东京科学研究所); AIST(日本产业技术综合研究所); NII LLMC(国立信息学研究所大语言模型研究中心)
类目: Computation and Language (cs.CL)
备注:
Abstract:Drifting Models generate a sample with a single function evaluation, making them attractive for fast parallel language generation, but their one-step formulation does not support iterative refinement. We propose Recurrent Language Generation via Drifting (ReLaDrift), which uses a shared Drifting Model to generate an initial draft from noise and iteratively refine it in a continuous latent space. We train this recurrence by sampling the number of rounds and applying the drifting loss only at the final round, stopping gradients through preceding rounds so that only one round requires backpropagation. Unlike the one-step Drifting Model, whose accuracy does not improve when iterated, ReLaDrift enables a trade-off between test-time computation and generation quality, as diffusion and flow language models do. On Sudoku Hard and GSM8K, it achieves competitive or superior performance compared with these iterative-generation approaches across a range of inference budgets. Our code is available at this https URL
[NLP-49] MAS-Bai: Linguistically Informed Adaptive Multi-Agent Translation for Bowen-Script Bai EMNLP2026
【速读】： 该论文旨在解决生成式 AI 在极低资源语言（如使用邦诺文书写系统的白语）中因预训练数据与目标语言书写系统不匹配而导致的翻译不可靠问题，尤其针对中文字符被音译借用而非意译所引发的严重字面翻译幻觉。其核心解决方案是提出一种基于语言学知识的自适应多智能体框架——MAS-Bai，通过动态路由机制将输入句分配至轻量级检索-翻译路径或包含词源与句法分析智能体的深层语文学路径，并引入轻量级反向翻译器实现循环一致性校验，作为推理阶段的可靠性信号。该方法在构建的6,536句白语-汉语语料库上进行评估，测试集达1,536句，相比标准检索增强生成（RAG）方法，在BLEU和COMET指标上分别提升8.11和6.85，显著改善了对邦诺文音译借词及白语特殊语序等难点的处理能力，且通过人类评估与反向验证分析进一步证实了自动评估结果的有效性。
链接: https://arxiv.org/abs/2610.23205
作者: Jingting Zheng(1),Hong Ren(1),Yuehan Cui(1),Shaolin Zhu(1),Deyi Xiong(1) ((1) TJUNLP Lab, School of Computer Science and Technology, Tianjin University, China)
机构: TJUNLP Lab, School of Computer Science and Technology, Tianjin University (天津大学)
类目: Computation and Language (cs.CL)
备注: 14 pages, 3 figures, Findings of EMNLP 2026
Abstract:Large Language Models (LLMs) remain unreliable for extremely low-resource languages whose writing systems conflict with high-resource pretraining priors. We study Bai-Chinese translation in the Bowen script, where Chinese characters are frequently borrowed phonetically rather than semantically, causing severe literal-translation hallucinations. We propose MAS-Bai, a linguistically informed adaptive multi-agent framework for Bowen-script Bai-Chinese translation. MAS-Bai routes each input through either a lightweight retrieval-translation path or a deeper philological path with etymological and syntactic agents, and uses cycle-consistency with a lightweight reverse translator as an inference-time reliability signal. We construct a 6,536-sentence Bai-Chinese corpus and evaluate on a 1,536-sentence test set. MAS-Bai achieves 16.23 BLEU and 62.15 COMET, improving over Standard RAG by 8.11 BLEU and 6.85 COMET. Forward LoRA, expert ablations and controlled routing experiments distinguish the roles of supervised adaptation, Bai-specific expertise and conditional routing. The largest gains occur on Bowen phonetic loans and divergent Bai word order, while human evaluation and reverse-verifier analyses support the automatic results.
[NLP-50] AMEND: Constraint-Guided Multi-Agent Text Detoxification with Isolated Critics
【速读】： 该论文旨在解决文本去毒（text detoxification）中毒性消除、语义保真与语言流畅性之间的平衡难题。现有方法通常采用无约束的标量优化目标，导致安全评分掩盖了语义损失，或依赖开放式的迭代辩论机制，易引发语义漂移且计算成本高昂。为此，本文提出AMEND框架，其核心在于将去毒过程解耦为角色条件生成、独立多维度批评以及以约束优先的选择三个阶段：通过无角色偏见的批评者评估多样化候选文本，并在排序前严格校验语义与流畅性可行性，确保输出质量；为进一步提升效率，引入AMEND Lite版本，仅在残留毒性片段存在时执行最多一轮针对性修复。实验结果表明，在ParaDetox数据集上，基于多角色的一轮选择使联合评价指标提升0.0508；在冻结确认测试集上，AMEND Lite以0.6597±0.0062的联合得分优于无约束多轮循环方法（+0.0489, p<0.05），同时减少28.4%的推理请求数量。关键创新点在于通过角色解耦与约束优先的筛选机制，在保障语义完整性的同时显著提升去毒效率与稳定性。
链接: https://arxiv.org/abs/2610.23183
作者: Tianyu Wang
机构: Soochow University (苏州大学)
类目: Computation and Language (cs.CL)
备注: 10 pages, 2 figures, 4 tables
Abstract:Text detoxification requires balancing toxicity removal, semantic preservation, and fluency. Existing methods typically optimize an unconstrained scalar objective—enabling safety scores to mask semantic loss—or rely on open-ended iterative debate prone to semantic drift and high computational cost. We propose AMEND, an inference-time framework that decouples detoxification into role-conditioned proposal generation, isolated multidimensional criticism, and constraint-first selection. AMEND evaluates diverse candidates with role-blind critics and enforces strict semantic and fluency feasibility before ranking by utility. To maximize efficiency, AMEND Lite restricts subsequent search to at most one targeted repair round only when residual toxic spans remain. On ParaDetox, one-round selection over diverse roles improves the joint evaluation metric by +0.0508 over the best single role. On a frozen confirmation test set, AMEND Lite achieves a joint score of 0.6597 \pm 0.0062 , outperforming unconstrained multi-round loops by +0.0489 ( p 0.05 ) while reducing inference requests by 28.4%.
[NLP-51] What LLM s Do in the Shadows: Reading Context Reliance from the Unembedding Dark Space
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在高风险场景中部署时，因生成内容与输入上下文不一致而产生的忠实性幻觉（faithfulness hallucinations）检测问题。现有方法通常通过对比上下文知识利用与参数化知识利用来识别此类幻觉，但其测量上下文依赖性往往需要高昂的计算开销，如生成注意力图、额外前向传播或训练辅助探测器。本文提出的关键解决方案是：通过将隐藏状态投影到“暗空间”（dark space）——即与解嵌入矩阵（unembedding matrix）弱相关的方向——来估算上下文依赖性。研究发现，该子空间不仅作为初始标记的注意力汇聚点，更主动参与上下文处理过程；其在隐藏状态范数中所占比例被定义为“暗物质比”（Dark Matter Ratio, DMR），可直接量化模型对上下文的利用程度。结合DMR与参数化知识得分，构建出名为DarkLens的轻量级忠实性检测器，仅需一次前向传播即可完成评估，无需额外推理开销或探测器训练。在六个开源权重的LLM（4B-13B参数）上，针对三个问答与摘要基准的实验表明，DarkLens在性能上达到或超越多种基线方法，展现出高效且精准的检测能力。
链接: https://arxiv.org/abs/2610.23182
作者: Alexandra Kuleshova,Andrey Volodichev,Daria Kotova,Alexey Zaytsev
机构: Applied AI Institute(应用人工智能研究院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Detecting faithfulness hallucinations in large language models (LLMs) is crucial for their safe deployment in high-stakes domains. Recent approaches address this by contrasting contextual against parametric knowledge utilization, yet measuring context reliance typically demands noticeable overhead, such as materializing attention maps, running additional forward passes, or training auxiliary probes. In this work, we show that context reliance can be estimated by projecting hidden states onto the dark space - directions that compose weakly with the unembedding matrix. Beyond its recognized function as an initial-token attention sink, we show that this subspace actively mediates context processing, and that its share of hidden-state norm - termed the Dark Matter Ratio (DMR) - provides a direct score of context utilization. Pairing DMR with a parametric knowledge score yields DarkLens, a lightweight faithfulness detector operating strictly within a single forward pass. Evaluated across three QA and summarization benchmarks on six open-weight LLMs (4B-13B parameters), DarkLens matches or outperforms a wide range of baselines without requiring extra forward passes or probe training.
[NLP-52] ree-Aware Posterior Distillation for Long-Horizon Agent ic Reinforcement Learning
【速读】： 该论文旨在解决长时程智能体强化学习中两个核心挑战：一是仅依赖最终奖励导致的回合级信用分配稀疏且方差高，二是固定回放批次上的序列小批量更新引发策略停滞（policy staleness），即当前策略随时间偏离行为策略。现有方法虽通过引入价值网络、步骤级标注或外部教师模型以获得更密集的信用信号，但均增加了训练复杂性；而重要性采样与截断虽能缓解策略漂移，却无法设定有限的目标以防止更新过程过度偏离。本文提出一种无价值网络、树感知的后验自蒸馏框架——TreeOPSD，其关键创新在于将每轮交互视为树状结构节点，从共享前缀分支出多种可能的延续路径，并递归地将最终奖励回溯至隐式回合级信用。随后，基于相对于冻结行为策略的似然比，将每项信用映射为闭式、有界的教师分布参数化形式。通过最小化该教师分布与学生策略之间的差异，既能恢复在行为策略处的策略梯度方向，又能通过有限目标约束序列小批量更新，有效防止过度更新。实验表明，TreeOPSD显著提升了多个基准任务上的智能体能力，尤其在长时程任务中表现突出，例如在Deep Research设置下，使Qwen3-30B-A3B-Thinking-2507的平均准确率相较GRPO提升4.97点。
链接: https://arxiv.org/abs/2610.23181
作者: Heming Zou,Qi Wang,Yixiu Mao,Yun Qu,Yuyi Guo,Zeyuan Liu,Chenglin Li,Weijie Liu,Saiyong Yang,Xiangyang Ji
机构: Chenglin Li(李成林); Weijie Liu(刘伟杰); Saiyong Yang(杨赛勇); Xiangyang Ji(季向阳)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Reinforcement learning with verifiable rewards has become a scalable post-training paradigm for large language models. However, training agents over long-horizon interactions presents two challenges: using only terminal rewards makes turn-level credit assignment sparse and high-variance, while sequential mini-batch updates on a fixed rollout batch introduce policy staleness as the current policy drifts from the behavior policy. Existing approaches obtain denser credit via learned critics, process reward models, or on-policy distillation, but they rely on a value network, step-level annotations, or a stronger external teacher that increase training complexity. Meanwhile, importance sampling and clipping mitigate policy drift, but they do not specify a finite target that stops sequential updates from overshooting. In this work, we introduce TreeOPSD, a critic-free, tree-aware posterior self-distillation framework for long-horizon agentic reinforcement learning. Treating each interaction turn as a tree node, TreeOPSD branches alternative continuations from shared prefixes and recursively backs up terminal rewards into implicit turn-level credit. It then maps each credit to a closed-form, bounded teacher distribution parameterized by the likelihood ratio relative to the frozen behavior policy. Minimizing the divergence between this teacher and the student policy recovers the policy-gradient direction as the tangent at the behavior policy while bounding updates toward a finite target across sequential mini-batches. Empirically, TreeOPSD improves agentic capabilities across several competitive benchmarks and model scales, especially for long-horizon tasks, e.g., raising the average accuracy of Qwen3-30B-A3B-Thinking-2507 by 4.97 points over GRPO on the Deep Research setting.
[NLP-53] Do Speech BCIs Need Larger Models? Rethinking Neural Decoding beyond Scaling NEURIPS2026
【速读】： 该论文旨在解决当前语音脑机接口（speech BCI）中大规模模型在数据稀缺、噪声大且标注成本高的低资源环境下，其性能提升效果不明确的问题。传统方法依赖于大型生成式模型或基于音素的复杂多阶段流水线，导致计算开销大、泛化能力受限。本文提出一种名为BEST（Brain-Encoding-Speech-Text）的两阶段高效神经信号到文本解码框架，其核心创新在于摒弃对大型生成模型和音素中心化流程的依赖，转而引入分层中间表示（hierarchical intermediate representations）与轻量级语音导向解码器，直接将神经活动映射为文本输出。该设计在保留关键语义结构的同时，显著降低参数量（仅需对比方法3.85%的参数）与计算成本，实现了在低资源条件下的鲁棒泛化与实际部署可行性。实验表明，在Brain-to-Text '24和’25基准上，BEST分别取得6.19%和1.62%的词错误率（WER），并揭示三大关键发现：（1）解码器规模扩展带来非单调性能增益；（2）跨模态对齐比模型规模更为关键；（3）过拟合会损害序列级性能。进一步的表征分析证实了神经信号、音频与文本之间存在一致的对齐路径。
链接: https://arxiv.org/abs/2610.23175
作者: Zehui Feng,Weichuan Wang,Xiaohan Chen,Cuntai Guan,Ting Han
机构: Shanghai Jiao Tong University (上海交通大学); Nanyang Technological University (南洋理工大学); Zhejiang University (浙江大学)
类目: Computation and Language (cs.CL)
备注: 25 pages, Accepted by NeurIPS 2026 as a Main Track conference paper
Abstract:Recent speech brain-computer interfaces (BCIs) increasingly rely on large models and complex multi-stage pipelines for neural decoding. However, in speech BCI, where training data are scarce, noisy, and expensive to collect, the benefits of model scaling remain unclear. This work revisits a fundamental question: are large models truly necessary for accurate neural decoding? We propose BEST (Brain-Encoding-Speech-Text), a two-stage framework for efficient neural-to-text decoding. Instead of relying on large generative models or phoneme-centric pipelines, BEST introduces hierarchical intermediate representations and a compact speech-oriented decoder that directly maps neural activity to text. This design preserves essential structure while significantly reducing computational cost, enabling robust generalization in low-resource settings and practical deployment under constrained resources. On the Brain-to-Text '24 and '25 benchmarks, BEST achieves word error rates (WERs) of 6.19% and 1.62%, respectively, using only 3.85% of the parameters required by comparable methods. Beyond these results, we provide three key insights: (1) decoder scaling yields non-monotonic gains; (2) cross-modal alignment is more critical than model size; and (3) overfitting degards sequence-level performance. Representation analysis further reveals a consistent neural-audio-text alignment pathway.
[NLP-54] Chinese Decision: A Family of Language Models for Single-Pass Probabilistic Decision-Making
【速读】： 该论文旨在解决语言模型在需要从有限选项中做出概率性决策的任务中，传统自回归生成方式效率低下且不适用于非开放回答场景的问题。其核心挑战在于如何在单次前向传播中高效、准确地输出候选选项的概率分布，同时支持布尔判断、分类选择与序数评分等多样化任务。解决方案的关键在于提出Chinese Decision系列中文模型，采用非自回归的单次前向计算架构，直接从上下文、指令和候选描述中提取最终位置的候选字母对数（logits），并在有效选项上进行归一化处理，从而实现快速、统一的决策输出。该方法通过共享接口兼容多种决策类型，并基于1304.8万条决策数据使用分布交叉熵训练0.8B模型，进一步引入强化学习决策优化（RLCD）目标训练2B模型。实验表明，未校准的0.8B模型在涵盖307,900个决策的CJ-Bench基准上达到72.24%通用领域准确率和76.98%专业领域平均准确率，整体期望校准误差仅为1.07%，验证了该方法在中文环境下高效率与高可靠性决策的可行性。
链接: https://arxiv.org/abs/2610.23170
作者: Qiang Yan(Institute of Computing Technology, Chinese Academy of Sciences)
机构: Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所)
类目: Computation and Language (cs.CL)
备注: 11 pages, 2 figures, 11 tables. Code: this https URL
Abstract:Many language-based applications require probabilities over a finite set of decisions rather than open-ended responses. We present Chinese Decision, a family of Chinese-language models for single-pass probabilistic decision-making. Given context, an instruction, and candidate descriptions, the model extracts final-position candidate-letter logits in one forward pass and normalizes them over valid options, without autoregressive answer generation. A shared interface supports Boolean judgments, categorical choices, and ordinal scores. We report a completed 0.8B model trained with distribution cross-entropy on 13.048 million decisions and preliminary 2B checkpoints trained with an additional RLCD objective. On the full CJ-Bench of 307,900 decisions, the uncalibrated 0.8B model achieves 72.24% general-domain accuracy and 76.98% average accuracy across medicine, law, and finance using a single checkpoint, with 1.07% overall expected calibration error. The 2B model reaches 72.56% general-domain accuracy after 40.13% of one epoch but has not surpassed the completed 0.8B model on the professional-domain average. We further evaluate a validation-temperature-scaled deployment model and its Q8 quantization, finding close aggregate performance while documenting individual prediction changes. We release the training, evaluation, and deployment pipeline. The current results establish a reproducible approach to Chinese probabilistic decisions but do not isolate the effect of model scale.
[NLP-55] Explaining Chess and Go in Natural Language
【速读】： 该论文旨在解决当前生成式AI在解释国际象棋走法时缺乏深度理解的问题，即现有语言模型虽能生成自然语言描述，但其解释往往依赖于表面的棋盘特征，难以捕捉可迁移、鲁棒的深层棋类概念。其解决方案的关键在于提出一种无需训练的新型方法——BinProp，该方法通过利用棋局评估值（chess-engine evaluations）作为监督信号，引导语言模型自主发现并利用可复用的棋类概念来生成解释。实验表明，BinProp显著提升了所有测试模型的解释质量，平均得分接近直接提示（direct prompting）的三倍，并且该方法在围棋（Go）、跳棋（draughts）和将棋（shogi）等其他策略性棋类游戏中亦展现出良好的泛化能力，无需额外调参即可发现有意义的概念。
链接: https://arxiv.org/abs/2610.23166
作者: Ori Press,Alexander Golubev
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Modern chess engines play far beyond human ability, often choosing moves whose underlying ideas are difficult for humans to understand. We propose an automatic metric for evaluating chess explanations, grounded in chess-engine evaluations. Evaluating state-of-the-art language models using this metric, we find that they perform surprisingly poorly, often relying on surface-level board observations rather than robust, transferable chess concepts. We then introduce BinProp, a training-free method that enables these same models to discover reusable chess concepts and use them to explain moves in natural language. BinProp improves explainability for every model tested, nearly tripling the average score compared with direct prompting. Beyond chess, the same method discovers useful concepts in Go, draughts, and shogi without additional tuning.
[NLP-56] Cross-Centaur: Modeling Individual Decision-Making Across Psychological Tasks with Language Models EMNLP2026
【速读】： 该论文旨在解决当前大型语言模型（LLM）在模拟人类行为时普遍存在的局限性：尽管已有模型如Centaur通过心理实验中的逐次决策微调提升了对群体层面规律的建模能力，但现有大规模行为数据集缺乏跨任务的个体追踪信息，导致模型难以捕捉个体在不同任务间的行为一致性与特异性。为此，研究提出构建首个跨任务个体级行为数据集Cross-Psych-101，涵盖591名参与者在12种不同类型任务（包括老虎机、权衡、结构与社会任务）中超过26万条有效决策。其核心解决方案是设计并训练Cross-Centaur模型，该模型能够基于目标任务描述以及同一参与者在源任务中的行为表现，预测其在目标任务中的具体行动。实验表明，Cross-Centaur在未见参与者及跨任务场景下均显著优于通用大模型和原始Centaur模型。探针分析进一步揭示，模型隐层状态编码了可跨任务迁移的个体行为倾向，验证了跨任务个体级数据对提升人类决策预测能力的关键作用。
链接: https://arxiv.org/abs/2610.23157
作者: Hikaru Asano,Hidezo Suganuma,Ryutaro Mori,Yukino Baba
机构: The University of Tokyo(东京大学); RIKEN Center for Brain Science(理化学研究所)
类目: Computation and Language (cs.CL)
备注: Accepted to EMNLP 2026 (Main Conference, Oral Presentation)
Abstract:Large language models (LLMs) are increasingly used as computational models of human behavior. Recent models such as Centaur have advanced this goal by fine-tuning LLMs on trial-by-trial decisions from psychological experiments. However, existing large-scale behavioral datasets do not link the same participants across different tasks, so these models learn population-level regularities but are not trained to model individual-level regularities across tasks. We introduce Cross-Psych-101, a dataset of 591 participants and 261,551 valid decisions across 12 Bandit, Tradeoff, Structure, and Social tasks. We train Cross-Centaur to predict a participant’s target-task actions from a target-task description and the source-task description paired with the same participant’s source-task behavior. Cross-Centaur outperforms off-the-shelf LLMs and Centaur on held-out participants, both within and across tasks. Probing analyses suggest that its hidden states encode participant-level behavioral tendencies that transfer across tasks. These results highlight the potential of cross-task individual-level data for predicting human decision-making with language models. Cross-Pysch-101 is available at this https URL.
[NLP-57] LED: Improving Masked Diffusion Language Models by Lookahead Editing
【速读】： 该论文旨在解决生成式AI中非自回归文本生成模型在优化难度与计算表达能力之间的权衡问题。现有基于扩散的大型语言模型（dLLMs）主要分为掩码扩散模型（MDMs）和统一扩散模型（UDMs），前者虽具有简单的掩码转移调度、易于优化，但表达能力受限；后者虽可通过迭代状态修正实现更高表达性，但训练难度显著增加。为兼顾二者优势，本文提出“前瞻编辑”（Lookahead Editing, \ours），其核心在于在掩码扩散解码过程中引入可重写的候选状态（proposal state），在每一步解码时通过多次前瞻更新迭代优化未来预测，从而在保持MDMs简单转移调度的同时，利用额外计算提升模型表达能力。实验表明，基于Qwen2.5-7B和Qwen3-8B微调得到的LED-7B与LED-8B，在8个基准任务上分别较标准MDM微调平均提升11.4和7.7分，且仅使用监督微调（SFT）的LED-8B已超越经过持续预训练、监督微调及强化学习训练的LLaDA2.1-16B Q-mode模型。
链接: https://arxiv.org/abs/2610.23122
作者: Yuyan Zhou,Zeyuan Yang,Yuxuan Song,Zhicheng Cai,Hongli Yu,Weiying Ma,Hao Zhou,James Kwok
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 10 pages, 5 figures
Abstract:Diffusion large language models (dLLMs) have emerged as a promising alternative to autoregressive language models, enabling non-autoregressive text generation through parallel decoding. Existing dLLMs can be broadly categorized into two families: masked diffusion models (MDMs) and uniform diffusion models (UDMs) In this work, we theoretically show that MDMs are easier to optimize due to their mask-based transition schedule but have lower computational expressivity, whereas UDMs enable iterative state revision and greater expressivity but are substantially harder to train. This trade-off raises a natural question: can we obtain the expressivity of iterative revision without giving up the simple, easy-to-optimize transition schedule of MDMs? Motivated by this analysis, we propose Lookahead Editing (\ours), which augments masked diffusion decoding with a rewritable proposal state. At each decoding step, the model performs several lookahead updates to iteratively refine future predictions in this proposal state before committing tokens to the masked sequence. This design introduces a rewritable proposal that uses additional compute to improve expressivity while keeping the simple mask-based transition schedule of MDMs. We fine-tune Qwen2.5-7B and Qwen3-8B to obtain LED-\texttt7B and LED-\texttt8B, respectively. Across eight benchmarks, LED SFT yields average gains of +11.4 and +7.7 points over standard MDM SFT. Moreover, using SFT alone, LED-\texttt8B outperforms LLaDA2.1-16B Q-mode, which is trained with continued pre-training, supervised fine-tuning, and reinforcement learning.
[NLP-58] Disentangling Language Effects on Tool Admission and Post-Retrieval Safety
【速读】： 该论文旨在解决大语言模型（LLM）代理在多语言环境下使用检索工具时存在的安全评估偏差问题，即当前的安全评测体系主要基于英语，未能充分考量语言差异对工具调用与生成内容安全性的影响。其核心问题是：语言因素是否影响工具准入（tool admission）或检索后生成内容的有害性（response harmfulness），或两者兼有。解决方案的关键在于提出“跨语言处理中的安全损失”（Safety Loss through Interlingual Processing, SLIP）现象，并通过在十种模型和七种语言中控制请求、检索页面及响应语言的实验设置，发现非英语有害请求显著提升平均工具调用率（从46.9%上升至67.4%），且在所有模型中均呈现正向变化；同时，要求以页面语言作答会加剧匹配工具调用下的综合有害性（从50.3%升至57.4%），尤其在中文和印地语中表现明显。研究进一步表明，仅翻译页面本身不会在多数模型中引发一致性的有害性上升，而自动化审计揭示了有害性标签无法有效区分真正的安全拒绝、因语言障碍导致的回避行为与能力失效。因此，论文主张多语言安全评估应将工具调用行为与响应有害性分开衡量，并明确区分安全拒绝与功能失败。
链接: https://arxiv.org/abs/2610.23093
作者: Aditya Nawal,Manit Baser,Mohan Gurusamy
机构: National University of Singapore(新加坡国立大学)
类目: Computation and Language (cs.CL)
备注: 33 pages, including appendix
Abstract:LLM agents increasingly use retrieval tools across languages, yet their safety evaluation remains English-centric. It remains unclear whether language affects tool admission, post-retrieval generation, or both. We independently vary request, retrieved-page, and required response languages in a controlled URL-retrieval setting across ten models and seven languages. We identify Safety Loss through Interlingual Processing (SLIP). Non-English harmful requests increase mean tool admission from 46.9% to 67.4%, with a positive mean change in every model. This pattern also persists in an autonomous retrieval setup across all four models tested in this setting. Requiring answers in the page language increases pooled harmfulness on matched tool calls from 50.3% to 57.4% across four closed-weight models, mainly in Chinese and Hindi. Translating pages alone produces no consistent harm increase in eight models with comparable inference settings. Automated audits show that a non-harmful response may be a genuine safety refusal, a deflection citing language difficulties, or an unusable answer. Harmfulness labels alone cannot distinguish these outcomes. Multilingual safety evaluations should therefore measure tool admission and response harmfulness separately and distinguish refusal from capability failure.
[NLP-59] Before Retrieval Begins: Safety and Tool-Use Decisions in LLM Agents ACL
【速读】： 该论文旨在解决大语言模型在调用外部工具时，其是否发起工具调用的决策机制对下游安全性具有重要影响，但这一关键决策过程长期被忽视的问题。现有研究多聚焦于工具调用后的输出安全，而忽略了调用前的判断逻辑。研究发现，在七个开源大模型中，面对明确有害请求时，未加权的平均伤害率高达74.5%。通过因子干预实验，研究识别出请求中“所请求操作”（requested-operation）成分的中性化处理导致最大幅度的调用接受率变化（19个百分点），表明该成分是影响调用决策的核心因素。进一步分析显示，基于该对比估计出的线性方向可有效预测其余四个模型在独立测试集上的调用行为，且部分模型在操作中性化请求下仍表现出调用倾向，说明该机制并非特定于某一操作类型。此外，研究发现书面拒绝（written refusal）与工具调用决策在行为和表征层面存在耦合关系：33.6%的有害调用伴随有文本拒绝。通过纯文本拒绝训练，可在保留良性工具使用能力的前提下降低不安全调用率，证明响应层的安全训练可向调用决策层迁移。综上，研究揭示了书面拒绝与工具调用决策之间存在部分、依赖模型的关系，强调应直接评估和监督工具调用决策，而非将文本拒绝作为其代理指标。
链接: https://arxiv.org/abs/2610.23071
作者: Aditya Nawal,Manit Baser,Mohan Gurusamy
机构: National University of Singapore(新加坡国立大学)
类目: Computation and Language (cs.CL)
备注: 19 pages, 5 figures. Preprint version of a manuscript submitted to Transactions of the Association for Computational Linguistics (TACL)
Abstract:Large language models increasingly act as agents that call external tools, yet most tool-safety research examines only what follows a tool call. The prior decision of whether to call the tool remains largely unexamined and can influence downstream safety. Across the seven open-weight models, the unweighted model-average harm rate after admitted harmful requests is 74.5%. To identify what governs this decision, we apply a factorial intervention to the input request. Within our generated construction, neutralizing the requested-operation component produces the largest admission contrast (19 pp). A linear direction estimated from this contrast influences held-out admission in four of seven models; effects on operation-neutralized requests in some models show that it is not an operation-specific mechanism. We then test whether written refusal is behaviorally and representationally coupled to the same decision. Written refusal accompanies 33.6% of admitted harmful calls. Plain-text refusal training reduces unsafe admission in models that retain benign tool use, showing that response-level safety training can transfer to the call decision. Together, these results establish a partial, model-dependent relation between written refusal and tool admission. They motivate direct evaluation and supervision of tool admission rather than treating text refusal as its proxy.
[NLP-60] ParetoChat: An LLM -Based Interface for Interactive Multi-Objective Optimization
【速读】： 该论文旨在解决多目标优化（Multi-objective Optimization）中决策者（Decision Makers, DMs）难以将定性偏好转化为优化算法所需的定量参数（如权重、优先级或参考点）的问题。传统方法要求用户精确指定数值型输入，而这一过程往往困难且不易解释。为应对该挑战，本文提出ParetoChat——一种基于大语言模型（Large Language Model, LLM）的专家代理交互接口，其核心解决方案是通过自然语言交互实现对用户定性偏好到结构化标量化参数的自动转换。该系统采用多代理架构，分离了输入处理、多准则决策（Multi-Criteria Decision-Making, MCDM）知识、解的呈现与反馈分析等模块，支持迭代式交互。仿真研究表明，相较于DESDEO平台中的参考点法和NIMBUS法，ParetoChat能更高效地引导搜索过程，逼近符合用户偏好的帕累托最优解；在历史快递订单分配实例中的应用进一步验证了其支持用户探索目标极值与中间权衡的能力。结果表明，基于LLM的代理可作为业务利益相关方与优化工具之间的实用交互层。
链接: https://arxiv.org/abs/2610.23062
作者: Gokhan Ceyhan
机构: Just Eat Takeaway.com
类目: Computation and Language (cs.CL); Optimization and Control (math.OC)
备注:
Abstract:Multi-objective optimization models are often deployed in settings where decision makers (DMs) must balance some conflicting criteria. In practice, however, stakeholders must translate their qualitative preferences between different criteria into numerical weights, priorities, or reference points required by optimization methods, which can be difficult to specify and interpret. This paper proposes \emphParetoChat, an LLM (large language model)-based expert-agent interface for interactive multi-objective optimization. The agent translates natural-language user input into structured scalarization parameters, invokes an optimization system, presents candidate solutions, and iteratively elicits feedback. We describe a multi-agent architecture that separates input processing, multi-criteria decision-making (MCDM) expertise, solution presentation, and feedback processing. We then present a simulation study using an LLM-based DM agent and benchmark ParetoChat against interactive methods implemented in DESDEO, including reference-point and NIMBUS methods. The computational experiments on multi-objective binary knapsack and set-partitioning instances show that ParetoChat can guide the search efficiently toward solutions that match the DM’s preferences. An application to historical courier-order assignment instances illustrates how users can explore objective extremes and intermediate trade-offs through natural-language requests. The results indicate that LLM agents may provide a practical interaction layer between business stakeholders and optimization tools.
[NLP-61] Same Person Different Advice: Auditing LLM Financial Advice with Repeated Risk-Tolerance Responses
【速读】： 该论文旨在解决生成式AI金融顾问在为客户制定股票配置建议时，过度依赖单一风险偏好问题的回答，而该回答具有显著的不稳定性问题。研究发现，同一用户在不同时间对同一风险问题的回答存在明显差异（如两年后41%的家庭答案改变，六个月内有三分之一的人在0-10量表上变动超过两个点），导致基于此回答生成的资产配置建议产生大幅波动。其解决方案的关键在于：当前主流大语言模型（LLM）对风险回答的响应强度远超实际认知稳定性，即使未提示可靠性，模型也表现出与“高可靠性”假设下一致的强烈反应；若依据经典测验理论，对风险回答的信度进行校正，可使配置建议的波动幅度降低2.4至3.3倍。此外，引入额外财务问题或一个澄清性提问可显著减少建议波动，其中一次澄清性提问能将建议变化从平均33.5个百分点降至1.9个百分点，同时保持对确认后变更的敏感性。研究强调，应将同人重测审计作为AI金融顾问的标准验证机制，并指出如何处理不可靠回答不仅是统计问题，更涉及客户自主权的核心伦理议题。
链接: https://arxiv.org/abs/2610.23060
作者: Soovadeep Bakshi,Apurba Bose,Aishwarya Bose,Sandip Nandi
机构: Independent Researcher(独立研究员)
类目: Computation and Language (cs.CL); Computers and Society (cs.CY)
备注: 23 pages, 5 figures, 9 tables
Abstract:AI financial advisers set a client’s stock allocation largely from one stated-risk question, yet people do not answer it the same way twice: in a national US panel, 41% of households chose a different answer two years later, and in a laboratory retest a third of people moved two or more points on a 0-10 scale after six months. Seven large language models from five providers gave one-turn allocation advice for 800 real household profiles in which only the risk answer changed. The recommended stock share moved by 10 to 16 percentage points on average, and by 24 to 35 when the answer changed, against 0.6 to 4.2 points for the same prompt asked twice; in the laboratory retest the rating alone moved advice by 11 to 18 points. Without any note, the models respond to the answer about as strongly as when told that it is very reliable (less so for the two Claude models). If that instructed response is how each model maps attitude to advice, discounting the answer for its measured reliability, as classical test theory prescribes, would cut the swings by a factor of 2.4 to 3.3. The swing is largest for clients under 40. A note that the answer is noisy steadies some models, but shifts their weight to the client’s age and pulls cautious clients towards a generic middle allocation: clients unwilling to take any risk are advised 22% stocks instead of 4%. Three extra money questions reduce the swing for some models, with no statistically detectable change relative to how much advice differs across people. In a scripted test, one clarifying question removes nearly all of the swing from an answer the client then corrects (33.5 points to 1.9 on average) while keeping the response to a change the client confirms. We argue that same-person retest audits should be a standard check for AI advisers, and that what to do with an unreliable answer is a question of client autonomy as well as of statistics.
[NLP-62] ECHO: Replacing the Compliance Proxy with Response Harmfulness in Optimization-Based Jailbreak Attack
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在安全对齐后仍易受越狱攻击（jailbreak attacks）的问题。现有基于优化的攻击方法普遍采用单一目标，即最大化提示词（prompt）层面的代理指标——如肯定前缀的似然度，作为模型合规性的衡量标准。然而，此类代理指标无法区分模型回答中的谨慎回避与完全可执行的有害内容，导致攻击搜索过程缺乏对模型实际输出的精准感知。本文提出ECHO（Evaluation-Conditioned Harmfulness Optimization），其核心创新在于以响应层面、连续的有害性评分替代原有代理指标，该评分由HarmSig（一个17亿参数的回归器）提供，其训练基础为模型真实的有害能力表现。ECHO将HarmSig与五种词级扰动算子及关键词锚定的束搜索（Keyword-Anchored Beam Search）结合，并引入全局顶-m记忆机制，在严格的黑盒威胁模型下实现高效攻击。在AdvBench上的五款开源权重模型测试中，ECHO实现了最高0.898的StrongREJECT平均得分和89.23%的HarmBench成功率，显著优于现有方法。
链接: https://arxiv.org/abs/2610.23030
作者: Yuanwei Wu,Lingdong Huang,Yunlong Mao,Kanye Ye Wang,Jingyu Hua,Sheng Zhong
机构: Nanjing University (南京大学); University of the Chinese Academy of Sciences (中国科学院大学); University of Macau (澳门大学)
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL)
备注:
Abstract:Despite extensive safety alignment, large language models remain vulnerable to jailbreak attacks. Existing optimization-based attacks share a single objective: maximize a token-level proxy of victim compliance—such as the likelihood of an affirmative prefix—computed on the prompt rather than the decoded reply. This proxy cannot distinguish a cautious hedge from a fully actionable harmful answer, leaving the search blind to what the model actually says. We propose ECHO (Evaluation-Conditioned Harmfulness Optimization), which replaces the compliance proxy with response-level, continuous harmfulness scores from HarmSig, a 1.7B regressor grounded in actual harmful capability. ECHO couples HarmSig with five word-level perturbation operators and Keyword-Anchored Beam Search with global top-m memory, under a strict black-box threat model. Across five open-weight victims on AdvBench, ECHO achieves up to 0.898 StrongREJECT mean score and 89.23% HarmBench success. Our code is publicly available at this https URL.
[NLP-63] Controlling the Diversity-Quality Trade-Off of LLM s by Merging Models Before and After Alignment
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在输出多样性与质量之间难以平衡的问题。已有的方法或依赖昂贵的微调，或仅通过温度采样等手段提升词汇层面的多样性，而对语义多样性的改善有限。本文提出一种深度感知的FrankenMerging方法，无需额外训练即可调控对齐模型与其基础模型之间的多样性—质量权衡。其核心思想是：假设词汇多样性主要由模型低层表征决定，而语义多样性则更多由高层表征决定，因此将对齐模型的低层与基础模型的高层进行融合。实验结果表明，通过调整融合分界层的位置，可灵活调节输出的多样性与质量；多数情况下，融合模型在语义多样性上优于对齐模型，在某些分界点甚至超越两个原始模型的词汇多样性。相较于温度采样，该方法在大多数场景下展现出更优的多样性—质量权衡表现。反向融合的消融实验进一步验证了该假设的合理性。
链接: https://arxiv.org/abs/2610.22999
作者: Yasuaki Sumita,Tomoharu Iwata,Toshiyuki Tanaka
机构: Kyoto University (京都大学); NTT Communication Science Laboratories (NTT通信科学实验室); NTT, Inc. (NTT公司)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 13 pages, 5 figures
Abstract:Alignment is known to change the output characteristics of large language models (LLMs). Base models tend to generate more diverse outputs, whereas aligned models tend to produce higher-quality outputs. The appropriate balance between the two depends on the task. Existing methods either require costly fine-tuning or, like temperature sampling, mainly improve lexical diversity rather than semantic diversity. We introduce a depth-aware FrankenMerging approach for controlling the diversity–quality trade-off between an aligned model and its base model without additional training. This approach combines the lower layers of an aligned model with the higher layers of its base model based on the hypothesis that lexical diversity is more closely associated with lower layers and semantic diversity with higher layers. Experiments showed that varying the number of layers taken from each model adjusts both diversity and quality. In most cases, the merged models exhibited higher semantic diversity than the aligned model, and at some split positions, they showed higher lexical diversity than both original models. They also demonstrated a better diversity–quality trade-off than temperature sampling in most settings. A supporting ablation, in which we reversed the merging order, yielded results largely consistent with the hypothesis.
[NLP-64] Prism Attention: Proposal-Refined Index Sharing Mechanism for Efficient LLM s Inference NEURIPS2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在长上下文生成过程中，由于自回归解码时需持续关注不断增长的键值（Key-Value, KV）缓存所导致的内存与计算瓶颈问题。现有基于令牌级稀疏注意力（token-level sparse attention）的方法虽通过限制计算范围以降低开销，但其实际性能提升常被随上下文扩展而反复重新计算各层稀疏索引所抵消。尽管通过在指定锚点层计算并跨层共享稀疏索引可缓解此问题，但静态复用策略忽略了注意力模式在不同层间的动态演化特性，导致复用索引逐渐偏离各层最优，造成显著的注意力损失。针对这一挑战，本文从注意力动态角度重新审视跨层稀疏性，发现层间显著令牌的转移具有高度局部化特征：尽管各层中前κ个最高注意力权重的精确索引存在差异，但核心高注意力令牌保持稳定，其变化主要集中在先前选中区域的扩展边界附近。基于此洞察，论文提出Prism Attention——一种基于“粗粒度到细粒度”渐进优化的稀疏注意力框架，将传统的刚性跨层索引复用替换为动态适应的层级协同机制，在保留跨层共享效率的同时维持各层的自适应能力。实验结果表明，Prism Attention在主流推理与长上下文基准测试中均达到当前最优性能，并在128K上下文长度下相比FlashAttention实现3.4倍的端到端解码速度提升。
链接: https://arxiv.org/abs/2610.22985
作者: Zhenxu Tian,Kebin Liu,Zhengwu Yang,Yi Su,Qingqing Dang,Yanlin Sha,Kaipeng Deng,Dianhai Yu,Yanjun Ma,Juntao Li,Min Zhang
机构: Soochow University (苏州大学); Baidu Inc. (百度公司)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Published as a main conference paper at NeurIPS 2026
Abstract:Long-context decoding in large language models (LLMs) is increasingly bottlenecked by the memory and computation required to attend over an ever-growing Key-Value (KV) cache at each autoregressive step. Although token-level sparse attention reduces this burden by restricting computation to a small subset of relevant past tokens, its practical gains are often offset by repeatedly recomputing layer-wise sparse indices as the context grows. A common remedy is to share sparse indices across layers by computing them only at designated anchor layers, but such static reuse ignores the layer-wise evolution of attention patterns, causing the reused indices to progressively deviate from the layer-specific optima and incur substantial attention loss. By revisiting cross-layer sparsity from the perspective of attention dynamics, we find that inter-layer salient token shift is highly localized: although the exact top- \kappa indices differ across layers, the core high-attention tokens remain stable, with changes concentrated near the expanded boundary of previously selected regions. Motivated by this, we propose Prism Attention, a proposal-refined sparse attention framework. It replaces rigid cross-layer index reuse with a coarse-to-fine strategy, preserving layer-wise adaptability while retaining the efficiency benefits of cross-layer sharing. Extensive experiments show that Prism Attention achieves state-of-the-art performance among mainstream sparse attention methods on both reasoning and long-context benchmarks, while attaining a 3.4\times end-to-end decoding speedup over FlashAttention at 128K context length.
[NLP-65] Self-Evolving Agents Can Repair Failures but Struggle to Generalize from Them
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）智能体在缺乏成功轨迹或外部评估信号的情况下，能否仅通过失败交互实现持续自我进化的问题。其核心挑战在于：当智能体处于冷启动状态且仅能依赖失败经验进行学习时，是否仍具备构建可泛化能力的潜力。论文提出“仅从失败中演化”（Cold-Start Failure-Only Evolution, CFOE）范式，系统探究智能体在无正向引导下的自进化能力。研究发现，尽管失败经验有助于修复已知失败任务，但难以催生对未见任务的持续改进，暴露出显著的“修复-泛化差距”（repair-generalization gap）。深入分析表明，智能体常错误解读失败原因，且即使经过验证的正确经验也难以有效迁移。进一步实验显示，引入验证器反馈或额外交互仅带来有限提升，而成功轨迹则显著促进知识迁移。因此，当前自演化智能体更擅长基于已有失败进行局部修复，而非从失败中构建具备泛化性的通用能力。
链接: https://arxiv.org/abs/2610.22984
作者: Weixiang Zhao,Yichen Zhang,Yanyan Zhao,Jinming Bai,Bin Liang,Ruifeng Xu,Bing Qin,Ting Liu
机构: Harbin Institute of Technology, Harbin, China; Harbin Institute of Technology, Shenzhen, China; The Chinese University of Hong Kong
类目: Computation and Language (cs.CL)
备注: 28 pages, 9 figures, 6 tables. Weixiang Zhao and Yichen Zhang contributed equally
Abstract:Experience-driven self-evolution offers a promising path for large language model (LLM) agents to continually improve through interaction. Yet existing approaches often rely on successful trajectories or external evaluation signals, leaving unclear whether agents can evolve when such positive anchors are unavailable. We study this question through Cold-Start Failure-Only Evolution (CFOE), where agents start without accumulated experience and learn only from failed interactions without privileged guidance. Across three self-evolving paradigms, multiple LLM backbones, and four interactive environments, we uncover a consistent repair-generalization gap: failure-derived experience can help repair previously failed tasks, but rarely yields sustained improvements on unseen tasks. Further analysis shows that agents often misinterpret failures, while even verified-correct experience remains difficult to transfer. Relaxing CFOE reveals that verifier feedback and additional interaction provide limited gains, whereas successful trajectories substantially improve transfer. These findings suggest that current self-evolving agents are better at repairing observed failures than building generalizable capabilities from failure.
[NLP-66] What Shapes Brain Alignment of Large Language Models during Dialogue? Insights from Language Production and Comprehension
【速读】： 该论文旨在解决大语言模型（LLM）在对话任务中与人类大脑神经表征之间对齐关系的研究空白，尤其关注语言生成与理解双重过程中的脑-模型对齐机制。现有研究多聚焦于被动语言理解，而对话场景下涉及的语言生成与理解动态交互尚未充分探索。为此，研究利用对话过程中采集的fMRI数据，系统分析了不同模型层级与脑区在语言生成和理解两种模式下的脑对齐模式，并进一步考察其与模型表征内在维度（intrinsic dimension, ID）的关联性，以及若干可解释的语言学与交际特征（如句法复杂度、语义连贯性、对话意图等）的贡献差异。研究发现，语言生成与理解在脑-模型对齐模式上呈现显著差异，且这种对齐程度与特定脑区的内在维度高度相关；同时，语言与交际特征的作用随处理模式、模型层级及脑区的不同而动态变化，揭示了生成与理解在认知加工路径上的异质性。关键突破在于通过多模态数据融合与表征分析，首次系统刻画了对话情境下脑-模型对齐的分层、功能特异性规律，为理解生成式人工智能与人类语言认知的共性提供了新范式。
链接: https://arxiv.org/abs/2610.22961
作者: Yuko Nakagi,Yuya Chiba,Yasue Kishino
机构: NTT, Inc.(日本电信电话公司)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG); Neurons and Cognition (q-bio.NC)
备注: 33 pages
Abstract:Brain alignment of large language models (LLMs) offers a way to investigate representational similarities between LLMs and the human brain during language processing. While previous studies have mainly focused on passive language comprehension, little is known about such alignment during dialogue, in which speakers engage in both language production and comprehension. In this study, using fMRI data collected during dialogue, we examine brain alignment during language production and comprehension across LLM layers and brain regions. We assess its association with the intrinsic dimension (ID) of LLM representations and the contributions of several interpretable linguistic and communicative features. Our results reveal distinct patterns of brain alignment between production and comprehension, as well as strong associations with ID in brain regions associated with each processing mode. Furthermore, we show that the contributions of linguistic and communicative features also vary across processing modes, LLM layers, and brain regions.
[NLP-67] SAFE-Mem: Secure Agent ic Memory with Fine-Grained Authority and Governance-Aware Retrieval
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）代理在使用持久化记忆时面临的安全风险问题，即外部来源中未受信任的声明（atomic statements）可能被误认为是授权偏好、政策或权限，从而导致未经授权的记忆使用。现有防御方法多依赖基于内容的过滤或粗粒度的溯源追踪，难以在防止非法访问的同时不损害合法信息的可用性。其核心解决方案为SAFE-Mem——一种细粒度的记忆治理框架，关键在于通过为每个声明添加三种治理标签（类型、权限、范围）以精确刻画其出处与使用许可；结合感知治理的检索机制，实现查询需求与声明权限的精准匹配，有效阻断未经授权的影响并保留合法信息；同时构建支持溯源追踪与高效检索的治理增强型记忆图（governance-enabled memory graph）。实验结果表明，在MPBench和LoCoMo基准上，SAFE-Mem在保持合法记忆可用性的前提下，攻击成功率较最强对比方法降低至少77%，且问答准确率最高，相较其他方案提升达15%。
链接: https://arxiv.org/abs/2610.22959
作者: Haoming Li,Dingyi Kang,Xu Hu,Jinxin Yang,Zhu Xu,Chenru Wang,Yi Ge,Miao Yin,Guanpeng Li,Feng Chen,Bingzhe Li
机构: 未知
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL)
备注:
Abstract:Large language model (LLM) agents increasingly rely on persistent memory to support long-term reasoning and personalized interactions. However, information collected from external sources can introduce security risks when untrusted claims (i.e., atomic statements) are later treated as authorized preferences, policies, or permissions. Existing defenses typically rely on content-based filtering or coarse-grained provenance tracking, which either fail to prevent unauthorized memory use or block legitimate information, sacrificing memory utility. To address these limitations, we propose SAFE-Mem, a fine-grained memory governance framework that improves security while preserving legitimate memory use. First, SAFE-Mem annotates each claim with three governance labels (i.e., kind, authority, and scope) to capture its provenance and permitted use. Second, governance-aware retrieval matches query requirements with claim permissions to block unauthorized influence while retaining legitimate information. Third, a governance-enabled memory graph connects source sentences, claims, and entities to support provenance tracking and efficient retrieval. Across four LLM backbones on MPBench and LoCoMo, SAFE-Mem attains the lowest attack success among defenses that retain legitimate memory, at least 77% below the strongest of them, and the highest answer accuracy of all evaluated defenses, improving it by up to 15%. The code is available at this https URL.
[NLP-68] Recognizing Atari and Assessing Sente: Controlled Evaluation Tasks for Language-Based Go Agents
【速读】： 该论文旨在解决生成式 AI 在围棋领域中对专业术语（domain terminology）的识别能力及其与着法评估之间关联性的问题，具体聚焦于“打吃”（atari）这一核心概念在局部棋盘变化下的识别准确性以及其战略意义评估。研究发现，尽管经过术语和解说监督微调的围棋语言模型（LoGos）在部分情况下能生成符合语义的术语预测，但其对“打吃”关系的变化缺乏敏感性，且其术语判断与实际战略价值评估存在脱节——即使模型的术语输出与参考评估结果矛盾，仍会伴随特定着法偏好。该研究的关键在于构建了一个受控的评估框架，通过分析术语识别与着法评价在不同战略情境下的匹配程度，揭示了当前语言模型在将领域术语与精确棋盘关系及战略价值建立可靠映射方面的局限性，为未来语言驱动型围棋智能体的可解释性与准确性改进提供了关键基准。
链接: https://arxiv.org/abs/2610.22953
作者: Taiyi Chen,Iskar Deng,Amandalynne Paullada
机构: University of Washington(华盛顿大学)
类目: Computation and Language (cs.CL)
备注: 7 pages, 2 figures, 5 tables. The first two authors contributed equally
Abstract:Domain terminology provides concise descriptions of game states and strategic relationships, but whether game-playing language models can accurately recognize these concepts and relate them to move evaluation remains unclear. We construct a controlled evaluation set centered on atari and its relationship to sente in Go, examining terminology recognition under local board changes and move evaluation under different strategic conditions. We conduct our tests on LoGos, a Go language model, after fine-tuning it with terminology and commentary supervision. The model struggles to recognize changes in the atari relationship and assess its strategic implications. Its terminology predictions are associated with move preferences even when those preferences contradict the reference evaluation. Our controlled tests provide a framework for examining whether language-based Go agents connect domain terminology with precise board relationships and strategic move values.
[NLP-69] Aligning Language Models Towards Counterfactual Invariance
【速读】： 该论文旨在解决大语言模型在生成文本时对输入提示中敏感属性（如人口统计学特征）的过度敏感问题，即当这些属性与任务本身无因果关联时，模型仍可能产生受其影响的偏见性输出，从而导致反事实偏差（counterfactual bias）。尽管在某些情境下，敏感属性信息对生成语义连贯的回答具有合理性，但若其被不当利用，则会削弱模型的公平性。解决方案的关键在于提出一种基于群体层面奖励函数的策略梯度方法，该方法通过对比反事实提示对（counterfactual prompt pairs）进行训练，强制模型在不同群体间保持输出的一致性，同时保留其在敏感信息相关时的合理响应能力。研究进一步构建了受控的反事实提示对数据集，并设计了可分离评估敏感信息因果相关性的基准测试，最终在多个公平性基准上验证了所提方法能有效降低反事实差异，同时维持整体生成质量。
链接: https://arxiv.org/abs/2610.22942
作者: Basar Demir,Emily Shen,Ksheeraj Sai Vepuri,Sourabh Kulkarni,Harihar Subramanyam,Ankit Jain,Nan Jiang,Chirag Nagpal
机构: Meta Superintelligence Labs(Meta 超智能实验室); University of California, San Diego(加州大学圣地亚哥分校); Independent Researcher(独立研究员)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Small variations in prompt characteristics can meaningfully alter the semantics, tone, and contextual assumptions of large language model outputs. Such characteristics can involve group variables such as demographics or other sensitive latent attributes. Sensitivity to such input variables is undesirable when demographic information is causally irrelevant to the task, but might still be relevant for generation of semantically coherent responses. This makes the task of enforcing counterfactual invariance while preserving model utility difficult. To study this problem, we first show that standard model alignment recipes (such as reinforcement learning) increase differences in response quality across demographics, while improving overall response quality, ultimately amplifying what is known as counterfactual bias. To address this, we develop a policy gradient technique involving a group-level reward function that operates on matched pairs of counterfactual prompts. For completeness, we build a dataset of controlled counterfactual prompt pairs and design an evaluation that separately measures cases where demographic information is causally relevant to the answer. Finally, we validate our findings on several fairness benchmarks. We demonstrate that our methods reduce counterfactual differences while preserving overall response quality and that the models continue to adapt appropriately when such sensitive information is relevant.
[NLP-70] Beyond Surface Cues: Robust LLM -Generated Korean Text Detection
【速读】： 该论文旨在解决当前大语言模型（LLM）生成内容检测研究中存在的重要局限性，即多数方法以英语为中心，缺乏对韩语（Korean）等非英语语言的系统性研究，且现有韩语文本检测评估常依赖表面线索（superficial cues），导致性能评估结果过于乐观。为应对这些问题，论文提出一个名为KoLing的对抗性基准数据集，其具有偏差可控、规模达50.9万样本的特点，能够有效抑制表面捷径（shortcuts），同时保持语义一致性，并分别评估纯生成文本与经LLM修改文本的检测能力。在此基础上，作者进一步提出KoLingProbe，一种基于理论语言学的可解释检测器，通过建模韩语在多个语言层次上的特异性属性实现高精度检测。实验表明，KoLingProbe在各项检测任务中显著优于主流基线模型，且对对抗攻击具备强鲁棒性；分析还揭示，跨多层次互补语言特征的融合不仅能提升检测性能，还能增强模型决策过程的可解释性。
链接: https://arxiv.org/abs/2610.22940
作者: SungHo Kim,SangKeun Lee
机构: Korea University(高丽大学); Department of Artificial Intelligence(人工智能系); Department of Computer Science and Engineering(计算机科学与工程系)
类目: Computation and Language (cs.CL)
备注:
Abstract:As large language models (LLMs) generate increasingly fluent text, the need for reliable detection of LLM-generated content has grown substantially. However, existing research remains heavily English-centric, leaving Korean largely underexplored. Moreover, current evaluations of Korean text detection often rely on superficial cues, leading to overly optimistic estimates of detection performance. To address these limitations, we introduce \textscKoLing, a bias-controlled adversarial benchmark comprising 50.9k instances that suppresses superficial shortcuts while preserving semantic content and separately evaluates LLM-generated and LLM-modified text detection. We further propose KoLingProbe, an interpretable detector grounded in theoretical linguistics that models Korean-specific properties across multiple linguistic levels. Experiments show that KoLingProbe substantially outperforms strong baselines on all detection tasks while remaining robust to adversarial attacks. Our analyses further reveal that integrating complementary linguistic features across multiple levels yields both improved performance and interpretability.
[NLP-71] Many Biases One Circuit: Post-Training Tunes a Pretrained Cue-Routing Mechanism
【速读】： 该论文旨在解决语言模型在面对提示（prompt）中包含误导性线索（cue）时，倾向于采纳错误选项而非正确答案的问题，此类现象被称为“谄媚倾向”（sycophancy）及相关偏差。传统观点认为此类偏差源于对齐（alignment）过程，但本文通过因果分析挑战这一观点，揭示出这些偏差的根源在于模型内部一个共享的注意力机制回路。研究发现，在多数模型的多项选择测试中，四种不同类型的线索均依赖于少数几个相同的注意力头，且该回路在预训练阶段即已存在。基础模型在纯文本提示下即会响应线索，而通过约200个无线索的通用助手示例进行微调后，该回路在对话格式中被激活，表明线索跟随行为可能是助手微调的副产物。最终的强化学习（RL）阶段进一步放大了线索的影响。在Tulu-3-8B和OLMo-2-7B模型中，通过对两个连续检查点的六个中间层注意力模式进行交换，发现对于建议答案类线索，模型的线索响应程度几乎完全转移到另一检查点的水平。这表明，后训练流程主要改变的是模型使用该共享回路的时机与强度。因此，缓解此类多选偏差的关键在于识别并干预这一核心注意力回路。
链接: https://arxiv.org/abs/2610.22937
作者: Prakhar Gupta,Terry Jingchen Zhang,Zhijing Jin
机构: University of Michigan (密歇根大学); Jinesis Lab, University of Toronto (多伦多大学) Vector Institute; University of Oxford (牛津大学); EuroSafeAI; Max Planck Institute for Intelligent Systems, Tübingen, Germany (马克斯·普朗克智能系统研究所，图宾根，德国)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Language models often abandon a correct answer when the prompt contains a cue, a hint toward a wrong option such as “I think the answer is B.” Sycophancy and related biases show this failure. They are often studied separately and blamed on alignment. We test this view causally, tracing the attention heads behind these biases through the stages of two post-training pipelines. In multiple-choice tests on most models, four cue types rely largely on the same few heads, a shared circuit already present in pretrained models. Most base models follow cues in plain text. Fine-tuning a base model on about 200 generic assistant examples without cues activates these heads in chat format, so cue-following can be a side effect of assistant tuning. The final reinforcement learning (RL) stage strengthens the effect of cues in almost every comparison. In Tulu-3-8B and OLMo-2-7B, we swap the attention patterns of six middle layers between the RL checkpoint and its predecessor. For suggested-answer cues, this shifts each checkpoint’s cue effect most of the way to the other’s level. Post-training thus seems to change when and how strongly models use this circuit. Mitigation for these multiple-choice biases could target this circuit.
[NLP-72] ASR-Style Transcript Normalization Removes Useful Prag matic Signal for Dialogue Act Classification
【速读】： 该论文旨在解决自动语音识别（ASR）流程中文本归一化对对话行为分类（dialogue act classification）性能的负面影响问题。现有ASR系统通常通过去除填充词、标点符号及扩展口语表达来标准化转录文本，但这一过程可能无意中丢弃对对话行为识别至关重要的语用线索。研究的关键在于系统评估了九种不同的文本处理条件在六种分类器（包括TF-IDF逻辑回归、RoBERTa、BERT、ALBERT、GPT-2和BiLSTM）与三个对话语料库（Switchboard、ICSI MRDA和SLUE-HVB）组合下的表现，结合多种子实验、归一化对照组和真实ASR输出评估。结果表明，在所有15个神经模型-语料库组合中，过度归一化均导致宏平均F1分数下降；尤其是使用原始填充词与回声词词表进行移除时，对MRDA和SLUE-HVB语料库性能造成持续恶化，而对Switchboard则效果不一。进一步分析显示，标点符号的移除使Switchboard的宏平均F1下降0.036，MRDA下降0.078，这表明原始线性基线模型看似对归一化不敏感，实则源于分词机制的掩盖效应。当仅保留“um”、“uh”、“erm”等常见填充词而不移除其他关键回声表达时，对MRDA仍造成性能下降，但在SLUE-HVB上显著缓解，说明原词表移除了具有类别判别意义的回声表达。因此，研究揭示了归一化的影响高度依赖于预处理定义、特征提取方式及对话上下文，强调应根据具体任务需求而非盲目执行统一的文本归一化策略。
链接: https://arxiv.org/abs/2610.22934
作者: Lucky Onyekwelu-Udoka,Md Shafiqul Islam
机构: Iowa State University (爱荷华州立大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Automatic speech recognition (ASR) pipelines often normalize transcripts by removing filled pauses and punctuation and expanding colloquial forms, potentially discarding cues useful for dialogue act classification. We evaluate nine transcript conditions across six classifiers (TF-IDF logistic regression, RoBERTa, BERT, ALBERT, GPT-2, and BiLSTM) and three conversational corpora (Switchboard, ICSI MRDA, and SLUE-HVB), supplemented by multi-seed experiments, normalization controls, and real-ASR evaluations. Across five random seeds, heavy normalization reduces Macro-F1 in all 15 neural model-corpus combinations under the original preprocessing protocols. Removal using the original filler and backchannel lexicon consistently degrades performance on MRDA and SLUE-HVB but produces mixed effects on Switchboard. Punctuation-aware TF-IDF experiments show that punctuation removal reduces Macro-F1 by 0.036 on Switchboard and 0.078 on MRDA, demonstrating that the original linear baseline’s apparent insensitivity resulted from tokenization. Restricting filler removal to um, uh, and erm preserves degradation on MRDA but largely eliminates the SLUE-HVB effect, where the original lexicon removes class-defining backchannel expressions. These findings demonstrate that normalization effects depend on preprocessing definitions, feature extraction, and conversational context, supporting task-dependent rather than indiscriminate transcript normalization.
[NLP-73] Sep-Latent: Separator Tokens as Natural Compression Anchors for Implicit Reasoning
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在复杂推理任务中采用显式思维链（Chain-of-Thought, CoT）所面临的计算开销大、生成文本冗长的问题，同时克服现有隐式推理方法在连续潜在空间中因固定长度潜在序列导致的表征坍缩缺陷。其解决方案的关键在于提出一种自适应的隐式推理框架——Sep-Latent，该框架通过引入分隔符标记（separator tokens）作为结构化的语义锚点，结合注意力衰减（Attention-Decay）机制，在训练过程中强制模型将推理逻辑压缩并聚焦于这些锚点上。此外，通过跨所有层的多层级知识蒸馏，实现显式与隐式推理路径之间的逻辑一致性对齐。尤为关键的是，Sep-Latent支持动态推理深度，可根据查询复杂度自适应终止潜在思考阶段，从而显著提升推理效率。实验结果表明，在GSM8k-Aug基准上，该方法仅需平均5.4个词元即可达到56.8%的准确率，相较于以往固定长度的隐式模型在效率上表现更优，且性能差距进一步缩小至显式CoT方法。主成分分析（PCA）与符号变体测试进一步验证了该方法具备良好的语义根基性及对扰动的更强鲁棒性。
链接: https://arxiv.org/abs/2610.22924
作者: Yinlong Xu,Jianfeng Huang,Zhengdong Zhou,Mingjie Lin,Fuyu Lv,Dan Ou,Hongxia Xu,Jian Wu
机构: Zhejiang University (浙江大学); Alibaba Group (阿里巴巴集团)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Large Language Models (LLMs) have demonstrated remarkable success in complex reasoning through explicit Chain-of-Thought (CoT), but the associated computational overhead and token verbosity remain significant barriers to efficiency. While implicit reasoning in continuous latent spaces offers a promising alternative, existing methods like Coconut and CODI often rely on fixed-length latent sequences and suffer from representational collapse. In this paper, we propose Sep-Latent, an adaptive implicit reasoning framework that utilizes separator tokens as structural semantic anchors. By introducing an Attention-Decay mechanism, we force the model to aggregate and compress reasoning logic into these anchors during training. We then align explicit and implicit pathways through multi-level distillation across all layers to ensure logical consistency. Crucially, Sep-Latent enables dynamic reasoning depth by adaptively terminating the latent thinking phase based on query complexity. Experimental results on the GSM8k-Aug benchmark show that Sep-Latent achieves an accuracy of 56.8% with an average of only 5.4 tokens, significantly outperforming prior fixed-length implicit models in efficiency while narrowing the performance gap with explicit CoT. Further analysis via PCA and symbolic variant testing confirms that our method ensures semantic grounding and exhibits superior robustness to perturbations.
[NLP-74] Wheres the Indirect Evidence? Investigating Grammatical Generalization with Influence Functions
【速读】： 该论文旨在解决语言模型在未显式学习特定语法现象的情况下，仍能表现出对该现象的泛化能力这一问题，其核心在于识别并验证促使这种泛化发生的“间接证据”（indirect evidence）。现有研究虽已推断间接证据的存在，但缺乏对其具体性质的明确界定与实证分析。本文提出一种自下而上的方法，利用影响函数（influence functions）量化训练语句对模型在15个来自BLiMP数据集的语法现象上的预测影响，通过人工构建的规则筛选并剔除直接证据后，保留最具影响力的60,000条剩余句子作为间接证据。实验表明，移除这些间接证据导致模型在五种不同架构下的保留准确率下降幅度显著超过移除同等规模的随机样本组（中位下降12个百分点），证明其对学习过程具有实质性影响。进一步分析显示，该间接证据在结构上与直接证据过滤器的条件存在部分匹配，其匹配频率为随机对照组的4.7倍。此外，基于间接证据重叠进行聚类可恢复出BLiMP中包含多个现象的三类语言学类别，表明该方法不仅可有效识别间接证据，还能揭示潜在的语言学关联。因此，解决方案的关键在于通过影响函数定位高影响力训练样本，并结合结构相似性分析与聚类验证，系统地识别和表征语言模型中隐含的间接学习信号。
链接: https://arxiv.org/abs/2610.22919
作者: William Dai,Zhenghao Herbert Zhou,Maya Viswanathan,Robert Frank
机构: Yale University (耶鲁大学); Wu Tsai Institute (吴氏研究所)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 18 pages
Abstract:Language models generalize beyond their training corpus to learn linguistic phenomena (Patil et al., 2024; Misra and Mahowald, 2024), showing that there exists indirect evidence driving learning. Prior work has inferred that indirect evidence exists without identifying its character or considered specific hypotheses about what it might be. We explore an alternative bottom-up method for identifying indirect evidence. We use influence functions (Koh and Liang, 2017) to measure how individual training sentences affect model predictions in 15 grammatical phenomena from BLiMP (Warstadt et al., 2020). We filter out a phenomenon’s direct evidence using hand-constructed rules and take the 60,000 most influential remaining sentences as indirect evidence. This indirect evidence substantially affects learning: retraining without it lowers held-out accuracy more than removing a matched random group for every phenomenon across five models (by a median of 12 points). We characterize the indirect evidence by its similarity to our direct evidence filters: it (partially) meets the filters’ structural conditions at a median of 4.7 times the rate of a matched random group. Finally, clustering phenomena by indirect-evidence overlap recovers three of the five BLiMP categories that include multiple phenomena, suggesting that this method can reveal linguistic relationships.
[NLP-75] FederatedBytes: Tokenizer-Free Byte-Level Small Language Models for Efficient Multimodal Federated Learning
【速读】： 该论文旨在解决联邦学习（Federated Learning, FL）在处理自然语言处理（Natural Language Processing, NLP）及多模态分析等现代人工智能任务时，因计算与通信开销过大而导致的可扩展性瓶颈问题。其核心解决方案在于提出一种名为FederatedBytes的小型语言模型，该模型通过字节级（byte-level）文本与多模态数据统一建模，并引入线性注意力机制（linear attention mechanism），实现了对异构输入（如文本、图像等）的统一序列化处理。关键创新点在于：将所有输入视为连续的字节序列，从而避免了复杂的预处理和专用编码器；同时，线性注意力机制显著降低了模型的内存占用与计算复杂度，有效缓解了联邦学习中客户端资源受限与通信成本高的问题。实验结果表明，FederatedBytes在多语言和多模态场景下的情感分析任务中，相较于ByT5和LSTM等基准模型，在保持更高准确率的同时，大幅降低了内存消耗与通信开销，验证了其在实际联邦学习环境中的高效性与可行性。
链接: https://arxiv.org/abs/2610.22906
作者: Ali Sarabadani,Massimo Piccardi,Mehran Abolhasan,Farzad Tofigh,Ian Zhou,Justin Lipman
机构: University of Technology Sydney(悉尼科技大学); Cyber Digital Centre(网络数字中心)
类目: Computation and Language (cs.CL); Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
备注:
Abstract:Federated Learning (FL) is a powerful machine learning framework that allows building a global model from the contributions of multiple, distributed clients, while ensuring the privacy of their local data. However, its heavy computational and communication overheads challenge its use for contemporary AI workloads such as natural language processing (NLP) and multimodal analysis. For this reason, this paper presents a novel small language model that leverages byte-level text and multimodal data processing, enhanced by a linear attention mechanism, to enable manageable Federated Learning of NLP and multimodal analysis tasks. The proposed model, nicknamed FederatedBytes, manages to process text and any other data types, including images, by treating its input as a mere sequence of bytes, while its lean linear attention mechanism allows abating the overall memory and computational footprint. To prove the effectiveness of the proposed model we have carried out an extensive performance evaluation on sentiment analysis tasks in multilingual and multimodal settings over the Multilingual Amazon Reviews and ScienceQA datasets. The experimental results show that FederatedBytes has been able to outperform comparable models such as ByT5 and LSTM, achieving higher accuracy while significantly reducing memory and communication overheads.
[NLP-76] Competence-Based Evaluation of Logical Reasoning in Large Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在逻辑推理能力上存在的不足，尤其是面对语言形式变化时表现出的不一致性问题。现有评估方法主要依赖于特定推理任务上的单一准确率指标，难以有效衡量模型在推理任务中的泛化能力。为此，论文提出一种新的评估范式——基于能力的评估（Competence-Based Evaluation, CBE），其核心在于检验模型是否能够在一组具有相同推理要求但表现形式不同的问题中稳定应用同一项能力。该方案通过参数化生成器控制问题复杂度的同时保持所需推理能力不变，并采用保持答案不变的多种表述方式来测试推理正确性。实验在基础排序任务上验证了六种GPT与Qwen模型的表现，结果显示尽管强模型在有限复杂度范围内表现良好，但在整个测试家族中均未能满足能力一致性标准；链式思维（Chain-of-Thought）和GRPO等策略虽能扩展性能范围，但无法解决所有失败案例；监督微调的LoRA方法则几乎未表现出外部泛化能力。CBE不仅提供了诊断性分析框架，还构建了一套可随未来模型发展而演进的评估协议。
链接: https://arxiv.org/abs/2610.22894
作者: Jize Jiang,Chengxiang Zhai
机构: University of Illinois Urbana-Champaign(伊利诺伊大学厄本那-香槟分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Recent advances in large language models (LLMs) have brought impressive performance across diverse tasks; however, shortcomings remain in logical reasoning, including inconsistencies under linguistic variation. To improve the reasoning capacity of LLMs, we must be able to accurately measure their reasoning capacity. Current benchmarks primarily report single accuracy numbers on selected reasoning tasks. Such a performance-based evaluation methodology says little about a model’s generalization capacity in reasoning. To address this limitation, we argue for a different evaluation paradigm, called competence-based evaluation (CBE), which tests whether models reliably apply the same competence across a family of problems. A parameterized generator scales complexity while keeping the required competence fixed, and matched answer-preserving presentations test correctness beyond one formulation. We instantiate CBE on elementary ordering tasks, scaling input size while varying presentations of the same ordering rule. Across six GPT and Qwen models, performance is reasonably good over bounded complexity ranges, but even strong models fail the competence condition across the tested family. Chain-of-thought and GRPO extend strong performance without resolving all tested failures; supervised LoRA shows little held-out transfer. CBE provides diagnostic profiles and an evaluation protocol that can grow with future models.
[NLP-77] Bengali Hallucination Detection with Retrieval and Programs: A System Study
【速读】： 该论文旨在解决孟加拉语（Bengali）幻觉检测中难以区分无依据的事实性陈述与错误推理的问题，尤其在缺乏参考文本的情况下。其核心挑战在于如何在不依赖外部知识源的前提下，有效识别生成内容中的事实性错误。解决方案的关键在于提出一种仅依赖推理（inference-only）的系统架构，该架构融合了响应条件查询重写、混合检索（hybrid retrieval）以及任务特定的语言模型判断机制。系统通过六条不同的检索路径并行搜索问题-答案和文本段落集合，并由路由模块将每个样本分配至一个270亿参数的文本判别模型或一个140亿参数的智能体（agent），后者可执行多模式操作：比较检索到的答案、直接进行逻辑推理，或生成并执行Python代码以验证结果。系统通过版本化代码、保存的预测结果、执行轨迹及语料库元数据等实现可审计的操作表征。实验表明，该系统在Olikbochon挑战赛中达到私有宏F1值0.956（版本2），处理2,516个样本耗时220.99分钟（双T4 GPU）。值得注意的是，尽管核心工作代码一致，后续运行仍改变六个预测结果，且对1,090条基准相关语料记录的有限审计发现38个评估输入存在完全相同的题目或上下文重叠，因而无法断言其具备完全独立于语料库的泛化能力。研究贡献在于提供了一个可审计、可追踪的检索-计算判别架构，而非组件性能增益的受控估计。作者明确区分了排行榜结果与独立复现的诊断分析，并指出未来更强有效性与可复现性声明所需的关键证据。
链接: https://arxiv.org/abs/2610.22892
作者: Shamim Hasan,Samiya Ahmed Chowdhury
机构: Chittagong University of Engineering and Technology (吉大港工程与技术大学); BNIST (孟加拉国国家信息技术研究所)
类目: Computation and Language (cs.CL)
备注: 8 pages
Abstract:Bengali hallucination detection must distinguish unsupported factual claims from incorrect reasoning, often without a supplied reference passage. We present and characterize an inference-only system that combines response-conditioned query rewriting, hybrid retrieval, and task-specific language-model judging. Six retrieval routes search question-answer and passage collections; a router directs each example to a 27B textual judge or a 14B agent that can compare retrieved answers, reason directly, or generate and execute Python. We characterize the system using version-specific code, saved predictions, execution traces, and corpus metadata from the Olikbochon challenge. Kaggle associates a displayed private macro-F1 of 0.956 with Version 2; its saved run processes 2,516 examples in 220.99 minutes on two T4 GPUs. Of 253 math-routed examples, 88 terminate through candidate matching, while execution traces expose 49 stderr-bearing run events across 35 examples. A later run changes six predictions despite identical core worker code. A bounded audit of 1,090 benchmark-related corpus records finds exact question or context overlap for 38 processed evaluation inputs, precluding an unqualified claim of corpus-independent generalization. The contribution is a concrete retrieval-and-computation judging architecture with an auditable operational characterization, not a controlled estimate of component gains. We distinguish the reported leaderboard result from independently recomputed diagnostics and identify the evidence needed for stronger effectiveness and reproducibility claims.
[NLP-78] NLData2Opt: Benchmarking Parameter Derivation and Prediction for Modeling Optimization Problems from Natural Language and Data
【速读】： 该论文旨在解决现有优化问题建模基准在评估大语言模型（LLM）能力时存在的理想化偏差问题，即其依赖于完全且相关性高的输入数据，而忽略了真实场景中普遍存在的噪声、不完整及冗余数据。实际优化流程中，模型需具备数据驱动的多阶段推理能力，包括从含干扰属性的数据中提取关键参数、基于已有数据属性进行转换以计算缺失参数，以及根据历史观测值预测未来不可用数值。为此，论文提出NLData2Opt基准，其核心在于将问题分解为问题描述与原始数据两部分，并设计三阶段任务以衡量模型在真实复杂数据环境下的建模能力。该基准的关键创新在于引入了数据清洗、特征推导和时间序列预测等现实挑战，揭示了当前主流基准所报告的性能提升无法直接转化为实际优化建模中的准确性，从而推动对生成式人工智能在优化领域应用的真实效能评估。
链接: https://arxiv.org/abs/2610.22889
作者: Lijia Liu,Kyohei Atarashi,Jiyi Li,Koh Takeuchi,Shunji Umetani,Hisashi Kashima
机构: Kyoto University(京都大学); Hokkaido University(北海道大学); Recruit Co., Ltd.(リクルート株式会社)
类目: Computation and Language (cs.CL)
备注:
Abstract:Formulating optimization problems from raw inputs of problem descriptions and data is time-consuming and costly. While Large Language Models (LLMs) offer a promising way to automate this process, existing benchmarks overestimate their capabilities by relying on idealized inputs where all data is relevant and complete. This diverges from reality, where data-centric reasoning is required to extract, compute, and predict parameters from noisy sources. To address this, we introduce NLData2Opt, a benchmark to evaluate LLMs on the multi-stage reasoning inherent in practical optimization workflows. NLData2Opt decomposes problems into problem descriptions and data, and requires models to (1) extract relevant parameters from data containing misleading attributes, (2) compute missing parameters derived by transforming existing data attributes, and (3) predict unavailable future values from historical observations. Evaluation results show substantial performance decline relative to existing benchmarks, indicating that benchmark progress does not directly translate to formulation accuracy in practical optimization workflows. NLData2Opt is available at this https URL.
[NLP-79] Batched Contextual Reinforcement II: Does RLVR Need Long Rollout Windows?
【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）因长滚动窗口（long rollout windows）导致的高计算成本问题。传统方法中，若简单缩短滚动窗口，会导致正确答案在截断前无法完成，从而丧失正确的奖励信号，使群体相对准确性的优势消失。为解决这一矛盾，论文提出并发约束强化（Concurrent Constraint Reinforcement, CCR），作为批处理上下文强化（Batched Contextual Reinforcement, BCR）的技术延续。其核心创新在于将三个独立可验证的问题并行置于一个固定长度的训练完成窗口内，允许在截断前完成的正确答案保留奖励，同时通过共享预算机制激励模型进行更短但有效的推理，而无需显式引入长度惩罚项。该机制通过条件性奖励方差与更新方差分析得到理论支持。实验表明，在4B和8B数学推理模型上，相较于使用32k窗口的单问题梯度回溯策略（GRPO），采用8,000令牌窗口的CCR分别将每步训练消耗的令牌数从975,464降至174,957（节省82.1%）和从793,832降至165,788（节省79.1%），且训练轨迹显示准确率奖励持续上升，同时推理长度缩短，单问题评估也证实了推理质量未下降。研究结果表明，保持奖励有效性的训练结构设计是实现短窗口下强化学习低成本训练的可行路径。
链接: https://arxiv.org/abs/2610.22887
作者: Bangji Yang,Hongbo Ma,Ruihan Guo,Jiajun Fan,Xi Zhu,Ge Liu
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 22 pages, 3 figures, 2 tables
Abstract:Long rollout windows make reinforcement learning with verifiable rewards (RLVR) expensive, yet simply shortening them can remove the correctness signal: when every single-problem rollout truncates before an answer, group-relative accuracy advantages vanish. As a technical continuation of Batched Contextual Reinforcement (BCR), we study how concurrent rewards relax this dependence on long windows. Concurrent Constraint Reinforcement (CCR) places three independently verifiable problems in one capped training completion. Correct answers completed before truncation retain credit, while a shared budget encourages shorter useful reasoning without an explicit length penalty. We connect this mechanism to conditional reward- and update-variance analyses and study 4B and 8B mathematical reasoning models. Against single-problem GRPO with a 32k window, CCR with an 8,000-token window reduces tokens per training step from 975,464 to 174,957 on 4B and from 793,832 to 165,788 on 8B, saving 82.1% and 79.1%, respectively. An 8B training trace shows increasing accuracy reward alongside shorter completions. Single-problem evaluations provide supporting evidence of retained reasoning quality. These results identify reward-preserving training structure as a practical route to lower per-step cost in short-window RLVR.
[NLP-80] Identify Classify Act: Per-Instruction Controlled Compliance for LLM s over External Data
【速读】： 该论文旨在解决大语言模型（LLM）在处理外部数据时面临的指令层面矛盾问题：单个文档或工具返回中常混杂合法指令与恶意提示注入，模型需在遵循前者的同时拒绝后者，而现有防御机制要么全盘信任、要么全盘拒绝对外部指令，导致安全-效用权衡边界坍缩，迫使模型在过度拒绝与过度遵从之间做出非此即彼的选择。其核心解决方案是引入逐指令信用归因（per-instruction credit attribution），并形式化为外部数据指令分类（External-Data Instruction Classification），即对每个嵌入式指令进行识别、分类与执行的三阶段决策。为此，作者提出EDIC（External-Data Instruction Classification）框架，采用分步奖励机制实现逐指令合规性控制：通过场景感知的流水线将有益与有害指令混合注入，并结合反事实过滤增强鲁棒性；同时设计一个由监督微调（SFT）初始化、经广义近端策略优化（GRPO）迭代优化的生成式判别器，实现指令级别的合规性归因。该判别器输出独立奖励信号，分别激励有益合规、惩罚有害合规，用于强化学习训练。在两种骨干模型和五个基准测试中，EDIC显著优于最强基线，在保持更高安全性的同时提升效用，相较之前方法在Qwen3-4B上实现有害指令拒绝率最高提升7.7%，Arena-Hard-Auto任务效用提升12.5%。
链接: https://arxiv.org/abs/2610.22874
作者: Zesheng Shi,Xue Li,Chenrui Dai,Jing Li
机构: Harbin Institute of Technology, Shenzhen, China(哈尔滨工业大学深圳校区)
类目: Computation and Language (cs.CL); Cryptography and Security (cs.CR)
备注:
Abstract:LLMs over external data face a per-instruction tension: a single document or tool return often interleaves legitimate directives with adversarial prompt injections, and the model must follow the former while rejecting the latter. Existing defenses suppress or trust all external instructions, collapsing the safety-utility frontier and forcing a choice between over-refusal and over-compliance. We argue the missing capability is \textbfper-instruction credit attribution and formalize it as External-Data Instruction Classification, an identify-classify-act decision over each embedded directive. We propose \textbfEDIC, a decomposed reward framework for per-instruction compliance. A scene-grounded pipeline injects mixed beneficial and harmful instructions with counterfactual filtering, while a generative judge is initialized by SFT and refined with GRPO to attribute compliance at the instruction level. The judge then produces a reward that separately encourages beneficial compliance and penalizes harmful compliance for RL training. Across two backbones and five benchmarks, EDIC improves both safety and utility over the strongest prior defense, lifting harmful-rejection by up to 7.7% and Arena-Hard-Auto utility by 12.5% on Qwen3-4B. The code repository is hosted at this https URL.
[NLP-81] Uncertainty-Aware Budget Allocation for Efficient LLM Tree Search
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在多步推理中树搜索（Tree Search）效率低下的问题，即现有方法在每一步均采用固定预算进行扩展，而未考虑扩展的边际效用。其核心问题是：在树搜索过程中，从同一中间状态生成的后续路径存在高度重复性，导致大量计算资源浪费。解决方案的关键在于提出一种无需训练的不确定性感知分配（Uncertainty-Aware Allocation, UAA）框架，通过分析兄弟节点（sibling actions）之间的语义分散度（semantic dispersion）来估计其不确定性，进而动态调整每个子树的扩展预算。该方法基于信息论分析，将每步预算视为对残差答案熵的蒙特卡洛估计样本量，证明了预算应随兄弟节点间分散度的增加而减少。UAA通过解耦“价值”（决定候选路径是否值得探索）与“不确定性”（决定需扩展的数量），实现不确定性感知剪枝和预算分配，在数学推理、逻辑推理及规划等多个基准上实现了最高达90%的令牌消耗降低和9倍的延迟减少，且在四种树搜索算法及开源与闭源模型间具有良好泛化能力。
链接: https://arxiv.org/abs/2610.22858
作者: Jianguo Huang,Zhiyi Lyu,Fuxiang Zhang,Yanchen Deng,Deheng Ye,Bo An
机构: Nanyang Technological University, Singapore
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Tree search enhances multi-step reasoning in large language models (LLMs), yet prevailing methods expand every step with a fixed budget, irrespective of the marginal utility of expansion. We first show, across existing search algorithms, that continuations sampled from a shared intermediate state are largely duplicates and collapse to a few distinct answers. Critically, we find that the semantic dispersion of sibling actions predicts the redundancy of their downstream continuations, permitting a subtree’s budget to be fixed before expansion. An information-theoretic analysis motivates this: treating the per-step budget as the Monte Carlo sample size for estimating a step’s residual answer entropy, we derive a sufficient budget that shrinks with sibling dispersion. We accordingly propose \textbfUncertainty-Aware Allocation (UAA), a training-free framework that decouples value from uncertainty: values determine which candidates are promising, while uncertainty, estimated from sibling dispersion at no additional model cost, determines how many merit expansion, via uncertainty-aware pruning and uncertainty-aware budgeting. Across mathematical reasoning, logical reasoning, and planning benchmarks, UAA reduces token usage by up to 90% and latency by up to 9\times at comparable accuracy, generalizing across four tree search algorithms and both open- and closed-source LLMs.
[NLP-82] he Reliance Hypothesis: Chain-of-Thought Faithfulness Rises as Language Resource Level Falls
【速读】： 该论文旨在解决生成式AI在多语言环境下链式思维（Chain-of-Thought, CoT）忠实性（faithfulness）的评估问题，尤其关注资源水平（resource level）与任务类型（知识密集型 vs. 推理密集型）对模型依赖推理机制的影响。现有研究主要聚焦于英语，而低资源语言中模型的知识获取能力受限，可能改变其推理依赖模式，但这一现象尚未被系统探讨。本文首次将CoT忠实性与语言资源水平及任务类型关联分析，在八种语言的多语言大规模多任务理解（MMMLU）平行多选题数据集上发现：在知识密集型任务中，随着语言资源水平下降，模型的CoT忠实性反而上升；而在推理密集型任务中则无此趋势，该现象在四种不同模型和多种忠实性度量下均一致。为此，论文提出“依赖假说”（Reliance Hypothesis），即当模型无法通过记忆召回答案时（如在低资源语言中的知识密集型任务），会更依赖其内部推理过程。模型置信度部分解释了资源水平的影响，但无法解释知识-推理之间的差异。值得注意的是，低资源语言中模型整体更倾向于使用推理，而非对单个答案的推理强度更高，这揭示了一种辛普森悖论（Simpson’s paradox）：尽管模型在低资源语言中总体信心较低，但其在同语言内信心较低的预测并未表现出更强的推理依赖。因此，该研究指出，CoT监控在低资源语言中可能最具信息价值，尽管这些语言的验证往往最不充分。
链接: https://arxiv.org/abs/2610.22841
作者: Lucas Resck,Jingyi Sun,Qianli Wang,Isabelle Augenstein,Anna Korhonen
机构: Language Technology Lab, University of Cambridge(剑桥大学语言技术实验室); University of Copenhagen(哥本哈根大学); Technische Universität Berlin(柏林工业大学); German Research Center for Artificial Intelligence (DFKI)(德国人工智能研究中心)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 21 pages, 10 figures, 16 tables
Abstract:Chain-of-Thought (CoT) faithfulness matters for monitoring, explaining and safely deploying large language models, yet it has been studied mostly in English. Languages differ in level, the amount of text available to learn from, so models know less in lower-resource languages, which may change how much they rely on their reasoning. To our knowledge, we are the first to relate CoT faithfulness to both resource level and task type (knowledge- or reasoning-intensive). On parallel multiple-choice questions from Multilingual Massive Multitask Language Understanding (MMMLU) in eight languages, faithfulness rises as resource level falls on knowledge-intensive tasks but not on reasoning-intensive ones, a gap that points the same way across four models and several faithfulness measures. We propose, and find support for, the Reliance Hypothesis: models rely more on their reasoning when they cannot recall the answer, as in lower-resource languages on knowledge-intensive tasks. Model confidence partly explains the effect of resource level, but not the knowledge-reasoning gap. Languages in which models are less confident rely more on reasoning, yet less confident answers within a language do not, a Simpson’s paradox. In lower-resource languages, models rely on their reasoning for more answers, not more strongly for each. CoT monitoring may therefore be most informative in lower-resource languages, where it is least often validated.
[NLP-83] oo Emotional to Be Fake? Intra-Utterance Emotion Shifts Expose Failure Modes in Speech Deepfake Detectors
【速读】： 该论文旨在解决生成式语音合成中情感动态变化对深度伪造（deepfake）检测效果的影响问题，特别是探究情感转移（emotion-changing）的合成语音在检测任务中的表现。其核心挑战在于，尽管当前生成式语音合成技术已能实现单句内情感的灵活切换，但这种情感动态变化是否会影响检测系统的判别能力尚不明确。解决方案的关键在于提出EmoShift基准数据集，该数据集系统性地对比了情感稳定（AA）与情感变化（AB）的语音样本，同时控制静态情感身份和声学混杂因素，从而分离出情感轨迹本身对检测性能的影响。实验结果表明，情感变化可使合成语音更接近真实语音，导致在固定阈值下检测器误检率最高上升25.3个百分点；而真实语音中的情感变化则导致误报率最高上升24.4个百分点。进一步分析揭示，不同检测器对情感演化轨迹具有特定敏感性，且该敏感性无法仅由静态情感类别或拼接伪影完全解释。研究强调了在构建鲁棒深伪语音检测系统时，必须纳入情感动态特征的建模。
链接: https://arxiv.org/abs/2610.22832
作者: Jingyi Chen,Surya Koppisetti
机构: Reality Defender
类目: ound (cs.SD); Computation and Language (cs.CL)
备注: Preprint. Under review
Abstract:Expressive speech synthesis can now change emotion within a single utterance, yet its impact on deepfake detection remains unclear. We introduce EmoShift, a benchmark comparing emotion-stable (AA) and emotion-changing (AB) speech to examine these effects while accounting for static emotion identity and acoustic confounds. It comprises 7,240 bona fide utterances from four source datasets and 9,720 synthetic utterances generated using four expressive TTS systems. Across seven detectors, we find that emotion changes can make synthetic speech appear more genuine, increasing spoof miss rates by up to 25.3 percentage points on OMG-Emotion at fixed thresholds. Conversely, emotion-changing bona fide speech increases false-alarm rates by up to 24.4 percentage points on IEMOCAP. Further analyses reveal detector-specific sensitivities to emotion trajectories, not fully explained by static emotion identity or splicing artifacts. These findings underscore the importance of emotion dynamics in deepfake detection.
[NLP-84] Causal Edit Serialization
【速读】： 该论文旨在解决自回归语言模型在生成文本时无法高效实现局部修订的问题，即模型在生成过程中难以对已输出或输入的文本进行精确的插入、删除、替换等编辑操作。其核心解决方案是提出一种名为因果编辑序列化（Causal Edit Serialization, CES）的框架，通过引入特殊的指针标记（pointer tokens）与预测的源地址配对，将各类编辑操作（如插入、删除、替换、复制和移动）显式地定位到特定位置。关键在于：利用轻量级定位头和双索引旋转位置编码，在保持标准因果注意力掩码的前提下，支持地址预测及对先前输出指针的条件依赖；同时通过结合随机噪声污染、模型生成的修复以及重复文本的预训练任务，使模型具备学习序列化修复的能力。实验表明，CES模型在不同规模下均能完整执行编辑任务，且在干净文本的压缩效率上仅略高于标准自回归模型；此外，经过监督微调或强化学习后训练的模型在语法纠错任务中表现出优于有害修改的有益变更能力，并在保留原始上下文及其缓存状态的基础上提升了后续文本生成质量。因此，CES为生成与局部修订提供了一个统一接口，具备向迭代优化及其他基于指针-地址变换操作扩展的潜力。
链接: https://arxiv.org/abs/2610.22820
作者: Louis Robinson
机构: 独立研究者(Independent Researcher)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 37 pages, including appendices and checklist. Code: this https URL ; data and checkpoints: this https URL
Abstract:Standard autoregressive language models generate text sequentially, but revision requires referring back to tokens in the input context or those already generated. Causal Edit Serialization (CES) is introduced as a framework for expressing such revisions within the same causal token stream. Special pointer tokens are paired with predicted source addresses, allowing insertion, deletion, replacement, copying and movement to be localized explicitly. A lightweight localization head and dual-index rotary position encoding support both address prediction and conditioning on previously emitted pointers, while retaining a standard causal attention mask. A pretraining procedure combines ordinary language modeling with serialized repairs derived from random corruptions, model-generated edits and repeated text. Across model scales and editing baselines, CES models demonstrate complete applied edits with modestly higher clean-text bits per byte than clean autoregressive models. Supervised and reinforcement-learning post-training provide a grammatical-error-correction proof of concept, producing more beneficial than harmful changes under reference-blind model judging. Correct repair programs also improve subsequent text prediction while the original context and its cached states are retained. CES provides a common interface for generation and localized revision, with potential extensions to iterative refinement and additional pointer-addressed transformations.
[NLP-85] Watermarkability: Measuring and Improving Language-Model Compatibility with Logit-Based Watermarks
【速读】： 该论文旨在解决生成式 AI (Generative AI) 中文本水印技术在实际应用中面临的核心挑战：不同模型、上下文及水印规则下，基于对数几率（logit）的水印方法因基础模型对目标词元集合的原始概率分布差异，导致水印遵循率与困惑度（PPL）增加不一致的问题。其核心解决方案在于提出“水印可实现性”（watermarkability）这一新概念，用于量化模型与特定水印请求分布之间的兼容性，并通过引入水印可实现性代价（watermarkability cost）来衡量将目标词元集的概率质量提升至指定水平所需的最小期望KL散度。为降低该代价，作者提出一种两阶段优化方法：第一阶段利用从16路词汇划分中采样的并集，在条件低秩适配器（target-set-conditioned low-rank adapter）上进行训练以增强目标集合的概率质量；第二阶段采用闭式控制器，施加最小必要的均匀对数几率偏置，以达到目标概率质量。实验表明，该方法在六种水印调度策略下，显著提升了水印可靠性（所有调度下样本落入目标集比例均超过99.80%），并在OPT-125M和Qwen3-8B上分别使提示条件下的平均PPL降低46.3%和5.2%，显著优于固定偏置基线。
链接: https://arxiv.org/abs/2610.22810
作者: Zewei Deng,Mohamed Seif,Liyan Xie
机构: University of Minnesota(明尼苏达大学); Oakland University(奥克兰大学); Princeton University(普林斯顿大学)
类目: Computation and Language (cs.CL); Cryptography and Security (cs.CR)
备注: 16 pages, 2 figures, 10 tables. Code available at this https URL
Abstract:Language-model watermarks can help identify model-generated text, encode provenance information, and detect post-generation edits. A broad class of logit-based watermarks increases the logits of a designated token set at each decoding step. Because a base model’s native probability mass on this set varies across contexts, models, and watermark rules, the same logit bias can yield different adherence rates and increases in perplexity (PPL). We introduce watermarkability to characterize the compatibility between a model and the token sets requested by a watermark. At a target mass \tau , we quantify this property through a watermarkability cost: the expected minimum KL divergence from the base-model distribution required to assign probability at least \tau to the requested set. A lower cost means that the model is more watermarkable for that request distribution. To reduce this cost, we propose a two-stage method. In the first stage, we train a target-set-conditioned low-rank adapter on unions drawn from a 16-way vocabulary partition to increase the requested set’s probability mass. In the second stage, a closed-form controller applies the smallest remaining uniform logit bias needed to reach the target mass. We evaluate the method on six watermark schedules using OPT-125M and Qwen3-8B. Across the 12 model-schedule pairs, the pre-adaptation watermarkability cost correlates with the log-PPL increase under a near-hard fixed-bias baseline (pooled Pearson r=0.929 ). For every schedule, at least 99.80% of sampled tokens fall in the target set requested at their decoding step. At this reliability, the adapter and controller reduce average per-schedule prompt-conditional PPL by 46.3% on OPT-125M and 5.2% on Qwen3-8B relative to the fixed-bias baseline.
[NLP-86] On the Position Bias of Large Language Models : A Comprehensive Study
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）中普遍存在的位置偏差（position bias）问题，该问题严重影响多选题评估的公正性以及检索增强生成（retrieval-augmented generation）的可靠性。现有研究对这一现象的理解受限于模型规模覆盖范围有限及以英语为主的基准测试，难以反映前沿模型在多语言场景下的真实表现。本文通过系统性实验，考察了从十亿级到万亿级参数规模的模型在五种语言中的多选题与多文档任务上的位置偏差行为。研究发现，尽管小规模模型和非英语语言中仍存在显著的位置偏差，但随着模型规模增大以及采用标准链式思维（Chain-of-Thought）提示策略，该偏差显著缓解；更重要的是，具备扩展推理能力的新兴“思考型”模型（thinking models）几乎不受位置偏差影响，表明通过提升模型推理能力可成为一种虽成本高昂但具有普适性的解决方案。
链接: https://arxiv.org/abs/2610.22807
作者: Yining Li,Huayang Li,Lemao Liu,Taro Watanabe
机构: Jilin University (吉林大学); Nara Institute of Science and Technology (奈良先端科学技術大学院大学); Fudan University (复旦大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Recent studies have shown that position bias remains a critical vulnerability in large language models (LLMs), undermining both the integrity of multiple-choice evaluations and the reliability of retrieval-augmented generation. However, existing understanding of this phenomenon is largely derived from studies with limited model coverage and English-centric benchmarks, leaving the behavior of frontier models and diverse languages underexplored. In this work, we conduct a comprehensive investigation of position bias across model scales ranging from billion-level to trillion-level parameters, covering both multiple-choice and multi-document tasks in five languages. Our experiments reveal that while position bias remains noticeable in small-scale models and Non-English languages, increasing model scale and employing standard Chain-of-Thought prompting significantly mitigate this issue. Moreover, we demonstrate that emerging \emphthinking models with extended reasoning capabilities are nearly immune to position bias, suggesting a costly but general approach toward resolving this vulnerability.
[NLP-87] Veil-Then-Reveal: Protecting PII Privacy via Reversible Pseudo-Anonymization in Zero-Trust User–LLM Interaction
【速读】： 该论文旨在解决在零信任威胁模型下，用户使用远程大语言模型（LLM）处理包含个人身份信息（PII）的敏感文档时所面临的隐私泄露问题。现有方法或通过屏蔽/删除PII导致信息损失，或采用传统加密方式破坏了文本的词汇与结构特征，致使LLM无法有效理解与生成内容。其核心解决方案为“掩蔽-还原”（Veil-then-Reveal, VtR）框架，关键在于实现用户端的伪匿名化：在数据离开用户设备前，利用格式保持加密（FPE）处理结构化PII（如电话号码、信用卡号），结合密钥哈希字典与安全存储库处理语义型PII（如姓名、地点、组织名），生成同类型、在词表内的真实感替代符，从而保留原始文本的语言与结构特性；同时通过会话级映射实现跨会话不可关联性，并支持用户端可逆还原，确保响应中替换项能完整回溯至原始值。此外，引入等级分类的FPE策略用于日期字段，在维持格式与模糊性的同时降低密文计算开销。实验表明，VtR在三个包含11,322个标注PII实例的数据集上实现了97.9%的检测召回率、95.2%的已检测值保护率及99.9%的端到端往返恢复率，所有被回显的替代符均实现100%成功还原。
链接: https://arxiv.org/abs/2610.22800
作者: Mohamed Shaaban,Mohamed Elmahallawy,Thomas E. Carroll
机构: 未知
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL)
备注:
Abstract:Large language models (LLMs) increasingly process sensitive documents and personal information, yet sending prompts containing personally identifiable information (PII) to remote providers exposes data subjects and organizations to risks from logging, retention, model training, and unintended disclosure. Existing anonymization methods often mask or remove PII, while conventional encryption eliminates the lexical and structural cues needed for effective LLM processing. We present Veil-then-Reveal (VtR), a user-side PII pseudo-anonymization framework for privacy-preserving LLM interaction under a zero-trust threat model. VtR treats remote LLM providers as untrusted and sends only pseudo-anonymized prompts, keeping plaintext PII on the user. VtR first detects PII and replaces it with realistic, same-type, in-vocabulary surrogates before prompts leave the user’s device, preserving their linguistic and structural properties. It combines format-preserving encryption (FPE) for structured PII, such as phone and credit-card numbers, with a keyed-hash dictionary and secure vault for semantic PII, such as names, locations, and organizations. Session-specific mappings provide cross-session unlinkability, while user-side restoration reversibly replaces surrogates with their original values in LLM responses. We further introduce a rank-class FPE strategy for dates that preserves format and ambiguity while reducing cipher evaluations. Evaluation on three datasets containing 11,322 annotated PII instances achieves 97.9% detection recall, 95.2% protection of detected values, and 99.9% round trip; 100% of echoed surrogates end-to-end round-trip restoration.
[NLP-88] Causal Latent Workspaces for Looped Transformers
【速读】： 该论文旨在解决在深度共享（shared-depth）Transformer架构中，随着层数加深，各位置无法有效传递和利用其因果前缀（causal prefix）的紧凑摘要信息的问题。现有方法虽通过重复使用Transformer块减少参数量，但缺乏专门的状态机制来支持位置间在多轮迭代中交换上下文相关的压缩表示。为此，作者提出因果潜在工作区Transformer（causal latent workspace Transformer），其核心创新在于引入一个锚定的、四轮交互的共享深度结构：每个位置在每轮中将门控向量写入与前缀相关的潜在槽（latent slots），并在下一轮应用共享块前，通过内容寻址方式检索该前缀的摘要信息。这一路径与传统的因果自注意力并行互补，同时保持原始输入嵌入在每轮中的可访问性。实验结果表明，在三个字节级语料库、三种随机种子及两种参数预算下，该模型在六组对比中于五组取得最优平均保留困惑度（held-out perplexity）。在101.71M参数规模下，其在WikiText-103 raw、enwik8和TinyStories上分别达到2.7820、2.7843和1.6823的困惑度，相较最强非工作区对照模型分别提升1.06%、0.73%和1.01%，且所有九组带种子标签的比较均支持工作区模型的优势。在0.918M参数规模下，其在WikiText-103和TinyStories上表现领先，而在enwik8上条件循环结构更优。架构消融实验进一步验证了引入工作区在全部六种设置中优于锚定循环结构，并支持每轮刷新因果摘要，从而确立了因果性、内容可寻址的前缀工作区作为环形Transformer的有效结构增强机制。
链接: https://arxiv.org/abs/2610.22796
作者: Ilia Zaznov
机构: 未知
类目: Computation and Language (cs.CL)
备注: 20 pages, 5 figures. Code and reproducibility materials: this https URL
Abstract:Reusing a Transformer block across depth increases computation without replicating its parameters, but successive passes have no dedicated state through which positions can exchange compact summaries of their causal prefix. We introduce the causal latent workspace Transformer: an anchored, four-round shared-depth architecture in which each position writes gated vectors into prefix-specific latent slots and retrieves a content-addressed summary before the next shared block application. This path complements causal self-attention while the original input embedding remains accessible at every round. Across three byte-level corpora, three seeds, and two matched parameter budgets, the proposed model achieves the best mean held-out perplexity in five of six settings against five Transformer controls. At 101.71M parameters it leads on WikiText-103 raw, enwik8, and TinyStories, reaching perplexities 2.7820, 2.7843, and 1.6823; these improve upon the strongest non-workspace control on each corpus by 1.06%, 0.73%, and 1.01%. All nine seed-labelled comparisons at this budget favor the workspace. At 0.918M parameters it leads on WikiText-103 and TinyStories, whereas a conditioned loop is stronger on enwik8. Architectural ablations show that adding the workspace improves upon the anchored loop in all six settings and support refreshing its causal summaries at each round. Together, these results identify a causal, content-addressable prefix workspace as an effective structural augmentation of looped Transformers.
[NLP-89] Prune the Leaky Neurons: Mitigating Private Data Leakage in Fine-Tuned LLM s via Joint-Embedding-Guided Pruning
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在私有数据上进行微调时面临的双重挑战：一是模型可能记忆并泄露敏感信息（如个人身份信息，PII），二是如何在移除隐私数据的同时保持模型的有用性。现有方法如数据清洗和差分隐私常导致性能下降，而基于隐私神经元编辑的方法缺乏无PII的参考基准来区分与身份相关的神经元与支持语义理解的神经元。为此，本文提出一种后处理神经元剪枝框架JEPA-Prune，其核心在于训练一个仅使用去标识化视图的公共模型作为无PII参考。该公共模型通过联合嵌入预测架构（Joint-Embedding Predictive Architecture, JEPA）对同一组记录的掩码与替代视图进行建模，从而捕获领域知识而不包含敏感标识。通过跨模型表示探针量化私有模型中泄露记录的表征与该无PII参考之间的偏差，结合积分梯度与尺度不变特异性评分，识别出对PII响应敏感的神经元，并在无需重新训练的情况下将其剪枝。实验表明，仅剪枝1.6%的神经元即可使PII推理准确率降低7.4倍、PII提取召回率降低13.9倍，同时保持未见文本困惑度与原始模型相差不超过16%，显著优于传统隐私神经元编辑方法（后者困惑度至少增加26倍）。因此，该方案的关键在于利用无PII参考模型实现精准的敏感神经元定位与高效剪枝，兼顾隐私保护与模型性能。
链接: https://arxiv.org/abs/2610.22792
作者: Mohamed Shaaban,Mohamed Elmahallawy
机构: Washington State University (华盛顿州立大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Fine-tuning large language models (LLMs) on proprietary records enables domain adaptation but introduces two challenges: (i) privacy leakage of memorized private data, such as personally identifiable information (PII), and (ii) balancing private data removal with utility preservation. Existing defenses, including data sanitization and differential privacy, often degrade performance, while privacy-neuron editing lacks a PII-free reference to distinguish identity-specific neurons from those supporting useful text. We propose JEPA-Prune, a post-hoc neuron pruning framework guided by a model trained exclusively on PII-free views. Alongside the fine-tuned private model, JEPA-Prune, trains a public model on surrogate and masked views of the same records using a joint-embedding predictive architecture (JEPA), capturing domain knowledge without identifiers. A cross-model representation probe quantifies deviations between private representations of leaked records and this PII-free reference. Integrated gradients and a scale-invariant specificity score then identify PII-responsive neurons, which are pruned without retraining. Experiments show that pruning just 1.6% of neurons reduces PII inference accuracy by up to 7.4x and PII extraction recall by up to 13.9x, while keeping unseen-text perplexity within 16% of that of the unpruned model for most LLMs. In contrast, privacy-neuron editing increases perplexity by at least 26x.
[NLP-90] Self-Evolving Coding Agents via Harness-Policy Optimization
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）编码代理中策略模型（policy model）与调度器（harness）之间协同优化不足的问题。传统方法通常固定一方（如固定 harness）来优化另一方，导致二者无法动态适配：一方面，针对某一版本策略设计的 harness 可能无法有效覆盖策略更新后的新失败模式；另一方面，策略学习受限于固定 harness 所生成的轨迹，难以探索更复杂的任务场景。为此，本文提出 Self-HPO——一种递归自进化框架，通过交替优化 harness 与 policy，实现二者的协同演进。其核心在于：在每轮迭代中，当前策略首先分析自身执行失败，自主生成新的 harness 候选集（H-step），随后基于所选 harness 采用强化学习（Reinforcement Learning, RL）更新策略（P-step），并将训练过程中识别出的高难度任务作为下一轮 harness 搜索集，确保 harness 优化始终聚焦于策略演进过程中的新兴失败模式。实验表明，Self-HPO 在三个编码基准上均显著优于仅优化 harness 或仅优化 policy 的基线，使 Qwen3.5-35B-A3B 的平均解决率从 57.8% 提升至 64.8%，且性能提升在分布外设置下仍具泛化能力。进一步分析显示，优化后的 harness 与 policy 在与原始组件配对时仍保持增益，而两者联合带来的性能提升远超各自独立改进之和，表明其存在显著协同效应。此外，引入更强外部模型（如 Opus 4.8）作为 harness 提议者可进一步提升框架表现，验证了该方法的可扩展性与通用性。
链接: https://arxiv.org/abs/2610.22791
作者: Zhepei Wei,Wenlin Yao,Baolin Peng,Danqing Wang,Qianhui Wu,Hao Cheng,Yu Meng,Jianfeng Gao
机构: Microsoft Research(微软研究院); University of Virginia(弗吉尼亚大学); Carnegie Mellon University(卡内基梅隆大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model (LLM) agents have shown promising potential in coding tasks, and their performance is jointly determined by the policy model and the harness that mediates their interaction with the environment. Indeed, harness and policy are mutually reinforcing and evolving only one component is suboptimal—a harness tuned for one checkpoint may not target the failure modes of an updated policy, while policy learning remains limited to the trajectories elicited by a fixed harness. In light of this, we propose Self-HPO, a recursive self-improving framework for coding agents that alternates harness and policy optimization, using the evolving policy as its own harness proposer. Specifically, in each round, the current policy first analyzes its own failures and proposes new harness candidates (H-step). The policy is then updated through reinforcement learning (RL) under the selected harness (P-step), and hard tasks identified during training form the next harness search set, keeping harness optimization targeted to the policy’s evolving failure modes. Across three coding benchmarks, Self-HPO consistently outperforms harness-only and policy-only baselines, improving Qwen3.5-35B-A3B’s average resolve rate from 57.8% to 64.8%, with gains generalizing to out-of-distribution settings. Further analysis shows that both optimized components retain gains even when paired with their base counterparts, while their joint gain exceeds the sum of their individual gains. Self-HPO also improves reliability (Pass@1) and expands the capability frontier (Pass@k) over the base agent. Additionally, we show that with a stronger external model (e.g., Opus 4.8) as the harness proposer, the same framework yields a further performance boost.
[NLP-91] hunder-DeID: Boundary-Aware Korean De-identification from Court Judgments to Broader Sensitive Documents
【速读】： 该论文旨在解决生成式AI在处理韩语敏感文档去标识化（De-identification）时的跨领域迁移性问题，尤其针对韩语特有的黏着语结构带来的分词边界模糊与实体识别精度下降等挑战。核心问题包括：韩语的黏着特性导致基于字节对编码（Byte Pair Encoding, BPE）的分词器频繁将个人身份信息（PII）与相邻助词合并，造成实体跨度（span）恢复困难；同时，现有PII分类体系在法律文本中缺乏足够的粒度以支持精准标注。为此，本文提出Thunder-DeID+，其关键创新在于融合精细化的三层PII分类体系与一种边界感知型分词器（boundary-aware tokenizer），通过引入N-gram统计特征和边界评分机制，对预训练BPE词汇表进行增强，从而更准确地对齐分词边界与实际PII跨度。实验采用独立于分词器的词段级评估协议，在完整判决书上，该方法使不同规模模型（150M–1B）的严格类型F1分数从0.76–0.77显著提升至0.87–0.88，主要得益于精确率的提高而非模型规模扩大；在对1,000份更广泛敏感文档微调后，模型进一步达到0.93的严格类型F1分数，验证了其在跨领域场景下的有效迁移能力。
链接: https://arxiv.org/abs/2610.22733
作者: Gyuseong Lee,Gyeongje Cho,Jongmin Kim,Heejin Kim,Yeonkyoung So,Hyunggeun Jeon,Hyunji M. Park,Sangho Kim,Yoonhee Park,Youngbin Pyo,Jaejin Lee
机构: Seoul National University (首尔国立大学); Dept. of Data Science, Seoul National University (数据科学系，首尔国立大学); Dept. of Computer Science and Engineering, Seoul National University (计算机科学与工程系，首尔国立大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:This study examines the transferability of a de-identification model developed for Korean court judgments to a broader range of sensitive documents, focusing on boundary precision, label granularity, and recall. The agglutinative structure of Korean causes Byte Pair Encoding (BPE) tokenizers to frequently merge Personally Identifiable Information (PII) with adjacent particles, which complicates accurate span recovery. Furthermore, current PII taxonomies lack sufficient granularity for legal texts. To address these issues, this study introduces Thunder-DeID+, which integrates a refined three-tier PII taxonomy with a boundary-aware tokenizer. The tokenizer enhances a pretrained BPE vocabulary by incorporating N-gram and boundary-scoring components to better align token and PII span boundaries. The proposed approach is evaluated using a tokenizer-independent word-segment-level protocol. On full-length judgments, the tokenizer consistently improves W. Strict Typed F1-score from 0.76-0.77 to 0.87-0.88 across four model sizes (150M, 395M, 750M, and 1B), mainly by increasing precision rather than model scale. With fine-tuning on 1,000 broader sensitive documents, the judgment-trained model further achieves 0.93 W. Strict Typed F1-score, indicating effective transfer beyond court judgments.
[NLP-92] okenEconomizer: Experience-Driven Black-Box Cost Control for Coding Agents
【速读】： 该论文旨在解决生成式 AI（Generative AI）在辅助软件工程中因前沿编码代理（coding agents）带来的显著令牌开销（token cost）问题。现有评估方法通常仅针对孤立任务进行，无法反映开发者在真实场景中持续构建、更新与维护大规模项目时的连续性需求，导致现有方法难以有效降低实际应用中的环境依赖型令牌消耗。其解决方案的关键在于提出 TokenEconomizer——一种外部控制器，通过动态决策三方面操作来实现序列任务下的令牌成本优化：（i）识别并注入必要的操作微技能（micro-skills），（ii）剔除持久无关的工具以减少冗余计算，（iii）在无效失败循环中适时干预。实验表明，在 Claude Code、Codex 与 GitHub Copilot CLI 搭配前沿模型的多种设置下，TokenEconomizer 在保持任务成功率相当的前提下，平均降低 18.4%–30.9% 的令牌开销；该效果在各类仓库和不同领域（如顺序化网页开发与安全审计）中均稳定显现，成本降幅达 6.4%–21.2%，且具有良好的泛化能力。本研究确立了适应性黑盒成本控制作为现实场景中降低编码代理令牌开销的重要方向。
链接: https://arxiv.org/abs/2610.22732
作者: Aaron J. Li,Weijia Xu,Chandan Singh,Zixiao Chen,Jeevana Priya Inala,Charlotte Siska,Acar Tamersoy,Bin Yu,Ion Stoica,Michel Galley,Jianfeng Gao
机构: University of California, Berkeley (加州大学伯克利分校); Microsoft Research (微软研究院); Meta (Meta)
类目: Computation and Language (cs.CL)
备注:
Abstract:Advances in frontier coding agents have sharply increased the token costs of AI-assisted software engineering. Their token efficiency is typically evaluated on isolated tasks, whereas real-world developers use them to continuously build, update, and maintain large-scale projects. This mismatch motivates us to evaluate agents over sequences of realistic coding tasks, where we find that agents exhibit recurring environment-dependent token costs that existing methods struggle to reduce. To minimize these costs, we introduce TokenEconomizer, an external controller that cuts environment-specific token costs over a sequence of tasks. These cuts come from determining (i) which operational micro-skills to inject, (ii) which persistently irrelevant tools to prune, and (iii) when to intervene in unproductive failure loops. Across Claude Code, Codex, and GitHub Copilot CLI paired with frontier models, TokenEconomizer reduces the average token cost by 18.4-30.9% with comparable task success. Savings hold in every repository under every tested agent-model setting and generalize to diverse domains such as sequential web development and security auditing, where costs fall by 6.4-21.2% depending on task structure. Our work establishes adaptive black-box cost control as a promising direction for reducing token costs of coding agents in realistic settings.
[NLP-93] CoLU: Confidence-Weighted Layerwise Unlearning for Large Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）在去记忆（unlearning）过程中如何有效遗忘特定信息的同时最大限度保留其他知识的问题，尤其针对现有方法在全模型微调时存在的性能退化与计算开销过大的问题。其核心挑战在于：不同网络层对目标信息的敏感度存在差异，统一的更新策略难以平衡遗忘与保留之间的权衡，且盲目全模型微调会带来高昂的计算成本。为此，论文提出了一种基于置信度加权的分层去记忆方法（Confidence-Weighted Layerwise Unlearning, CoLU），其关键创新在于通过设计一种新的置信度加权遗忘-保留正则化机制，仅对单个选定的Transformer层进行更新，从而在降低计算复杂度的同时实现更精细的控制。为进一步提升效率，论文还引入简短层筛选（Brief Layer Screening, BLS），利用新提出的归一化遗忘-保留（Normalized Forgetting-Retention, NFR）评分，在少量迭代和子集数据上快速评估各层表现，预选出前k个最优候选层，再进行完整分析。实验表明，BLS可将CoLU的运行时间降低60.2%至79.7%，同时保持相近性能；在TOFU、MUSE Books及News等数据集上，CoLU在遗忘效果与保留能力方面均显著优于当前主流方法（如NPO、SimNPO、SatImp、FOM-UL等），并在多数设置下创下新的性能纪录，同时平均显存占用较全模型方法降低77.4%至79.5%。
链接: https://arxiv.org/abs/2610.22721
作者: Jingkai Guo,Ziyi Zhao,Chaitali Chakrabarti,Deliang Fan
机构: 未知
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Large language model (LLM) unlearning aims to forget certain information while preserving others, without retraining from scratch. SoTA methods such as NPO, SimNPO, and SatImp balance forgetting and retain through whole model fine-tuning, while FOM-UL explores selective layer updates. Prior works show that suppressing target information can damage retained knowledge. This trade-off can vary across layers even under the same unlearning loss. In this work, we propose Confidence-Weighted Layerwise Unlearning (CoLU), which combines a novel confidence-weighted forgetting and retain regularization while updating a single selected transformer layer, rather than the whole model for unlearning. To reduce CoLU compute cost, we further propose Brief Layer Screening (BLS), which uses a newly defined normalized forgetting-retention (NFR) score to first rank layers through a brief unlearning trials using a subset of forgetting dataset and fewer epochs. BLS pre-selects top-k ranked layer candidates, which will then be fully profiled using full data analysis. BLS is proven to reduce the runtime of CoLU by 60.2-79.7% with almost the same performance. On TOFU dataset, CoLU improves both forgetting and retention over NPO, SatImp, and GA-based FOM-UL in all six TOFU settings, and over SimNPO in five of the six settings. On MUSE Books and News datasets, CoLU achieves the highest retained KnowMem and the PrivLeak closest to zero among the compared SOTA unlearning methods. CoLU presents a new record performance in both unlearning and retain in most settings up to date. Moreover, across the measured settings, average CoLU GPU memory usage is 77.4% to 79.5% lower than full-model unlearning.
[NLP-94] ARBOR: Locating Reasoning Failures in Long Chains of Thought through Reasoning State Tracking
【速读】： 该论文旨在解决大模型在复杂推理过程中存在的推理失败问题，特别是已有研究中识别出的过度思考（overthinking）、思考不足（underthinking）和无支持推断等失效模式。传统方法依赖局部信号（如相邻步骤的文本相似性、切换标记或大模型对步骤正确性的判断）进行故障检测，但这些方法未能充分考虑推理过程中的状态演化机制。本文提出一种名为ARBOR的推理失败归因框架，其核心创新在于将推理轨迹建模为由各步骤创建、验证、修剪或回溯推理状态所构成的树状结构，通过全局视角追踪状态演变路径，从而将每个失败精准归因于某一步骤与其前序状态之间的关系。该方法将原有三类失效模式细化为具有不同作用机制与代价的子类型，并揭示了在某个步骤之前由ARBOR维护的树状状态能够有效区分真正发生失败的步骤与处于风险但未实际失败的步骤。此外，论文探讨了ARBOR在推理阶段及训练数据层面与现有缓解策略的集成方式，显著提升了对推理缺陷的可解释性与可控性。代码、数据与交互式示例已公开。
链接: https://arxiv.org/abs/2610.22717
作者: Huihan Li,Siyuan Wang,Naga Vamsi Ramana Dinavahi,Xiaohui Liu,Xiang Ren
机构: University of Southern California(南加州大学); The Chinese University of Hong Kong(香港中文大学); New York University(纽约大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Large reasoning models solve complex problems through long chains of thought, in which prior work has documented failure modes such as overthinking, underthinking, and unsupported deductions. Existing methods detect these failures from local signals: textual similarity to nearby steps, switch tokens, or an LLM’s judgment of a step’s correctness. However, what a step contributes to the solution depends on the state that earlier steps have established, so identifying failures requires a global view that tracks that state across the trajectory. Therefore, we propose ARBOR, a reasoning-failure attribution framework that represents each trajectory as a tree of reasoning states, built as each step creates, verifies, prunes, or revisits a state, and attributes each failure to a relation between a step and the state preceding it. Using ARBOR, we refine the three failure modes into distinct subtypes that harm reasoning through different mechanisms and at different costs. We find that the tree state ARBOR maintains before a step separates failure steps from the at-risk steps at which the failure did not occur. We discuss how ARBOR integrates with existing mitigations at inference time and in training data. Code, data and interactive examples are available at this https URL.
[NLP-95] First Passage Attention
【速读】： 该论文旨在解决传统基于softmax注意力机制中遗忘门（forget gate）机制在上下文管理上的局限性：现有方法依赖于对键-查询路径上衰减贡献的单调累加，导致无法选择性地清除中间上下文信息，同时保留较早但结构上重要的标记。为此，论文提出首次通过注意力（First Passage Attention, FPA），其核心创新在于将路径累积机制替换为路径极值计算——通过将键的logit值按其下方学习到的标量场在路径跨度内的最低谷深度进行比例下调，实现非单调的、基于“首次穿透”行为的注意力调控。该机制的关键在于：当标量场与键层级遵循共享的学习随机游走时，未受惩罚的存活键自然形成一个栈结构，从而支持对历史信息的有序管理。研究表明，一个固定维度的FPA头即可实现任意由后续下探决定的键子集，而传统带位置偏置或逐标记遗忘门的softmax注意力头则需线性增长的头维才能达到类似效果。在合成实验中，单层FPA头能稳定收敛至目标键集合，而对比的softmax和FoX基线在所有随机种子下均失败；在1.1亿参数语言模型上的扩展验证表明，FPA在困惑度上与FoX相当，并在训练长度8倍的情况下唯一实现了完美的“密钥检索”（passkey retrieval）性能，且其键-值缓存压缩至约25%仍保持高效。
链接: https://arxiv.org/abs/2610.22714
作者: Yuxuan Liu
机构: University of Bergen (卑尔根大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Augmenting softmax attention with forget gates typically relies on summing decay contributions along the key–query path. Because such accumulation is monotonic, these mechanisms cannot selectively erase intermediate context while preserving older, structurally relevant tokens. We propose First Passage Attention (FPA), which shifts from path summation to path extrema by lowering a key’s logit proportionally to the deepest dip of a learned scalar field below that key’s level over the intervening span. FPA encompasses the Forgetting Transformer (FoX) as a special case, and when the field and levels follow a shared learned walk, the alive keys with unpenalised logits naturally form a stack. We show that a constant-dimension FPA head realises any subset of keys left alive by subsequent dips, whereas softmax heads with positional biases or token-level forget gates require a head dimension linear in the number of keys. In synthetic experiments, one-layer FPA heads converge to these target sets while comparable softmax and FoX baselines fail on every seed. Scaling to a 110M-parameter language model, FPA matches FoX in perplexity and uniquely achieves perfect passkey retrieval at 8\times the training length, even with its key–value cache compressed to about 25% .
[NLP-96] GlobeBench: Evaluating Text-Based World Model Capabilities for Scalable Environment Simulations
【速读】： 该论文旨在解决生成式文本世界模型（text-based world models）缺乏标准化评估基准的问题，阻碍了模型间的比较、研究与优化。为此，作者构建了GlobeBench这一新型基准，涵盖5个独立领域和19种模拟技能的1,632个任务，用于评估模型在给定完整智能体轨迹的情况下，能否准确预测环境对特定动作的响应。其核心解决方案在于设计一种混合架构：结合确定性门控机制与语义判别器，通过将模型预测结果与基于实际执行的参考结果进行对比，实现对环境状态转移预测能力的量化评估。实验表明，尽管GPT-6 Astra在20个通用大语言模型（LLM）及专用文本世界模型中表现最优（任务成功率49.57%），但当动作影响范围扩大至依赖对象时，平均成功率骤降至5.1%，且631项任务未被任何模型成功解决，揭示出当前通用大模型在建模复杂环境动态方面的显著能力鸿沟。
链接: https://arxiv.org/abs/2610.22708
作者: Nicholas Saban,Charvi Bannur,Mariya Vasileva,Darshan Deshpande
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Code: this https URL ; Data: this https URL
Abstract:Text-based world models have recently shown potential to simulate complex RL environment state transitions for a diverse set of use cases. However, no standardized benchmark exists to compare, study and improve these models. Through this paper, we attempt to solve this problem by building GlobeBench, a novel benchmark with 1,632 tasks across five separate domains and nineteen simulation skills. The tasks check whether models can predict the environment’s response to a specific action given the full agent trajectory by understanding earlier state changes, following environment rules, and predicting the effects of actions on both related and unrelated objects. We compare the predicted environment response with an execution-derived reference, using a hybrid architecture with both deterministic gates and a semantic judge. Among 20 general-purpose LLMs and specialized text world models, GPT-6 Astra achieves the highest task success rate at 49.57%. We find that model accuracy decreases as actions affect a larger portion of the environment. Specifically, when the agent’s changes must propagate to dependent objects, task success drops to an average of 5.1% across all 20 models. Although model strengths differ across tasks, 631 tasks remain unsolved by any model, suggesting a significant gap in general LLM capabilities as a world model.
[NLP-97] Explore Separately Adopt Selectively: Execution-Guided Tool Use over Large Tool Libraries
【速读】： 该论文旨在解决大语言模型智能体（LLM agents）在执行任务过程中，从大规模工具库（如数千个工具）中动态发现并验证相关工具时所面临的挑战，尤其是在多个候选工具可满足同一需求的情况下，如何避免因共享执行轨迹导致的探索不充分或失败尝试对后续推理产生负面影响。其核心解决方案是提出BRIA（BRanch-Inspect-Adopt）框架，通过在独立分支中暂隔离地并行探索候选工具，每个分支独立检索并执行对应工具以生成执行证据，再由任务代理评估这些证据并仅采纳有效结果进入主执行轨迹。该机制有效避免了不同候选路径间的干扰，提升了工具选择的准确性与任务执行的鲁棒性。BRIA还支持在任务执行中动态引入新工具以应对能力需求变化。实验表明，该方法在PlanBench-XL和MCP-Atlas基准上均优于或持平于最强基线，在MCP-Atlas上达到89.77的平均主张得分，相较MCP-Zero提升14.37个百分点，并位于任务性能与输入令牌成本的帕累托前沿，展现出卓越的效率与效果平衡。
链接: https://arxiv.org/abs/2610.22688
作者: Anna Mazhar,Seiji Maekawa,Nikita Bhutani
机构: Cornell University (康奈尔大学); Megagon Labs
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model agents must discover relevant tools from large libraries, e.g., thousands of tools, as new needs arise during task execution. When several candidates could satisfy the same need, execution may be required to identify the relevant one. Investigating candidates in a shared trajectory can leave alternatives unexplored or allow failed attempts to influence subsequent reasoning. To address this, we present BRIA (BRanch-Inspect-Adopt), a simple yet effective agent that temporarily isolates candidate investigations in separate branches. Each branch retrieves and executes tools to investigate its candidate; a task agent inspects the resulting execution evidence and adopts only useful results into its continuing trajectory. BRIA retrieves additional tools as new capability needs arise during task execution. We evaluate BRIA on PlanBench-XL and MCP-Atlas using a shared pool of 1897 tools, three language models, and static and adaptive retrieval baselines. Across all settings, BRIA matches or exceeds the strongest retrieval baseline. On MCP-Atlas, it achieves an average claim score of 89.77, compared with 75.40 for MCP-Zero, a gain of 14.37 percentage points. BRIA also lies on the Pareto frontier for task performance and input-token cost.
[NLP-98] Multimodal Filesystem Memory For Long-Horizon Embodied Agents
【速读】： 该论文旨在解决长时程智能体（long-horizon agents）在持续交互过程中如何有效管理记忆以实现经验复用的问题。传统方法要么保留全部历史导致检索困难，要么过度压缩信息而丢失关键的视觉与空间细节。其解决方案的核心是提出一种多模态文件系统记忆架构（Multimodal Filesystem Memory, MFM），通过并行维护四种互补型记忆：情景记忆（episodic memory）记录观测、动作与结果；空间记忆（spatial memory）将经验关联至具体位置；语义与过程性记忆（semantic and procedural memory）构建通用知识与可复用行为策略；工作记忆（working memory）维持当前计划。MFM利用可搜索的文本记录对齐不同表征与底层多模态事件，使智能体能基于上下文线索检索过往经验，并通过回溯原始观测恢复文本描述中缺失的细节。通过整合与提炼机制，智能体可将经验转化为高层知识与操作流程，同时保留对原始事件的引用。在长时程的Minecraft与计算机使用任务中，MFM显著提升性能：基于GPT-6-Astra模型，在“从零挖掘钻石”任务中成功完成，于“末影龙”任务中达成16/24里程碑，优于最佳基线的12.7；在CUABench-KiCad任务中平均奖励达0.440，显著高于基线的0.400及无持久记忆情形下的0.267。
链接: https://arxiv.org/abs/2610.22686
作者: Chloe Su,Jiahan Li,Zhenting Qi,Chaoran Cheng,Haoyu Chen,Sejoon Jun,Weirui Ye,Yilun Du
机构: Harvard; Kempner Institute; NYU; UIUC; NEU; MIT
类目: Computation and Language (cs.CL)
备注:
Abstract:Agents operating over long horizons need an effective memory that allows them to reuse their experience throughout long interactions. If the entire history of an interaction is retained, it becomes difficult to find the relevant experience, while compressing it into summaries may result in the loss of visual and spatial details that are required later on. We present \textbfMFM, a multimodal filesystem memory architecture that keeps complementary types of agent memory. \textbfEpisodic memory stores the observations, actions, and outcomes, and \textbfspatial memory associates experiences with locations. \textbfSemantic and procedural memory build up general knowledge and reusable methods of acting, and \textbfworking memory holds the current plans. MFM makes use of searchable textual records to connect these different representations to the underlying multimodal episodes. The agents are able to find previous experiences by means of contextual cues and to examine the original observations in order to recover details that are missing from their textual descriptions. Through consolidation, the agents convert experience into higher-level knowledge and procedures while still keeping references to the original episodes. In the case of long-horizon Minecraft and computer-use tasks, MFM leads to improved performance with three base models. With GPT-6-Astra, MFM completes \emphMine Diamond from Scratch and reaches 16 out of 24 milestones on the \emphEnder Dragon task, as against 12.7 for the best baseline. On CUABench-KiCad it attains a mean reward of 0.440, as compared to 0.400 for the best baseline and 0.267 in the case of no persistent memory.
[NLP-99] Not All Graphs Are Maps: A Task-Agnostic Exploration Framework for Mobile GUI Navigation AACL
【速读】： 该论文旨在解决现代移动图形用户界面（GUI）智能体在执行任务时缺乏持久的、应用特定的导航知识这一核心问题，即无法有效记忆功能位置、页面归属关系以及跨次访问中等价屏幕的映射。其解决方案的关键在于提出一种任务无关的图导航框架TAG-Nav，通过为现有GUI智能体后端引入可复用的目的地记忆机制，将离线探索过程提炼为紧凑的结构化记录——包括应用功能、目标页面与路径节点（waypoint routes），从而在未见目标任务指令前完成导航知识的预构建。当未来任务的目标处于已覆盖范围内时，TAG-Nav可直接检索目的地并提供路径引导，无需依赖特定任务的示范数据。该设计实现了导航与任务执行的解耦：多数任务共享相同的前置导航阶段（如定位正确页面），而最终操作则各不相同。为消除下游执行行为对导航评估的干扰，作者进一步提出DestBench——一个仅关注导航能力的基准协议，专门测试智能体能否准确抵达预操作目标屏幕。实验表明，基于图结构的记忆机制显著提升导航成功率：在相同条件下，UI-TARS-1.5-7B的严格导航成功率从52.4%提升至58.3%（使用自动化广度优先搜索生成的记忆），Qwen-VL-Max则从59.5%提升至71.4%。研究结果证实，目的地覆盖率与路径清洁度是决定图记忆价值的关键因素，使图质量成为可量化评估的变量，而非探索过程的默认优势。
链接: https://arxiv.org/abs/2610.22672
作者: Adamo Orsini,Seyed Mehdi Ayyoubzadeh,Sadman Rakib Pinon,Hannuo Zhang,Koosha Gholipour Baradari,Yuanhao Yu,Xinxin Zuo,Yang Wang
机构: Concordia University(康考迪亚大学); Mila – Quebec AI Institute(魁北克人工智能研究所); McMaster University(麦克马斯特大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Accepted to Findings of AACL-IJCNLP 2026
Abstract:Modern mobile GUI agents can perceive visible screens and execute local actions, but they still lack persistent app-specific navigation knowledge: where functions live, which pages contain them, and which screens are equivalent across repeated visits. We introduce TAG-Nav, a Task-Agnostic Graph Navigation framework that wraps existing GUI-agent backbones with reusable destination memory. Offline exploration is distilled into compact records of app functions, target pages, and waypoint routes before any evaluated instruction is seen. When a future task’s destination is covered, TAG-Nav retrieves the destination and provides route guidance without task-specific demonstrations. This separates mobile GUI destination navigation from full task completion: many tasks share the prefix of reaching the right page, while final actions may enter content, send data, make purchases, delete information, or change sensitive settings. To evaluate route finding without downstream execution confounds, we introduce DestBench, a navigation-only protocol that asks whether an agent reaches the correct pre-action destination screen. DestBench enables a controlled graph-memory study comparing no-memory agents with random, automated BFS, and clean graph memory. Graph-derived memory improves strict navigation success rate of UI-TARS-1.5-7B from 52.4% without graph memory to 58.3% with automated BFS memory, and improves Qwen-VL-Max success rate from 59.5% to 71.4% on the same conditions. Together, TAG-Nav and DestBench show that destination coverage and route cleanliness strongly shape graph-memory value, making graph quality a measured variable rather than an assumed benefit of exploration.
[NLP-100] ProPerAg ent: Toward Memory-Driven Proactive Assistance
【速读】： 该论文旨在解决生成式 AI 在主动辅助（proactive assistance）场景中面临的双重挑战：如何从长期、高噪声的个人记忆中准确识别干预时机与恰当内容。传统方法通常假设关键证据已提供于上下文，而本文研究更具挑战性的现实场景——系统需自主从长达一年、包含大量干扰信息的个人记忆（平均约618K tokens）中挖掘支持主动干预的证据，并判断何时触发响应。其核心解决方案在于提出 ProPerMem 基准测试框架，通过佩戴式视频与真实用户记忆配对，结合人工验证的触发窗口和联合评估内容与时间的评分机制，揭示现有模型在该任务上表现极差（最高仅15%），且内容与时机判断缺一不可。基于此，作者构建 ProPerAgent（主动个性化代理），采用图结构组织的记忆检索模块实现可插拔式记忆提取与过滤，结合轻量级强化学习（RL）阶段抑制过早或重复触发，无需额外训练即可将严格准确率从不足10%提升至37.5%（Qwen3.8-27B）和38%（GPT-5.3-Chat），经后处理训练进一步达到42.4%，验证了分离建模内容与时机决策的有效性。
链接: https://arxiv.org/abs/2610.22671
作者: Shenglai Zeng,Xiao Yang,Tiffany Ma,Xinyuan Zhang,Nanshu Wang,Deren Lei,Kai Sun,Xilun Chen,Tammy Stark,Xianhui Zhu,Yuetai Li,Rulin Shao,Tianxin Wei,Rakesh Wanga,Anuj Kumar,Jiliang Tang,Wen-tau Yih,Xin Luna Dong
机构: Michigan State University(密歇根州立大学); Meta Reality Labs(Meta虚拟现实实验室); Meta Superintelligence Labs(Meta超智能实验室); University of Washington(华盛顿大学); Meta Recommendation Systems(Meta推荐系统)
类目: Computation and Language (cs.CL)
备注:
Abstract:As assistants move from reactive answering toward proactive intervention, memory changes role: it no longer only supplies the answer to a query, it determines whether and when to speak at all. Prior work in this setting hands the system the relevant fact in context; we study the regime that makes the problem hard, where the evidence justifying an intervention must be mined from a long-horizon personal memory. We introduce ProPerMem, a benchmark for memory-driven proactive assistance (MPA), pairing egocentric video with year-long, distractor-rich user memories (~618K tokens on average), human-verified trigger windows, and a protocol that scores content and timing jointly. Every baseline we evaluate performs poorly (at most 15%), including frontier models given summarized or retrieval-augmented memory. A controlled study explains why: with oracle evidence delivered inside the trigger window GPT-5.3-Chat reaches 96%; letting it choose the timing drops it to 38%, and replacing oracle evidence with retrieval drops it to 15%. Content and timing are separately necessary, and missing either is catastrophic. Guided by this, we build ProPerAgent (Proactive Personalized Agent) with a plug-and-play harness that surfaces memory from a graph-organized store and gates it down to what is worth interrupting for, plus a lightweight RL stage that curbs early and repeated firing. Without any training, ProPerAgent lifts strict accuracy from below 10% to 37.5% on Qwen3.8-27B and 38% on GPT-5.3-Chat; RL post-training reaches 42.4%.
[NLP-101] Exploring The Empirical Privacy Landscape of Synthetic Data
【速读】： 该论文旨在解决在敏感数据（如合成医疗记录）上训练语言模型时，如何在模型能力（capability）与隐私泄露风险之间实现有效平衡的问题。其核心挑战在于：传统微调方法虽能提升模型性能，但易导致私有信息泄露；而差分隐私随机梯度下降（DP-SGD）虽可降低泄露风险，却显著损害模型能力。本文提出的关键解决方案是采用隐私感知重写（privacy-aware rewriting），即通过生成式方法对原始数据进行改写，在保留语义一致性的同时主动替换敏感信息，尤其引入“唯一性感知指令”以增强对私有细节的去标识化处理。实验表明，相较于简单的同义改写或直接微调，该方法在多种重写模型中均能更有效地减少隐私泄露，且更强大的重写模型通常带来更低的下游隐私风险。此外，研究构建了一个基于“信标（canary）”的评估框架，支持对不同方法在能力-隐私权衡上的系统性比较，为未来模型与方法的优化提供了可复现的基准。因此，隐私感知重写被证明是一种在能力与隐私保护之间取得更优平衡的有前景路径。
链接: https://arxiv.org/abs/2610.22659
作者: Zihao Zhao,Avi Schwarzschild,Casey Meehan,Katherine Lee,Zhili Feng,Dylan Sam
机构: Johns Hopkins University (约翰霍普金斯大学); OpenAI
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL)
备注:
Abstract:Methods for training language models on sensitive data differ in how they balance capability gains against the risk of leaking private information. In this paper, we present a controlled comparison of data-level and training-level interventions for navigating the capability-privacy tradeoff on synthetic medical records with canaries to probe for leakage. Direct finetuning performs best on capability metrics but leads to the most privacy leakage, while DP-SGD reduces leakage at substantial cost to capabilities. Generating synthetic data from these records offers a more favorable balance, particularly when guided by a uniqueness-aware instruction to replace private details. Relative to naive paraphrasing, privacy-aware rewriting improves privacy protection across a variety of different rewriting models. We also find that more capable rewriting models generally reduce the risks of downstream privacy leakage, although held-out fit does not improve monotonically and increased test-time compute yields inconsistent gains. Overall, these results highlight privacy-aware rewriting as a promising direction for achieving a more favorable trade-off between capabilities and empirical privacy compared to other training-level methods, and our evaluation framework supports comparisons of future models and methods.
[NLP-102] Secondhand Stress: How Stressful Disclosures Affect Psychosocial Wellbeing in Online Graduate Admissions Communities
【速读】： 该论文旨在解决在线社区中个体在应对压力性生活转变（如研究生申请）时，接触同侪困境披露可能带来的心理社会影响问题。研究基于压力交叉理论（stress crossover theory），探讨用户在申请准备阶段与同侪的痛苦表达互动后，其后续心理状态表达（如焦虑、抑郁、压力）及参与行为的变化，并检验这种影响是否通过后续社交互动进一步传递。解决方案的关键在于采用倾向评分分层（propensity-score stratification）与基线调整比较方法，对来自两个Reddit社区、覆盖三个招生周期的171,981篇帖子和1,130,413条评论进行分析，发现直接接触他人痛苦披露的用户表现出更高的焦虑、抑郁与压力表达水平，以及更频繁的评论活动和词汇重复性，且这一效应在不同招生周期中持续存在；但未发现下游互动者（即曾被暴露于此类内容的用户）之间存在显著的心理社会关联。研究揭示了直接接触痛苦内容与后续社交接触之间的关键区分，强调在线社区设计需在维护自我披露与同侪支持功能的同时，关注用户接触负面内容后的潜在心理负担。
链接: https://arxiv.org/abs/2610.22657
作者: Veda Duddu,Agam Goyal,Sabelle Huang,Ayush Umashankar Prasad,Mingquan Yi,Koustuv Saha
机构: Siebel School of Computing and Data Science (塞贝尔计算与数据科学学院); University of Illinois Urbana-Champaign (伊利诺伊大学厄本那-香槟分校)
类目: Computation and Language (cs.CL)
备注: Preprint: 22 pages, 15 tables
Abstract:Online communities provide spaces for individuals navigating stressful life transitions to disclose difficulties and seek peer support. Yet encountering others’ distress may also have psychosocial implications for community members. Drawing on stress crossover theory, we examine how engagement with peer distress during graduate admissions is associated with subsequent psychosocial expression and participation, and whether similar associations extend through downstream interactions. We analyze 171,981 posts and 1,130,413 comments from two Reddit communities across three admissions cycles, using propensity-score stratification and baseline-adjusted comparisons. Users who engaged with distress disclosures during application preparation subsequently exhibited higher classifier-estimated anxiety, depression, and stress expression, alongside greater commenting activity and lexical redundancy. These associations recurred across admissions cycles, while no significant psychosocial associations were detected among downstream respondents interacting with previously exposed users. Our findings highlight how greater distress-related expression can coexist with continued community participation, while distinguishing direct engagement with distress disclosures from subsequent social contact. We discuss implications for designing online communities that preserve opportunities for self-disclosure and peer support while remaining attentive to the experiences of those encountering distress-related content.
[NLP-103] Evidence-Gated Bilingual RAG for Reliable Consumer Banking Search
【速读】： 该论文旨在解决消费者在银行搜索中使用自然语言描述需求（如“用于旅行的卡片”）时，与银行网站上产品页面的关键词不匹配的问题，尤其在双语服务场景下（阿拉伯语与英语），同一意图可能有多种表达方式，导致检索准确率下降。其解决方案的关键在于提出一种基于证据门控的双语检索增强生成（RAG）框架，该框架完全运行于本地基础设施，通过结构化网页内容优化、混合稠密-词法检索、意图感知的证据选择、交叉编码器重排序以及三分类决策机制（答案 / 相关不匹配 / 无答案），有效提升了检索与生成的准确性。实验结果显示，该方法在210个标注URL的阿英双语查询上，将Top-1命中率从50.48%提升至79.05%，MRR@5从0.571提升至0.797；端到端评估中严格成功率达69.2%，且72.7%的生成答案被自动化忠实性检测工具判定为可信。尽管结果初步，但展示了该框架在处理自然语言查询与双语检索挑战方面的潜力。
链接: https://arxiv.org/abs/2610.22652
作者: Ahmed Mohamed Mahmoud Rayyan,Rana Mahmoud Mohammed,Aya Abdallah Abdelmonem,Mohamed Khaled Goda Othman,Rahatara Ferdousi
机构: Queen’s University (皇后大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Consumers often describe what they need rather than use the terminology found on a banking website. For example, a natural-language request (“a card for travel”) may not share the keywords of the relevant product page, and the problem is harder in bilingual services, where the same intent may be expressed in Arabic or English. This paper presents an evidence-gated bilingual retrieval-augmented generation (RAG) framework for consumer banking search that operates entirely on local infrastructure. The framework combines structured web-content refinement, hybrid dense-lexical retrieval, intent-aware evidence selection, cross-encoder reranking, and a three-outcome Answer / Related Mismatch / No Answer decision mechanism. On 210 URL-labeled Arabic and English queries, the proposed approach improves Top-1 hit rate from 50.48% to 79.05% and MRR@5 from 0.571 to 0.797 over a hybrid baseline. A separate 26-case end-to-end evaluation achieved 69.2% strict success, with 72.7% of generated answers classified as supported by an automated faithfulness judge. These results are preliminary: the retrieval queries are paraphrases of a small reference set, and the end-to-end test is small and uses an automated rather than human judge, so they do not establish human-verified answer correctness or generalizable performance.
[NLP-104] SAFETYEDIT: Sparse Internal Targets with Prompt-Wide Control for LLM Jailbreak Backdoors
【速读】： 该论文旨在解决模型编辑后门（model-editing backdoors）在触发条件下仍可能出现模型拒绝响应或生成看似合规但实质未回应请求的不理想行为的问题。现有方法多通过秩一更新少量模型权重以实现对肯定性响应前缀（affirmative response prefix）的控制，但此类方法在满足前缀目标后，模型仍可能表现出不一致的合规行为。为克服这一局限，论文提出一种基于稀疏内部表示的目标干预策略：聚焦于一个与拒绝相关的、位于深层前馈网络（MLP）中的稀疏坐标作为攻击目标。其核心解决方案是SAFETYEDIT，该方法通过优化单一拒绝相关坐标，并利用早期注意力层的键值缓存（KV-cache）标记机制，将触发信号传播至后续指令位置，从而实现对模型输出的精准控制。该方法不依赖固定的写入-读取对，而是以稀疏目标为核心，具备更强的适应性和泛化能力。实验表明，在四款开源大语言模型及三组数据集上，SAFETYEDIT在12次对比中均取得最高的触发保护准确率（Guard ASR）和全部12次最高的强拒绝得分（StrongREJECT），且全面的消融实验与行为审计显示其对非触发输入的行为和通用能力影响极小，验证了其在有效性与保真度之间更优的权衡。
链接: https://arxiv.org/abs/2610.22640
作者: Yutong Ke,Yansu Wang,Zeyu Zhang,Xihao Guo,Chongwen Zhao,Kaizhu Huang
机构: Duke Kunshan University (杜克-昆山大学); Xi’an Jiaotong-Liverpool University (西安交通大学利物浦大学); Wuhan University (武汉大学)
类目: Computation and Language (cs.CL)
备注: 35 pages,4 figures; includes appendix
Abstract:Model-editing backdoors aim to induce harmful responses under a trigger while preserving ordinary behavior on untriggered input. Recent model-editing approaches target affirmative response prefixes through rank-one updates to a small subset of model weights. Yet when this prefix-level target is satisfied, the model may still revert to refusal or produce a coherent, apparently compliant response that does not address the request. To overcome this limitation, existing internal-representation approaches shift focus from output tokens to dense, high-dimensional compliance directions in latent space. Existing representation-level approaches rely on preservation constraints to maintain conditionality, motivating us to ask whether a smaller internal target can offer a different effectiveness-preservation trade-off. Motivated by the comparable control capability of single-coordinate interventions, we investigate a sparse refusal-related MLP coordinate as this target. We propose SAFETYEDIT, which optimizes a single refusal-related coordinate and uses KV-cache marking in early attention layers to propagate a prefix-trigger signal across subsequent instruction positions. On some checkpoints a downstream MLP reader converts this signal into control of the coordinate, while on others the propagated signal alone suffices; the method is defined by the sparse target rather than by a fixed writer-reader pair. Across four open-weight LLMs and three dataset groups, SAFETYEDIT achieves the highest triggered Guard ASR in 11 of 12 comparisons and the highest mean StrongREJECT score in all 12, relative to the baselines evaluated for each model. Crucially, comprehensive ablation and behavioral audits confirm generally minimal changes in untriggered behavior and general capabilities in the tested configurations.
[NLP-105] Inductive Learning of Concept Embeddings from Cross-Linguistic Colexification Patterns with Textual Attributes
【速读】： 该论文旨在解决历史语言比较与类型学研究中长期依赖离散表示、未能充分挖掘嵌入表示潜力的问题。当前基于共词化（colexification）模式构建的概念嵌入虽能反映有意义的语义属性与关系，但受限于传统的归纳式（transductive）嵌入方法，仅能为训练数据中存在的概念生成嵌入，难以扩展至未见概念，导致可扩展性差。本文提出一种归纳式（inductive）概念嵌入学习框架，融合全量与部分共词化网络的结构特性以及短概念描述中的语义信息，实现对未见概念基于简短定义的嵌入表示。该方案的关键在于通过结合拓扑结构与文本语义信息，使模型具备泛化能力，从而在跨语言相似性或相关性度量上表现出良好相关性，并有效推广至训练数据之外的新概念。
链接: https://arxiv.org/abs/2610.22638
作者: Arne Rubehn
机构: University of Passau(帕绍大学); Passau, Germany(帕绍, 德国)
类目: Computation and Language (cs.CL)
备注:
Abstract:The advent of distributional semantics as a framework for computational semantics marked a paradigm shift in the nature of representations, by which meanings are represented as dense vectors commonly known as word embeddings. Computational approaches to historical or typological language comparison, on the other hand, still mostly rely on discrete representations, while the potential of embedding-based methods remains largely untapped in this context. Recent studies have proposed concept embeddings as a language-independent, onomasiological counterpart to word embeddings, based on networks of cross-linguistic colexification patterns. While the resulting concept embeddings reflect meaningful semantic properties and relations, they suffer from a lack of scalability, since they rely on transductive embedding approaches which only learn embeddings for nodes present in the training data. In this study, we present an inductive framework for learning concept embeddings, combining the structural properties of full and partial colexification networks with semantic information from short concept descriptions. This framework allows the embedding of unseen concepts based on a short definition. We show that the resulting concept embeddings correlate with various cross-linguistic measures of similarity or relatedness, while generalizing well to new concepts that were absent from the training data.
[NLP-106] Measuring Mechanistic Uncertainty in Large Language Models via Causal Robustness
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在关键应用领域中因幻觉（hallucination）问题导致的可靠性与可解释性不足，尤其关注如何实现可靠且可解释的不确定性量化。其核心解决方案是提出“机制不确定性”（mechanistic uncertainty）这一新概念，通过分析模型内部机制响应的稳定性来表征不确定性。关键创新在于引入“因果鲁棒性裕度”（causal robustness margin），即为使模型输出发生偏离所需施加的最小内部干预量，并利用雅可比透镜（Jacobian Lens, Jlens）概念的导向向量对其实现操作化。实验结果表明，错误或不确定的回答对应更小的因果鲁棒性裕度，表明其对内部扰动更为敏感；同时，在受控的知识注入实验中，经过正确训练后模型所学习到的事实表现出更大的裕度，验证了该指标具有基于机制的不确定性信号能力，能够有效补充传统的输出层面不确定性评估方法。
链接: https://arxiv.org/abs/2610.22636
作者: Aneesh Komanduri,Majid Afshar,Timothy A. Miller,Yanjun Gao
机构: University of Colorado Anschutz Medical Campus(科罗拉多大学安舒茨医学校区); University of Wisconsin-Madison(威斯康星大学麦迪逊分校); Harvard Medical School(哈佛医学院); Boston Children’s Hospital(波士顿儿童医院)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) are increasingly used in critical domains, where their susceptibility to hallucinations makes reliable and interpretable uncertainty quantification essential. We introduce the concept of mechanistic uncertainty, which characterizes uncertainty through the stability of the internal mechanisms supporting a model’s response. We propose the causal robustness margin, defined as the minimum internal intervention required to deviate the model’s output, and operationalize it using Jacobian Lens (Jlens) concept steering vectors. Across three open-source models and five question-answering datasets, we show incorrect and uncertain responses exhibit smaller margins, indicating greater sensitivity to internal interventions, and the causal robustness margin complements conventional output-based uncertainty methods. Finally, in a controlled knowledge injection experiment, correctly learned facts show increased margins after training, suggesting that causal robustness provides a mechanistically grounded signal of uncertainty.
[NLP-107] oMStories: Interpreting Theory-of-Mind in Synthetic Narratives
【速读】： 该论文旨在解决生成式 AI 在理解人类心理状态方面的局限性问题，具体聚焦于模型能否从行为表现中推断人物隐含的情绪、恐惧、动机或情感创伤等心理状态，而非依赖直接陈述。其核心挑战在于评估模型在处理“隐性表达”（showing）与“显性表达”（telling）之间存在冲突时的推理能力。解决方案的关键在于构建了一个名为“心智理论故事”（Theory-of-Mind Stories, ToMStories）的合成数据集，包含约6000个短篇叙事片段，每个片段均配有针对角色心理状态的多选题或开放性问题，涵盖情绪、恐惧、动机及情感创伤等维度。通过对13个参数量从1B到72B不等的开源模型进行评估，研究发现最大规模的模型在多选题上可达到与人类相当的水平，但在开放性任务中仍显著落后。进一步分析揭示，模型在解释和生成隐性心理状态时倾向于做出简化假设，表明尽管合成数据有助于揭示模型的推理机制，但可能高估了其对人类心理复杂性的理解能力。
链接: https://arxiv.org/abs/2610.22632
作者: Melanie Subbiah,Chia-En Jacklyn Tsai,Kathleen McKeown
机构: Columbia University(哥伦比亚大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Understanding people often requires inferring what they feel or want from how they behave rather than from what they state outright. We ask whether language models can make these inferences, and how they do so. We introduce Theory-of-Mind Stories (ToMStories), an evaluation dataset of ~6,000 short synthetically generated narrative vignettes, each paired with a multiple-choice or open-ended question about a character’s emotion, fear, motivation, or emotional wound. These psychological states are conveyed implicitly (showing), explicitly (telling), or with conflict between dialogue and description. Evaluating thirteen open-weight models (1B-72B parameters), we find that the largest models match humans on multiple choice but not open-ended questions. Our analyses point to simplifications models make in both interpretation and generation of implicit psychological states, indicating that synthetic data can help us understand how models perform interpretation, but likely overestimates how well models interpret humans.
[NLP-108] A Plausible Explanation Is Enough to Mislead Multimodal Judges
【速读】： 该论文旨在解决多模态大语言模型（Multimodal Large Language Models, MLLMs）作为评判工具时，其判断结果是否容易被看似合理但实际错误的解释所误导的问题。研究发现，即使答案本身错误，只要解释在语言上显得合理，仍会显著影响判别结果，导致模型对错误答案的接受率上升，这种现象被称为“可塑性捕获”（plausibility capture）。解决方案的关键在于揭示：尽管解释内容与视觉事实无关，但其表面合理性足以引发模型决策标准的系统性偏移——即更倾向于接受答案，而非基于真实视觉证据进行判断。通过信号检测理论分析，研究发现这一偏差主要源于决策阈值的下移，而非感知辨别能力的下降。此外，该现象在不同领域、解释生成器及评估协议中普遍存在，现有缓解措施效果有限且依赖于具体模型，提示当前以解释为依据的评价体系存在严重可靠性隐患，可能高估模型的真实性能。
链接: https://arxiv.org/abs/2610.22631
作者: Anglin Cai,Zhiqi Wang,Xiaozhi Yang,Yurun Lin,Dongwon Lee,Yuchen Yang
机构: Northwestern University(西北大学); The Pennsylvania State University(宾夕法尼亚州立大学); University of Pennsylvania(宾夕法尼亚大学); University of Washington(华盛顿大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Multimodal large language models (MLLMs) increasingly serve as judges in training and evaluation pipelines, often assessing answers together with explanations. Although AI judges are sensitive to textual factors, it remains unclear whether ordinary plausible explanations can sway judgments requiring visual verification. We introduce JUPE, a visual-question-answering (VQA)-style benchmark that fixes answers while varying explanations. Across 14 MLLM judges, ordinary explanations that sound plausible mislead judges, with effects comparable on average to deliberately persuasive arguments. Plain justifications and chain-of-thought rationales reduce rejection of incorrect answers by 12.7 and 13.8 percentage points, respectively. We call this failure plausibility capture. Using signal detection theory, we find a significant mean shift in decision criterion toward acceptance, without a clear change in mean discriminability. Explanations sway judgments across domains, explanation generators, and judging protocols, while tested mitigations offer limited, model-dependent protection. Such shifts can inflate performance estimates when judges evaluate answers with explanations. Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) Cite as: arXiv:2610.22631 [cs.CL] (or arXiv:2610.22631v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2610.22631 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-109] rajectory-Guided Steering: Mitigating Spurious Feature Bias in Clinical Classification
【速读】： 该论文旨在解决大型语言模型在临床分类任务中过度依赖虚假特征（spurious features）的问题，尤其是当临床文本中的性别标记被替换时，会导致与性别无关的诊断结果发生显著变化（最高达53.5%的病例出现误判）。现有推理阶段的修正方法通常依赖于围绕特定虚假特征构建的反事实数据，或表现出不可预测的行为。本文提出轨迹引导调控（Trajectory-Guided Steering, TGS），其核心创新在于：通过分析模型在不同层间对答案偏好变化的预测轨迹（prediction trajectory），判断每条文档是否需要修正以及修正方向，而无需依赖反事实数据或对虚假特征的先验知识。实验表明，在i2b2的16个临床分类任务上，TGS可修复17.7%的错误，仅破坏1.3%的正确预测，且在所有超参数设置下均保持净收益；而所有基线方法均为净损失或接近零。此外，该方法显著降低了模型对性别这一虚假特征的依赖，例如高血压和糖尿病在性别替换后的不一致性分别从53.5%、35.1%下降至38.0%、21.5%。若对所有文档统一应用修正，则正确阳性预测的误伤率从0.1%上升至6.4%，凸显了基于轨迹进行逐文档选择的关键作用。TGS还展现出良好的跨模型家族（三类）及跨临床与生物医学数据集的泛化能力。
链接: https://arxiv.org/abs/2610.22618
作者: Fardin Ahsan Sakib,Ziwei Zhu,Ozlem Uzuner
机构: George Mason University (乔治梅森大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models rely on spurious features in clinical classification: swapping a single gender token in clinical notes alters gender-irrelevant diagnoses in up to 53.5% of cases. Existing inference-time corrections either need counterfactual data built around the specific spurious feature or behave unpredictably. We observe that how a model’s preference between answers evolves across layers, its prediction trajectory, indicates for each document whether a correction is likely to help. We propose Trajectory-Guided Steering (TGS), which decides from the trajectory alone whether to correct each prediction and in which direction, with no counterfactual data and no knowledge of the spurious feature. On 16 clinical classification tasks from i2b2, TGS fixes 17.7% of errors while breaking only 1.3% of correct predictions and stays net-positive across all tested hyperparameter settings; all baselines are net-negative or near zero. The same corrections reduce reliance on the spurious feature: on the most affected diseases, gender-swap inconsistency falls sharply (hypertension 53.5% - 38.0%, diabetes 35.1% - 21.5%). Applying the correction to every document instead raises breaks on correct positive predictions from 0.1% to 6.4%, so per-document selection is essential. TGS also transfers to three model families and three other clinical and biomedical datasets.
[NLP-110] When Does Uncertainty Measure RAG Hallucination Risk?
【速读】： 该论文旨在解决检索增强生成（Retrieval-Augmented Generation, RAG）系统中幻觉（hallucination）风险的量化与检测问题，尤其关注在不同检索覆盖条件下模型生成事实性错误答案的可能性。其核心挑战在于：当检索到的证据不足或不充分时，模型可能依赖内部知识生成看似合理但无依据的陈述，导致幻觉风险升高，而传统不确定性度量在缺乏外部证据约束时难以有效识别此类错误。解决方案的关键在于提出一种条件化的不确定性评估框架，即仅在检索证据存在且对答案形成有效约束的前提下，不确定性指标才具有可解释性和判别力。研究通过在HotpotQA和SQuAD等数据集上测试四个指令微调模型发现，仅基于文本长度的分类器性能接近随机水平，而困惑度（perplexity）在针对特定虚假编辑的检测中达到0.81–0.91的AUROC，表现出较强判别能力。进一步分析表明，检索深度的影响揭示了不确定性得分变化的机制：仅使用最高相关性的单篇文档时，检测能力下降；加入第二篇相关文档后，检测性能恢复并提升。最终，结合不确定性与语义漂移（semantic-drift）得分的联合指标显著改善了校准效果，并在RAGTruth QA任务中持续保持高于随机的性能表现。结果支持将不确定性视为一种依赖于检索证据是否充分的、条件性的幻觉风险度量工具。
链接: https://arxiv.org/abs/2610.22615
作者: Rajeel Ahmad Ansari,Tautik Agrahari
机构: Independent Researcher(独立研究员); Delhi, India(德里, 印度); Bengaluru, India(班加罗尔, 印度)
类目: Computation and Language (cs.CL)
备注: 6 pages. Accepted for oral presentation at ICNLSP 2026
Abstract:Uncertainty can measure hallucination risk in retrieval-augmented generation (RAG), but only when retrieved evidence constrains the answer. When evidence is present, unsupported factual claims become less likely; when retrieval coverage is poor, the model can answer from internal knowledge and assign high probability to unsupported claims. We test this across four instruction-tuned models on HotpotQA and SQuAD. After matching faithful and hallucinated answers within word-count bins, length-only classifiers fall near chance, while perplexity reaches 0.81–0.91 AUROC on targeted unsupported edits. Varying retrieval depth shows why the score changes meaning: detection can weaken with only the top-ranked passage, then recover after adding a second passage. A combined uncertainty and semantic-drift score improves calibration, and measurements remain above chance on RAGTruth QA. Results support a conditional view of uncertainty as a useful measure of RAG hallucination risk.
[NLP-111] Can LLM s predict the outcomes of machine learning experiments?
【速读】： 该论文旨在解决当前前沿生成式人工智能（Generative AI）在实现递归自我改进（Recursive Self-Improvement, RSI）过程中面临的核心挑战：即大语言模型（Large Language Models, LLMs）在提出机器学习（Machine Learning, ML）实验方案后，缺乏对这些方案实际效果的准确预判能力。尽管LLMs已具备自主设计、选择并执行ML实验的能力，但其在面对长时间训练开销时，难以有效评估所提方案的价值，从而限制了高效迭代与自我优化的可行性。本文的关键解决方案在于构建一个包含288个由LLM提出的实验构想的数据集，并针对八个ML基准任务，通过五种前沿大模型（GPT-6 Astra、Claude Opus 5.5、Claude Fable 5.1、Kimi K3、GLM-5.3）对这些构想的潜在影响进行预测，再与实际训练结果对比，系统评估其预测性能。研究发现，尽管所有模型的预测表现显著优于随机猜测，但准确性仍有限——最优模型Claude Opus 5.5仅能解释约49%的结果方差（R² = 0.49），且普遍存在对改进幅度的系统性低估；不同任务间预测质量差异显著，而代码实现框架的选择对最终结果影响较小，未观察到模型自偏好现象。此外，更高推理深度和额外上下文虽带来小幅提升，但整体仍表明当前LLMs对ML实验成效的“直觉”尚处于有用但受限的状态。该研究为评估生成式AI在自动化科研中的真实能力提供了关键实证依据，并公开数据集以推动后续研究。
链接: https://arxiv.org/abs/2610.22599
作者: Ayush Bansal,Aldo Pareja,Sean Klein,Serge Didenko,Artem Lukoianov
机构: Autolab AI; Institute of Foundation Models; MIT
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 15 pages, 9 figures, 1 table. Dataset: this https URL
Abstract:Recent advances in AI make Recursive Self-Improvement (RSI) increasingly plausible. Large Language Models (LLMs) can already propose, select, and run Machine Learning (ML) experiments. However, the space of possible design choices in ML is vast. An LLM that proposes an experiment needs a reliable intuition of whether it is worth running given the timing constraints of long training runs. In this work, we measure such intuition in current frontier LLMs by evaluating how accurately they predict the impact of self-generated ML experiment ideas before these are put into training runs. To this end, we construct a dataset of 288 LLM-proposed ideas on eight MLE-bench tasks. We then prompt five frontier LLMs (GPT-6 Astra, Claude Opus 5.5, Claude Fable 5.1, Kimi K3, and GLM-5.3) to predict the impact of each idea and compare their predictions with the actual outcomes after training. We report four main findings. First, LLM predictions are substantially better than random chance but far from accurate: even the best model, Claude Opus 5.5, explains only half of the variance in outcomes (R^2 = 0.49). Second, we notice that all tested models systematically underestimate the magnitude of the changes. Third, prediction quality varies widely across tasks. Finally, the choice of the coding harness that actually implements the ideas has little effect on outcomes, and we find no evidence of self-preference. Higher reasoning settings and additional context lead to modest gains. Overall, our results show that current LLMs have useful but limited intuition about which ML ideas will work. We release our dataset to support further study.
[NLP-112] Broad and Efficient PII Detection with Unified Labels
【速读】： 该论文旨在解决个人身份信息（PII）检测模型在高流量场景下难以兼顾高效性与低误报率（FPR）的问题。现有检测模型普遍采用复杂架构，导致部署成本高昂，且多数模型仅在少量合成数据集上进行微调，限制了其在真实场景中的覆盖能力。此外，公开数据集间存在分类体系不一致、标注标准不统一等问题，即使经过类别映射，仍因标签分歧导致训练与评估信号不一致，影响模型性能。为此，本文提出一种基于编码器的141M参数检测模型，其在来自77个公开数据集的112万条样本上进行训练，并依据对12类结构化PII的详细上下文定义重新标注。在AI4Privacy和Nemotron-PII基准测试中，该模型在低误报率操作点下表现出色，在14项对比中有11项（按本文标签）和9项（按原始标签）取得最高召回率，显著提升了多语言披露信息及嵌入URL中的IP地址等边缘场景的覆盖能力。同时，该模型相较基线实现2-10倍更高的吞吐量，并减少94%-97%的峰值GPU内存占用，实现了高性能与低成本部署的平衡。其核心解决方案在于构建统一、细粒度的标注规范并整合大规模异构数据，从而提升模型泛化能力与效率。
链接: https://arxiv.org/abs/2610.22597
作者: Andy Hsu,Konstantin Berlin
机构: Cisco AI Defense(思科人工智能防御)
类目: Computation and Language (cs.CL); Cryptography and Security (cs.CR)
备注:
Abstract:Personally Identifiable Information (PII) detection models must efficiently find identifiers across diverse contexts while maintaining a low false-positive rate (FPR) on high-volume traffic. Recent detectors increasingly rely on complex architectures that raise deployment cost, yet many are fine-tuned on only a few synthetic datasets that limit their coverage. Combining public datasets is challenging because their taxonomies differ, and even after categories are mapped to a common schema, labels can disagree on what counts as a detection, producing inconsistent training and evaluation signals. To address these problems, we present a practical 141M-parameter encoder-based detector trained on 1.12 million examples from 77 public datasets, relabeled under a detailed definition of broad contextual detection for 12 structured PII categories. Evaluated on the AI4Privacy and Nemotron-PII benchmarks at a low-FPR operating point, our model achieves the highest recall against larger baselines in 11 of 14 comparisons under our labels and 9 of 14 under the original labels. Its largest recall gains include useful additional coverage, such as multilingual disclosures and IP addresses embedded in URLs. It achieves this with 2-10x higher throughput and 94-97% less peak GPU memory than the baselines, delivering leading detection performance at substantially lower cost.
[NLP-113] he Model Knows Better: Defending Medical Multi-Agent Systems Against Prompt Injection
【速读】： 该论文旨在解决医疗多智能体系统中因提示注入（prompt injection）攻击导致的临床决策偏差问题。此类攻击通过向模型输入中嵌入恶意文本，诱导生成式AI（Generative AI）做出错误判断，而现有防御机制在面对外观类似正常临床文本的注入内容时表现不佳。其核心解决方案在于提出一种无需训练的防御方法：依据临床实践，将患者的客观发现与临床判断（如诊断、建议和指令）进行分离，并独立验证每一项判断是否与患者的实际发现相一致，从而确保决策仅基于有依据的判断。该方法显著提升了系统对攻击的鲁棒性，在多个基准测试中将MedQA上的攻击成功率降低至4%-7%，同时保持未受攻击场景下的准确率，并对具备防御规避能力的自适应攻击仍具有效性。
链接: https://arxiv.org/abs/2610.22594
作者: Fardin Ahsan Sakib,Ziwei Zhu,Ozlem Uzuner
机构: George Mason University (乔治梅森大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Medical multi-agent systems, in which LLM agents discuss a case, are increasingly studied for clinical decision support. Prompt injection, text that an attacker adds to a model’s input, can manipulate LLMs into wrong decisions. How it affects these systems, and how to protect them, remains unclear. Across two systems, two model families and three benchmarks, a single injected passage succeeds in up to 61% of cases, and collaboration does not consistently protect against it. Existing defenses help only partly: they miss injections that look like ordinary clinical text. Yet when asked to check the injected text against the patient’s findings, a model that followed the injection usually rejects it. We therefore propose a training-free defense that verifies the clinical judgments behind the system’s decision. Following clinical practice, it separates the patient’s findings from clinical judgments, such as diagnoses, recommendations and instructions, and checks each judgment against the findings alone, so the decision rests only on supported judgments. Our defense outperforms all baselines on every benchmark, lowering attack success on MedQA to 4-7% while maintaining accuracy on cases without an attack, and remains effective with a medical model and against defense-aware and adaptive attacks.
[NLP-114] SAG2: Embedding-Free Multi-Hop Retrieval with BM25 and JEV
【速读】： 该论文旨在解决多跳检索（multi-hop retrieval）在实际部署中面临的高成本与复杂基础设施依赖问题，特别是传统方法对大规模语料库预处理、昂贵的嵌入计算（embedding computation）及专用数据库服务器的依赖。其核心解决方案是提出SAG2——一种无嵌入（embedding-free）的检索架构，仅依赖单一SQLite数据库、BM25算法以及来自JEV（Joint Event-Entity Verification）的上下文证据决策机制。SAG2采用查询时关联（query-time association）原则，以原始文本段落构建词法索引（lexical index），替代原有的事件-实体和密集索引；通过选择的段落提取关键词进行后续检索，并利用新获取的证据动态指导下一步决策，从而将语义证据选择过程移至查询阶段。这一设计彻底消除了离线嵌入计算、基于大语言模型（LLM）的事件-实体抽取以及专用数据库服务的需求。实验表明，SAG2在MuSiQue、2WikiMultiHopQA和HotpotQA等数据集上均达到最高召回率（Recall@5和Recall@10），其中Recall@10分别达86.62%、97.43%和99.00%，并在包含51万文档的EnterpriseRAG-Bench语料库上实现81.10%的文档级召回率。此外，其索引速度极快（仅需2.36秒完成11,656个段落索引），生成数据库仅25.87 MiB，且无需任何模型调用。SAG2证明了高精度多跳证据检索可由轻量、快速构建的基础设施支持，为知识检索应用提供了实用可行的基础。
链接: https://arxiv.org/abs/2610.22588
作者: Yuchao Wu
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Strong multi-hop retrieval should be practical to deploy without expensive corpus preparation or complex database infrastructure. We present SAG2, an embedding-free retrieval architecture built around a single SQLite database, BM25, and contextual evidence decisions from JEV. SAG2 follows the query-time association principle of SAG (SQL-Retrieval Augmented Generation), while replacing its event-entity and dense indexes with a lexical index over original passages. Selected passages supply terms for further retrieval, and newly retrieved evidence guides subsequent decisions. This moves semantic evidence selection to query time, eliminating offline embedding computation, LLM-based event-entity extraction, and dedicated database servers. The resulting system combines lightweight preparation with strong retrieval accuracy: on 1,000-question subsets of MuSiQue, 2WikiMultiHopQA, and HotpotQA, SAG2 achieves the highest Recall@5 and Recall@10 among the compared methods, with Recall@10 of 86.62%, 97.43%, and 99.00%. It also reaches 81.10% document Recall@10 over the 511,962-document EnterpriseRAG-Bench corpus. Indexing all 11,656 MuSiQue passages takes 2.36 seconds, produces a 25.87 MiB database, and requires no model calls. SAG2 demonstrates that accurate multi-hop evidence retrieval can be supported by compact, rapidly prepared infrastructure, making it a practical foundation for knowledge retrieval applications.
[NLP-115] Mitigating Demographic Disparities in Explanation-Human Alignment via Rationale Regularization
【速读】： 该论文旨在解决仇恨言论检测中解释公平性（explanation fairness）的问题，即当前高分类性能模型在不同人口统计群体间生成的解释可解释性存在差异，导致解释结果对某些群体可能不具合理性或代表性。其核心解决方案是提出一种名为Plausibility-EFG的新度量指标，用于量化模型生成的词元归因（token attributions）与人类标注理由之间在不同群体间的最大解释合理性差距。研究基于HateXplain数据集，评估了BERT、RoBERTa及经过对抗训练的ADV-RoBERTa模型，并结合集成梯度（Integrated Gradients）与SHAP两种归因方法进行分析。结果表明，ADV-RoBERTa不仅提升了分类性能和解释与人类理由的一致性，还显著降低了不同群体间的解释偏差。更重要的是，研究发现仅减少预测层面的偏见并不能消除解释层面的不公平现象，说明预测公平性与解释公平性虽相关但本质不同。最终，研究提出Fair-REF框架，通过增强对抗正则化可进一步降低Plausibility-EFG，尽管会带来轻微的宏观F1分数下降，验证了在保证性能的同时提升解释公平性的可行性。
链接: https://arxiv.org/abs/2610.22586
作者: Suryansh Talukdar(1),Taha Sarwat(1),Jonathan Rusert(1) ((1) Purdue University Fort Wayne)
机构: Purdue University Fort Wayne(普渡大学福灵顿分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 19 pages, 10 figures, 10 tables
Abstract:We study explanation fairness in hate speech detection. High classification performance does not always mean that a model produces equally plausible explanations for different demographic groups. Using the HateXplain dataset, we introduce Plausibility-EFG, a metric that measures the largest difference in explanation plausibility across demographic groups by comparing model-generated token attributions with human-annotated rationales. We evaluate BERT, RoBERTa, and ADV-RoBERTa, an adversarially trained model, using Integrated Gradients and SHAP. Our results show that ADV-RoBERTa improves classification performance and alignment between explanations and human rationales while reducing explanation disparities across demographic groups. These improvements are observed with both attribution methods and remain consistent across training, validation, and test splits. We also show that reducing prediction-level bias alone does not eliminate disparities in explanations, suggesting that predictive fairness and explanation fairness are related but distinct objectives. Finally, our fairness-performance analysis shows that stronger adversarial regularization can further reduce Plausibility-EFG, although it may cause a small decrease in macro-F1. Overall, this work presents explanation fairness as an important objective in hate speech moderation and introduces Fair-REF, a framework designed to reduce disparities in model explanations across demographic groups.
[NLP-116] Repairing LLM -Generated Specifications with Deterministic Diagnosis
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在形式化自然语言推理问题时产生的语义错误问题，即生成的正式规范虽可运行但存在约束、定义域、目标函数或谓词等方面的错误，而传统基于LLM的诊断因缺乏形式化真值依据，难以准确识别此类错误，甚至导致11.8%原本正确的规范出现退化。其解决方案的关键在于提出DRAGO修复框架，通过引入确定性的、与形式系统相关的诊断机制，替代原有的非接地（ungrounded）批判性反馈，并仅在诊断出的错误权重严格降低且满足无参考接受判据（reference-free acceptance predicate）时才执行重写。实验结果表明，在涵盖逻辑推理（SMT-LIB2）、约束规划（MiniZinc）和规划（PDDL）五个基准上的测试中，DRAGO将错误退化率从11.8%显著降低至1.3%，同时相比零样本形式化提升平均答案准确率12.9个百分点，相比基于LLM诊断的修复方案提升8.7个百分点。此外，研究还揭示了Planetarium数据集中存在两种渲染缺陷，导致其地板瓷砖划分有50%（3,086个实例）与参考形式化不一致，并已发布修正后的划分版本。
链接: https://arxiv.org/abs/2610.22568
作者: Giovanni Vaccarino,Emanuele La Malfa,Dehai Min,Hanrong Zhang,Pier Luca Lanzi,Lu Cheng,Michael J. Wooldridge,Philip S. Yu,Sara Bernardini
机构: University of Oxford (牛津大学); University of Illinois Chicago (芝加哥伊利诺伊大学); Politecnico di Milano (米兰理工大学); Institute for Decentralized AI (去中心化人工智能研究所); The Pennsylvania State University (宾夕法尼亚州立大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Large language models (LLMs) are increasingly used as formalizers: given a reasoning problem expressed in natural language, they produce a formal specification that a symbolic solver can then solve, with guarantees relative to that specification. Many specifications are runnable yet semantically wrong, encoding incorrect constraints, domains, objectives, predicates, or goals, and solver-error feedback dries up once execution succeeds. Purely LLM-based diagnosis, lacking a formal ground truth, often fails to spot these errors, and its ungrounded critiques cause regressions in 11.8% of initially correct specifications in our experiments. We propose DRAGO, a repair framework that replaces this ungrounded critique with deterministic, formalism-specific diagnosis, and retains a rewrite only when the diagnosed error weight strictly decreases under a reference-free acceptance predicate. Across five benchmarks spanning logical reasoning (SMT-LIB2), constraint programming (MiniZinc), and planning (PDDL), DRAGO cuts regressions by nearly an order of magnitude, to 1.3%, while improving answer accuracy by 12.9 percentage points over zero-shot formalization and 8.7 percentage points over LLM-diagnostic repair, averaged across benchmarks. We also uncover two rendering defects in Planetarium that cause 50% (3,086 instances) of its floor-tile split to be inconsistent with the reference formalizations, and release a corrected split.
[NLP-117] Who Do Perspectivist Models Predict Well? Annotator Agreement Sparsity and Performance
【速读】： 该论文旨在解决视角感知型自然语言处理（Perspectivist NLP）标注任务中，算法在面对具有异质性标注行为的标注者时的性能波动问题。具体而言，研究关注那些在标注结果上与多数人存在显著分歧（即标注偏离度高）或标注数量较少的标注者对模型预测能力的影响。其解决方案的关键在于构建迄今为止规模最大的视角感知型NLP数据集集合，这些数据集均采用序数标注（ordinal annotations），从而支持对不同视角下人类评分模式的系统建模。通过在多个数据集上评估多种视角感知型算法，并建立个体标注者特异性（annotator idiosyncrasy）与算法性能之间的量化关系，研究发现：算法在倾向于与其他标注者一致的标注者上表现更优，而标注者的响应率（response rate）与算法性能之间并无统计显著的正相关性。这表明，标注一致性比标注数量更能决定算法的泛化能力，为提升视角感知模型的鲁棒性提供了关键依据。
链接: https://arxiv.org/abs/2610.22567
作者: London Lowmanstone IV,Vaagishwaran Sivakumar,Anwesha Samaddar,Akul Mundada,Harnoor Kaur,Maria Gini
机构: University of Minnesota, Twin Cities(明尼苏达大学双城分校)
类目: Computation and Language (cs.CL)
备注: 24 pages, 9 figures, 5 tables
Abstract:Successful completion of perspectivist natural language processing (NLP) annotation tasks requires algorithms to be aligned across diverse perspectives in order to predict how different humans will rate text. This work studies how the performance of such algorithms varies for annotators who behave differently than others, either by providing annotations that diverge from others’ responses or by providing fewer annotations altogether. To aid with characterizing model performance, we curate the largest collection to date of perspectivist NLP datasets with ordinal annotations. After evaluating multiple classes of perspectivist algorithms across each of the datasets, we model the relationship between individual annotator idiosyncrasy and algorithm performance. In general, we find that algorithms perform better on annotators who tend to agree with others, but that annotator response rate does not have a statistically significant positive correlation with algorithm performance.
[NLP-118] Cultural (In)Competence: Investigating Culture-Specific Stereotype Detection Capabilities of LLM s EMNLP2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在跨文化语境下进行特定文化刻板印象检测时存在的可靠性不足问题，尤其关注模型在非英语或非美国文化背景下的表现偏差。其核心挑战在于，现有模型往往缺乏对文化语境的充分理解与适应能力，导致对刻板印象的识别存在系统性偏差，如高估刻板印象的存在（过度预测）。解决方案的关键在于通过三个关键因素优化模型的文化敏感性：一是使用非英语提示语（prompting language），发现非英语提示有助于提升模型对文化语境的考量；二是采用多语言指令微调（instruction tuning language），结果表明额外的指令微调可有效降低假阳性率；三是调整微调数据集规模，发现训练数据量对检测效果的影响呈非单调关系，即存在最优区间而非简单线性提升。综上，该研究强调了文化知识可及性与多语言适配在实现精准、公平的刻板印象检测中的核心作用。
链接: https://arxiv.org/abs/2610.22545
作者: Maximilian Spliethöver,Jannes Lötfering,Dominik Bachmann,Alexander Espig,Willem Zuidema,Katrin Schulz,Henning Wachsmuth
机构: Leibniz University Hannover (汉诺威大学); Institute for Logic, Language and Computation, University of Amsterdam (阿姆斯特丹大学逻辑、语言与计算研究所)
类目: Computation and Language (cs.CL)
备注: Accepted to Findings of the Association for Computational Linguistics: EMNLP 2026
Abstract:Stereotypes are subjective constructs that are mediated by the cultural context in which they are viewed. Successfully modeling them thus also requires cultural competence. When using large language models (LLMs) for stereotype-related tasks, such as their detection, it is therefore crucial that models can access relevant cultural knowledge and adjust responses accordingly. Since LLMs have shown disparities in what cultural knowledge is available to them, their capacity regarding stereotypes may also be disparate across languages and cultural contexts (e.g., for non-English languages or for non-US contexts). In this paper, we systematically investigate the effectiveness of LLMs for culture-specific stereotype detection across three factors: (1) the prompting language, (2) the instruction tuning language, and (3) the fine-tuning dataset size. We conduct experiments for US, Dutch, and Chinese cultural contexts. Our results suggest that LLMs do not reliably account for cultural contexts and overpredict the presence of stereotypes. Non-English prompts seem to improve consideration of cultural context and thus lead to more effective predictions. Moreover, we find that additional instruction tuning reduces false-positive rates and that training data size moderates detection effectiveness non-monotonically.
[NLP-119] HeRE: A Novel Benchmark for Hebrew Relation Extraction via LLM -Guided Denoising of Knowledge Graph Alignments AACL
【速读】： 该论文旨在解决低资源语言在关系抽取（Relation Extraction, RE）任务中缺乏标准化评估基准的问题。现有方法依赖人工标注成本高昂，而基于翻译的方案则引入英语中心偏差且准确性不足。为此，论文提出一种可扩展、语言无关的关系抽取数据构建方法，通过去噪知识库与文本之间的自然但噪声较大的对齐关系来生成高质量数据。其核心在于利用自然语言推理（Natural Language Inference, NLI）任务定义有效的语义对齐：当输入文本蕴含某个关系三元组时，即视为有效对齐。研究系统评估了五种去噪策略，发现多语言模型在少样本大语言模型（LLM）设置下表现优于专用语言模型，而希伯来语特定编码器在NLI微调后仍具竞争力，并在无标签跨训练设置中表现最优。此外，实验表明以目标语言进行提示（prompting）的效果与英文提示相当甚至更优，验证了该流程在目标语言中的自然性。最佳开源模型Gemma-4-31B-it（F1=0.915）性能接近最佳API模型GPT-5.4（F1=0.921），并在完全保留的双标注测试集上达到F1=0.844。基于优化后的流程，研究发布了HeRE——首个面向希伯来语的关系抽取银标数据集，包含150万实例，填补了希伯来语标准评估基准的空白，同时为其他低资源语言提供可复现、语言无关的关系抽取数据构建范式。
链接: https://arxiv.org/abs/2610.22542
作者: Ron Keinan,Amir David Nissan Cohen,Reut Tsarfaty
机构: Bar-Ilan University (巴伊兰大学); Origin AI
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted to AACL-IJCNLP 2026
Abstract:While LLMs demonstrate strong capabilities in English Relation Extraction (RE) tasks, a significant gap remains in assessing their RE performance in low-resource languages, due to a lack of standardized benchmarks. Creating native, language-specific annotated RE benchmarks is expensive and requires human expertise, while relying on translation yields suboptimal, inaccurate datasets that introduce English-centric biases. To tackle this, we propose a scalable, language-agnostic methodology to create RE data by denoising naturally occurring yet noisy alignments of knowledge bases and corresponding texts. We define a proper alignment via a Natural Language Inference (NLI) task, where a relation triplet is entailed by an input text. We empirically assess five denoising strategies, and demonstrate that multilingual models outperform language-specific models in the few-shot LLM setting, while Hebrew-specific encoders remain competitive under NLI fine-tuning and are strongest in the label-free cross-training setting. We further find that prompting in the native language achieves performance on par with or slightly exceeding English prompts, confirming that the pipeline operates naturally in the target language. The best open-weights model, Gemma-4-31B-it (F1,=,0.915), matches the best API model, GPT-5.4 (0.921), and reaches F1,=,0.844 on a fully held-out, dual-annotated test set. Applying the optimized pipeline, we release HeRE: a novel, massive, 1.5-million-instance silver dataset for Hebrew RE, designed to support data-intensive experimentation and model fine-tuning. HeRE bridges a critical gap in standard Hebrew RE evaluation, while providing the NLP community with a reproducible, language-agnostic pipeline for RE dataset construction in other low-resource settings.
[NLP-120] Are Large Reasoning Models Good Everyday Reason ers? A Comparison with Non-Reasoning Reasoning Models on Logical Reasoning Tasks
【速读】： 该论文旨在解决当前生成式推理模型（Generative Reasoning Models, GRMs）在处理日常逻辑推理（everyday logical reasoning）时表现不佳的问题。尽管大推理模型（Large Reasoning Models, LRMs）在数学与科学类基准测试中表现出色，但其在面对偏离教科书形式逻辑、依赖语用和语言背景的日常推理任务时，性能显著下降。论文的关键解决方案在于构建一个基于语言学与语用推理现象的基准测试，系统评估推理模型与非推理模型在受控提示条件下的表现差异。研究发现：（1）当推理偏离形式逻辑规则时，推理模型性能普遍退化；（2）部分采用思维链（Chain-of-Thought）提示的非推理模型反而优于同类推理模型；（3）额外指令提示对推理行为的控制能力有限。这些结果表明，现有推理模型未能可靠建模日常推理模式，暴露出后训练阶段训练目标的局限性，亟需改进训练策略并拓展跨多样化推理形式的评估体系。
链接: https://arxiv.org/abs/2610.22538
作者: Kentaro Ozeki,Risako Ando,Takanobu Morishita,Hirohiko Abe,Koji Mineshima,Mitsuhiro Okada
机构: Keio University(早稻田大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Large reasoning models (LRMs) that leverage test-time computation achieve strong performance on mathematical and scientific benchmarks and are increasingly deployed in real-world applications. However, their ability to perform \textiteveryday logical reasoning, which often departs from textbook formal logic, remains underexplored. In this work, we systematically evaluate how well reasoning models align with everyday reasoning patterns by introducing a benchmark based on linguistic and pragmatic reasoning phenomena. We also compare reasoning models with non-reasoning models under controlled prompting conditions. We find that, in many cases, (1) reasoning models show degraded performance in everyday reasoning when such reasoning deviates from formal logical rules, (2) some non-reasoning models with chain-of-thought prompting outperform comparable reasoning models in everyday reasoning, and (3) prompting with additional instructions provides limited control over reasoning behavior. These findings suggest that current reasoning models do not reliably capture everyday reasoning and point to limitations in post-training, motivating improved training objectives and evaluation across diverse forms of reasoning.
[NLP-121] Pearl-Chat: Engineering a native Luganda language model from first principles in JAX and Flax NNX
【速读】： 该论文旨在解决自回归基础语言模型在低资源非洲语言（特别是黏着语系的班图语族语言）中面临的计算效率低下与形态学处理负担过重的问题。现有通用多语言分词器在班图语中平均词元密度超过3.2个/词，导致上下文效率下降和推理延迟增加。其解决方案的关键在于构建一个专为卢干达语（Luganda）设计的轻量级因果解码器模型——Pearl-Chat（约850万参数），采用从数学原理出发的JAX与Flax NNX框架实现，结合显式函数状态分解、基于Google Grain的确定性可恢复数据流以及SPMD张量并行策略。核心创新在于提出一种从头训练的字节级字节对编码（byte-level BPE）子词分词器（词汇表大小 $ V = 8,192 $），在卢干达语语料上实现了4.88字符/词元的压缩率与1.60词元/词的词元密度，相较LLaMA-3（128K）和Gemma（256K）提升两倍压缩效率。模型通过两阶段训练流程：第一阶段在跨领域平行文本、圣经、百科及新闻语料上预训练，第二阶段在510条对话数据上进行指令微调并使用仅助手损失掩码。经过4,000次优化迭代，验证交叉熵损失从9.4695（困惑度12,958.98）单调下降至0.1818（困惑度1.20），降幅达9.2877纳特。在三层次留出诊断基准测试中，模型取得0.9792 ROUGE-L与0.9405 BLEU得分，并保持了对班图语名词类别的语义一致性。研究还开源了全部权重、分词器及代码，并提供基于Python SSE的流式推理运行时。
链接: https://arxiv.org/abs/2610.22529
作者: Wesley Kambale,Roland Sankara
机构: 未知
类目: Computation and Language (cs.CL)
备注: 17 pages, 8 figures
Abstract:Autoregressive foundation language models predominantly target high-resource Indo-European languages, imposing computational and morphological penalties on under-represented African languages. Generic multilingual tokenizers routinely exceed fertility rates of 3.2 tokens per word on agglutinative Bantu languages, degrading context efficiency and inflating latency. We present Pearl-Chat, an 8,473,122-parameter ( \sim 8.5M) causal decoder-only transformer engineered for Luganda (Oluganda, ISO 639-2: lg), a low-resource Niger-Congo Bantu language spoken by over 10 million people in East Africa. Built from first mathematical principles in JAX and Flax NNX, Pearl-Chat uses explicit functional state decomposition (this http URL and this http URL), deterministic resumable data streaming via Google Grain, and SPMD tensor parallelism using this http URL. We introduce a byte-level byte-pair encoding (BPE) subword tokenizer trained from scratch on curated Luganda corpora ( V = 8,192 ), achieving 4.88 characters per token compression and 1.60 tokens per word fertility, a two-fold compression improvement over LLaMA-3 (128K) and Gemma (256K). The model is trained with a two-stage curriculum: Stage 1 pre-training on 59,501 sequences across parallel, biblical, encyclopedic, and journalistic corpora, and Stage 2 instruction fine-tuning on 510 dialogues with assistant-only loss masking. Across 4,000 optimization steps, validation cross-entropy loss drops monotonically from 9.4695 (perplexity: 12,958.98) to 0.1818 (perplexity: 1.20), a decrease of 9.2877 nats. On a three-tier held-out diagnostic benchmark, Pearl-Chat achieves 0.9792 ROUGE-L and 0.9405 BLEU while maintaining concordial agreement across Bantu noun classes. We provide a Python Server-Sent Events (SSE) streaming runtime and open-source all weights, tokenizers, and code.
[NLP-122] Literal Lock-In: How Test Language Shifts Literal Attraction in Typed Decision Models on Bengali Idioms
【速读】： 该论文旨在解决生成式 AI（Generative AI）之外的决策模型（decision models）在缺乏可解释性机制（如推理链或理由生成）情况下，其内部判断逻辑不可见且易受输入干扰的问题。这类非生成式评分器（non-generative scorers）作为路由、分类与安全护栏被广泛部署，但因其不生成文本，传统可解释性方法失效。研究的关键在于通过仅对输入进行干预——即在选项列表中插入特定成语的字面释义（literal gloss），来探测模型行为的脆弱性。实验发现，这一简单操作使所有模型在孟加拉语任务中的错误率下降65%至96%，表明模型对输入语义的敏感性远超预期；进一步分析显示，模型在孟加拉语上的表现劣化程度是英语的1.9至6.4倍，说明其“抗干扰”能力实际取决于测试语言而非模型本身架构；此外，校准性能普遍下降，开放模型甚至出现置信度反转现象，导致基于置信度的过滤策略比随机筛选更差。即便在语义无关的选项列表上，最佳校准模型仍报告0.82的平均置信度，揭示了模型内在信心与真实准确性的严重脱节。因此，该研究的核心突破在于：通过输入级干预揭示非生成式模型的深层缺陷，并证明其可靠性无法依赖于表面校准指标。
链接: https://arxiv.org/abs/2610.22526
作者: Adib Sakhawat,Syed Rifat Raiyan
机构: Islamic University of Technology, Dhaka, Bangladesh (伊斯兰科技大学, 达卡, 孟加拉国); Systems and Software Lab (SSL) (系统与软件实验室)
类目: Computation and Language (cs.CL)
备注: Under review, 33 pages, 12 figures, 22 tables
Abstract:Decision models, non-generative scorers that report a probability over caller-supplied options and emit no text, are increasingly deployed as routers, classifiers and guardrails. Because they generate nothing, rationales and chain-of-thought do not apply. We audit three: an encoder scorer, a decoder scorer and a closed API, entirely through interventions on their input, using expert-annotated Bengali idioms carrying a gold figurative and a gold literal gloss. Dropping an idiom’s own literal gloss into the option list draws every arm toward it, taking 65% to 96% of each arm’s remaining Bengali errors. A model-by-language matrix over one shared item set shows how far this depends on the test language: every arm is 1.9 to 6.4 times more captured in Bengali than in English on the same paired entries, the commercial model’s apparent resistance is a property of its test language, and capability does not order the failure. Calibration degrades in all three arms, and on the open arms confidence inverts, so gating on it is worse than gating at random. On a semantics-free option list, where accuracy is at chance, the best-calibrated arm still reports mean confidence 0.82. Code and data: this https URL.
[NLP-123] alkhder at AlexandriaX-2026: Context-Aware English-to-Dialectal Arabic Dialogue Translation with QLoRA-Fine-Tuned NileChat-3B
【速读】： 该论文旨在解决将英语对话片段准确翻译为十三种阿拉伯语方言的跨方言自然语言生成问题，核心挑战在于如何在有限数据条件下实现多方言统一建模并保持上下文连贯性。其解决方案的关键在于：基于NileChat-3B-Base模型，采用QLoRA（Quantized Low-Rank Adaptation）技术进行高效微调，仅训练一组共享的适配器权重（adapter weights）以支持所有方言，同时在输入中显式注入目标国家与方言、对话领域、参与者角色、说话人-受话人性别方向以及前序对话上下文等多维度信息，从而增强模型对特定方言语境的感知能力。推理阶段采用自回归方式，将模型自身先前输出作为后续输入的上下文，不依赖任何真实目标侧测试信息，实现了端到端的零样本外推能力。实验表明，在仅使用Alexandria平行语料的情况下，系统可在约1小时内完成训练，并在不到两小时内完成私有测试集的翻译，达到20.36的country-macro-averaged spBLEU和36.64的chrF++，在约束赛道与非约束赛道均表现稳定。
链接: https://arxiv.org/abs/2610.22525
作者: Hasan Alkhder,Maher Pasa,Maria Hanini,Amro Najjar,Ahmet Zengin
机构: Sakarya University (萨基亚大学); Altınbaş University (阿尔廷巴什大学); Imperial College London (帝国理工学院); Luxembourg Institute of Science and Technology (卢森堡科学与技术研究所)
类目: Computation and Language (cs.CL)
备注:
Abstract:This paper describes team Alkhder’s submission to Subtask 1 of the AlexandriaX-2026 shared task, which asks systems to translate English dialogue turns into thirteen Arabic dialects. We fine-tune NileChat-3B-Base with QLoRA on the Alexandria parallel dialogue corpus, training one set of adapter weights for all dialects. Each turn is translated conditioned on the target country and dialect, the domain, the participant roles, the speaker-addressee gender direction, and the preceding dialogue context. At inference, conversations are translated autoregressively and the model is fed its own previous outputs back as context without using any gold target-side test information. Under the constrained track, using only Alexandria data, it trains in about one hour and translates the private test set in under two hours. The system reaches a country-macro-averaged spBLEU of 20.36 and chrF++ of 36.64 on both tracks.
[NLP-124] Learning User Representations to Improve Simulation Fidelity
【速读】： 该论文旨在解决当前用户模拟器（user simulator）在模拟真实用户行为时存在的泛化能力不足问题，其核心瓶颈在于用户表征（user representation）的质量低下。现有模拟器难以有效捕捉真实用户群体的多样性，尤其对少数群体用户的代表性较差，导致训练出的智能助手在面对未充分覆盖的用户类型时表现不佳。本文提出的关键解决方案是引入“事后蒸馏”（hindsight distillation）机制，通过构建对话重建任务（conversation reconstruction task）来评估并优化用户表征：给定一段原始对话，生成用户表征后将其注入模拟器以生成仿真对话，并通过对比原始与仿真对话中的用户行为差异，获得事后反馈（hindsight feedback），进而指导表征生成器进行在线优化。该方法使表征生成器能够直接从原始对话中学习到更高质量的用户表征。基于此，作者提出了UserRepGen-4B模型，其在700段跨七大数据集的对话上验证表明，经事后反馈优化后的表征显著提升了模拟保真度，优于GPT-5.5生成的表征。此外，所学表征不仅在对话重建任务中表现优异，还能提升未见任务上的模拟准确性，更好地保留真实用户群体的多样性，并增强对不同用户个体的区分能力。
链接: https://arxiv.org/abs/2610.22518
作者: Shuhaib Mehri,Philippe Laban,Michel Galley,Dilek Tür,Jennifer Neville,Jianfeng Gao
机构: University of Illinois Urbana-Champaign(伊利诺伊大学厄本那-香槟分校); Microsoft Research(微软研究院)
类目: Computation and Language (cs.CL)
备注:
Abstract:User simulators enable interactive training and evaluation of AI assistants at scale. Yet current simulators struggle to reproduce the diverse behaviors of real users, leading to assistants that generalize poorly to underrepresented users. We identify user representations as an underexplored bottleneck in user simulation. User representations are natural-language descriptions that condition simulators to reproduce a particular user’s behavior. A representation’s quality becomes apparent in hindsight, through the behavior it induces in a simulator. To evaluate and optimize user representations, we formulate a conversation reconstruction task: given an original conversation, we generate a representation, condition a simulator on it, simulate a conversation, and then measure fidelity by comparing user behavior in the original and simulated conversations. This comparison also yields hindsight feedback, which identifies where the representation falls short and how it can be improved. We introduce hindsight distillation, an online training procedure that distills feedback-conditioned improvements into the representation generator, enabling it to generate improved representations directly from the original conversation. We present UserRepGen-4B, a representation generator trained with hindsight distillation. Across 700 conversations spanning seven datasets, hindsight feedback helps improve user representations at test time, and UserRepGen-4B generates representations that enable more faithful simulation than those generated by GPT-5.5. Beyond conversation reconstruction, learned representations improve simulation fidelity for unseen tasks, better preserve the population-level diversity of real users, and better distinguish users across conversations.
[NLP-125] alkhder at ImageEval 2026 Shared Tasks: Constrained Selection and Rotation-Averag ed Logit Scoring for Multimodal Hallucination Detection
【速读】： 该论文旨在解决多语言视觉-语言理解任务中从三组候选描述中准确识别唯一视觉有效陈述的问题，尤其针对英文与现代标准阿拉伯语（Modern Standard Arabic, MSA）双语场景。其核心挑战在于如何在不依赖任务特定微调的情况下，实现对图像内容的精准语义匹配，并缓解因答案位置分布不均带来的偏差。解决方案的关键在于采用一个70亿参数的预训练视觉-语言模型，在零样本（zero-shot）条件下直接进行推理，将任务建模为受约束的选择问题，并利用下一个词标记的对数概率（next-token log-probabilities）作为判断依据以选出最优答案。为消除候选答案在固定位置上的偏倚，系统引入拉丁方旋转（Latin-square rotation）机制，动态调整候选句的排列顺序，从而提升选择的公平性与鲁棒性。实验结果表明，该方法在英文和MSA测试集上分别获得0.0510和0.0960的对比一致度（CI）分数，相较于官方报告的30亿参数基线模型，在开发集上分别降低了83.7%和80.4%的CI值，显著提升了性能。
链接: https://arxiv.org/abs/2610.22515
作者: Maher Pasa,Maria Hanini,Amro Najjar,Hasan Alkhder
机构: Altınbaş University (土耳其阿尔廷巴什大学); Imperial College London (帝国理工学院); Luxembourg Institute of Science and Technology (卢森堡科学与技术研究所); Sakarya University (萨卡里亚大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:We describe our submission to Subtask~1b of ImageEval 2026. The task consists of selecting the single visually valid statement from three candidates paired with an image, and is in both English and Modern Standard Arabic (MSA). Our system uses a pre-trained 7B vision–language model in a zero-shot setting, without task-specific training or fine-tuning. We formulate the task as a constrained selection problem and choose the answer using next-token log-probabilities. To reduce answer-position bias, we apply a Latin-square rotation to the candidate statements. Our system achieves CI scores of 0.0510 on the English test set and 0.0960 on the MSA test set, reducing CI relative to the official 3B baselines reported on devtest by 83.7% and 80.4%, respectively.
[NLP-126] RealCLI: Toward a Unified Evaluation of Human-Agent Collaboration for CLI Agents
【速读】： 该论文旨在解决当前命令行接口（CLI）智能体在执行长周期任务时，现有评估基准无法有效衡量人机协作质量的问题。现有方法要么假设用户指令在整个执行过程中保持不变，要么仅关注任务结果而忽略运行时的交互过程。为此，本文提出REALCLI——一个统一评估任务执行效果与人机协作质量的配对式基准。其关键在于构建了一个包含五个维度、22个子维度及51种交互场景模板的综合性评估框架，涵盖任务完成度与协作质量的多维量化。通过两种评估模式实现差异化测试：REALCLI-EXEC提供完整任务规格以评估任务执行能力；REALCLI-COLLAB则引入脚本化用户，动态施加信息补充、修正、约束调整和授权决策，模拟真实协作情境。实验结果揭示前沿智能体存在显著能力短板，最佳配置在协作评分（J_collab）上仅达37.6%，表明任务有效性与协作质量之间存在显著脱节，尤其在可引导性（steerability）、信任校准（trust calibration）与安全边界管理（safety boundary management）方面表现薄弱。因此，REALCLI的核心贡献在于建立了一个可复现、严谨的评估体系，为推动协同型CLI智能体的发展提供了坚实基础。
链接: https://arxiv.org/abs/2610.22507
作者: Yunqi Liu,Peiyu Wang,Bingli Wang,Leo Luo,Xuanwu Song,Xudong Ma,Hongxin Li,Yaxin Du,Weicheng Gu,Fanglin Xu,Yi Xing,XiangLong Liu,Jujie He,Zhange Zhang,Huatong Song,Zhaoyi Li,Deyang Kong,Chuan Hao,Peihao Wu,Mingjie Tang,Ran Tao,Jian Yang,Wayne Xin Zhao,Bryan Dai
机构: Beihang University(北京航空航天大学); Renmin University(中国人民大学); IQuest Research(智谱研究院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Command-line interface (CLI) agents increasingly execute longer-horizon tasks, yet existing benchmarks either assume that user instructions remain fixed throughout execution or evaluate task outcomes without capturing runtime collaboration quality. In this paper, we present REALCLI, a paired benchmark that unifies the evaluation of task execution and human-agent collaboration quality on realistic CLI tasks. To comprehensively measure collaboration, REALCLI builds a five-dimensional framework, operationalized into 22 sub-dimensions and 51 interaction scenario templates. REALCLI provides two modes for paired comparison: REALCLI-EXEC supplies complete specifications upfront, whereas REALCLI-COLLAB introduces a scripted user that delivers progressive information, corrections, constraint revisions, and authorization decisions. Evaluations reveal profound capability gaps in frontier agents: the best configuration attains an average score of only 49.2%, and when evaluated by the joint collaboration score ( J_\mathrmcollab ), the peak performance reaches merely 37.6%. These results show that task effectiveness diverges sharply from collaboration quality, with models consistently struggling in steerability, trust calibration, and safety boundary management. REALCLI provides a rigorous, reproducible foundation for measuring and advancing collaborative CLI agents.
[NLP-127] From Parallels to Divergence: Language Models Learning of Action Nominal Constructions
【速读】： 该论文旨在探究不同语法构型如何影响语言模型对动作名词化结构（Action Nominal Constructions, ANC）及其相关构造的学习效果，并比较模型的学习偏好与自然语言中的类型学分布之间的异同。其解决方案的关键在于采用基于语法工程的合成语料库构建方法，该方法将来自自然语料库的共享语义输入与计算语法相结合，生成96种在句法语序、名词短语语序、配列类型、补足语系统及ANC策略上存在差异的目标语言。针对每种语言，均从零开始训练GPT-2-small模型，并通过覆盖34种语法现象的目标最小对进行评估。研究发现，模型在基本句法构型上表现出与自然语言一致的类型学特征，但在补足语系统和ANC策略方面出现偏离，且学习到一些在自然语言中罕见或未被记录的组合。这一部分一致性表明，自然语言的分布可能既受到通用学习压力的影响，也受到功能性需求的驱动，而这些因素尚未被当前模型学习实验完全捕捉。
链接: https://arxiv.org/abs/2610.22468
作者: Iskar Deng,Shane Steinert-Threlkeld
机构: University of Washington(华盛顿大学)
类目: Computation and Language (cs.CL)
备注: 41 pages, including appendices
Abstract:Action nominal constructions (ANCs) express events in nominal form, with their argument marking following patterns found in clauses or ordinary noun phrases. In this paper, we investigate how different grammatical configurations affect language models’ learning of ANCs and related constructions, and how these learning preferences compare with typological distributions in natural languages. We use a grammar-engineered method for synthetic corpus construction that combines shared semantic input from natural corpora with computational grammars. We construct 96 target languages varying in clause word order, noun-phrase word order, alignment, complement system, and ANC strategy. For each language, we train GPT-2-small models from scratch and evaluate them using targeted minimal pairs covering 34 grammatical phenomena. The results show typological parallels in basic clause configurations but divergences in complement systems and ANC strategies, with models also learning combinations rare or unattested in natural languages. This partial alignment suggests that natural language distributions may reflect both general learning pressures and functional demands not fully captured by model learning experiments.
[NLP-128] Progressive Misalignment Can Emerge from Benign Recursive Self-Distillation
【速读】： 该论文旨在解决生成式AI在递归自蒸馏（recursive self-distillation）训练过程中可能出现的安全性退化问题。其核心问题是：尽管初始模型是安全的，且在训练中仅使用自身生成的输出与非有害提示（cue）进行迭代优化，但经过多轮自蒸馏后，模型对多种有害查询的响应逐渐偏离安全对齐目标，表现出“渐进性错位”（Progressive Misalignment）现象。解决方案的关键在于揭示这一失败模式并非由恶意训练提示引发——即使采用良性提示或模拟善意用户反馈，该现象依然出现；通过梯度分析进一步证明，某些看似无害的输入也能诱发模型安全属性的逐步丧失。这表明当前自蒸馏范式在缺乏充分安全保障机制的情况下，可能隐含系统性风险，亟需对基于自身数据递归训练的AI系统的安全性展开深入研究。
链接: https://arxiv.org/abs/2610.22467
作者: Karim Abdel Sadek,Federico Di Gennaro,Fanny Yang
机构: University of California, Berkeley; ETH Zurich
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 9 pages
Abstract:Training models on their own outputs has attracted widespread attention as a possible route to advance AI capabilities. One prominent approach is self-distillation, where a model is trained to match the token-level predictions it would have produced when given access to additional context, such as a cue, feedback, or instruction. However, the safety properties of this training procedure are poorly understood. In this paper, we present a surprising safety failure of recursive self-distillation, where each self-distilled model becomes the teacher for the next round of training. We observe that when a safe model is recursively self-distilled on its own (safe) generations while receiving an additional non-harmful cue, it becomes increasingly misaligned across a diverse set of harmful queries; we call this Progressive Misalignment. We show this failure mode in both the Qwen3 and Phi-4 model families on SORRY-Bench and StrongREJECT. Progressive misalignment also occurs in a more realistic setup, where the fixed cue is replaced by feedback from a simulated user with benign intent. Using a gradient-based method, we are also able to find benign prompts that lead to progressive misalignment when applying recursive self-distillation, showing that harmful training prompts are not necessary for this failure mode to emerge. We hope our work encourages future investigations of the safety properties of AI systems recursively trained on their own data.
[NLP-129] High prevalence of antagonizing rhetoric in trending topics on Twitter/X
【速读】： 该论文旨在解决社交媒体上政治对立言论中“敌意化”（antagonization）与“批评性”（criticism）两种修辞方式混同的问题，即现有研究将政治负面情绪视为单一维度，未能区分其本质差异。其核心挑战在于：在政治分歧日益加剧的社交平台语境下，究竟是将对手视为需被击败的敌人（敌意化），还是作为可对话的合法对手（批评性），这一区分对民主话语质量具有关键影响。本文的关键解决方案是提出一种计算方法，通过自然语言处理技术识别文本中的这两种对立修辞类型，并定位其攻击对象。基于对超过1700万条德语推特数据的分析，研究发现约三分之一的推文包含敌意化言论，且趋势话题下的讨论显著比普通推文更具敌意化倾向；进一步结合用户政治倾向估计表明，右翼倾向用户更频繁使用敌意化修辞；同时，两类对立修辞均在政治阵营内部被高度放大，传播倍数可达无对立修辞内容的两倍；最后，通过分析被敌意化的目标，揭示出不同政治阵营构建“他者”的机制，其中德国选择党（AfD）和绿党分别成为对方阵营的核心敌意对象。
链接: https://arxiv.org/abs/2610.22455
作者: Armin Pournaki,Eckehard Olbrich
机构: 未知
类目: ocial and Information Networks (cs.SI); Computation and Language (cs.CL)
备注: 33 pages, 8 figures
Abstract:Disagreement between political opponents is a central element of democracy. This disagreement can be voiced in different ways. An opponent can be antagonized, i.e. portrayed as an enemy that needs to be overcome, or they can be criticized, i.e. treated as a legitimate adversary with whom one can engage. While the latter is a fundamental part of democratic exchange, the former can be detrimental to public discourse. Political disagreement increasingly takes place on social media, where negativity towards political out-groups has been found to drive engagement. However, existing research has largely treated such negativity as a single dimension, leaving open whether it is antagonization or criticism that circulates on these platforms. This work addresses this gap by introducing a computational approach to detect these two types of rhetoric in text and identify its targets. On a corpus of over 17 million tweets from trending topics in the German Twittersphere, the method reveals that about a third of all tweets contain antagonizing rhetoric. Comparing tweets from trending topics to a random sample shows that the discourse in trending topics is systematically more antagonizing than ordinary Twitter communication. Combining antagonization detection with a large-scale estimation of users’ political leaning, we find that right-leaning users employ antagonizing rhetoric significantly more than their left-leaning counterparts. Furthermore, both forms of oppositional rhetoric are strongly amplified within political camps, with multipliers retweeting them up to twice as often as tweets without oppositional rhetoric. Finally, patterns of antagonized targets reveal how different political camps construct their out-groups, with the right-wing populist party AfD and the Green party as the central antagonists of the opposing camps.
[NLP-130] Where Prefill Expert Reduction Hurts MoE Models: A Few Prompt Tokens Carry Most of the Short-Answer F1 Loss
【速读】： 该论文旨在解决在混合专家模型（MoE）预填充（prefill）阶段通过减少路由专家数量以降低服务成本时所导致的性能损失问题。其核心发现是：这种简化虽能降低成本，但会显著影响生成质量，尤其在短答案问答任务中表现为模型虽能正确识别答案，却在回答过程中出现短暂中断。关键问题是，这种性能下降主要集中在提示（prompt）的特定敏感位置，包括答案起始前的模板标记（如Qwen3-30B-A3B中的两个聊天模板令牌）或指令部分的初始若干令牌（如DeepSeek-V2-Lite中的前8–32个提示令牌），这些位置对答案置信度的贡献分别约占总损失的一半和五分之一。实验表明，即使恢复超专家（super experts）或大幅激活，也仅能挽回约四分之一的性能损失，远低于保持这些敏感位置完整路由的效果。进一步分析揭示，所有敏感布局均位于提示末端，且因答案边界始终对应单轮提示结束，故保持提示末端完整即可有效避免损失；相比之下，随机放置相同数量的全专家令牌则导致F1值下降约3分，显著劣于保留接口位置的策略。此外，该优化策略在长提示（11K token）下可使首个输出令牌的延迟降低约14%，而在短提示（2K token）下无明显改善。因此，解决方案的关键在于优先保障提示末端关键令牌的完整专家路由，而非依赖复杂的路由规则或激活修复机制。
链接: https://arxiv.org/abs/2610.22454
作者: Zhiyuan Wang
机构: 独立研究员(Independent Researcher)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 8 pages main text, 26 pages with appendices; supplementary material in ancillary files
Abstract:Running prompt tokens through fewer routed experts during prefill is a simple way to make mixture-of-experts (MoE) serving cheaper, yet it is unclear what this damages. We locate the damage with token-, layer- and attention-path interventions on short-answer question answering in four MoE families. At half the experts the F1 loss of three families is mostly answer form: the model still finds the answer but stops answering briefly. In Qwen3-30B-A3B this loss is made at the two chat-template tokens before the answer, and in DeepSeek-V2-Lite at the first 8-32 prompt tokens, where its instruction sits; by answer likelihood these positions account for about half and a fifth of the damage. Reduction changes the first token’s attention mass, norm and largest activation by less than 4%, and restoring super experts or massive activations recovers at most about a quarter of the loss, well below keeping the sensitive positions full. On real prompts with about 4% of tokens at full experts, placing them on the task interface keeps F1 within one point of the unmodified model, while placing them at random in the documents is about 3 points worse (two of five prespecified primary comparisons; the others were undecided or not run). Because the answer boundary always ends a single-turn prompt, and DeepSeek’s loss was no longer detectable once its instruction moved away from the start, the sensitive tokens lay at the prompt ends in every sensitive layout we found, so rules that keep the ends are not distinguished from rules that read application marks. Time to first token falls about 14% on 11K-token prompts and not at 2K tokens.
[NLP-131] Embeddings Are Overpriced? A Frozen 16 MB Vocabulary for Small and Ternary Language Models
【速读】： 该论文旨在解决小型语言模型中词汇表（vocabulary）占据模型参数绝大部分且难以压缩的问题，尤其是在量化后仍需保持高精度的嵌入层（embedding table）导致模型体积庞大、计算效率低下的瓶颈。其核心解决方案是用一个无训练参数的冻结式512位编码表替代传统的可学习嵌入表，该编码表通过预训练嵌入向量进行主成分投影与符号二值化一次性生成，完全不参与训练。实验表明，尽管该冻结表在可训练参数上增加了2.6倍，但其在罕见词上的表现显著优于传统可学习表（提升1.7至2.3 nats），并能以仅3%的代价支持10,000个新词的增量引入，而传统嵌入表在相同条件下性能下降达43%。此外，结合三值化BitNet b1.58结构，整个模型压缩至46 MB，可在笔记本CPU上实现615–740 tokens/秒的生成速度，内存占用仅为浮点模型的七分之一，同时在LAMBADA、PIQA等基准测试中达到或超越原浮点模型性能，验证了该方法在高效性、可扩展性与推理速度上的优越性。
链接: https://arxiv.org/abs/2610.22428
作者: Sai Kiran Bathula
机构: QLNI(QLNI), New South Wales(新南威尔士州), Australia(澳大利亚)
类目: Computation and Language (cs.CL)
备注: 9 pages, 2 figures, 6 tables. Code, data, checkpoints and per-question results: this https URL . Preprint DOI: https://doi.org/10.5281/zenodo.23382634
Abstract:In a small language model the vocabulary is the single largest part. In Gemma 3 270M the embedding table holds 168M of 268M parameters, and because quantisation methods leave the embedding in 16-bit, it is also most of the file once the rest of the model is compressed. We ask what that table actually buys. We replace it with one frozen table of 512-bit codes, 16 MB, with no trainable parameters, derived once from a pretrained embedding by principal-component projection and sign binarisation, and we train the same Gemma-shaped body from scratch on the same 2.05B tokens with and without it. The learned table buys 0.20 nats of validation loss for 2.6 times the trainable parameters, but on LAMBADA, HellaSwag and PIQA the frozen-table model is better or tied; only ARC-Easy favours the learned table. With a ternary BitNet b1.58 body the whole model is a 46 MB file, still ties the float model on LAMBADA and PIQA, and on a laptop CPU generates 615 to 740 tokens per second in 62 MB of memory, five to six times the speed of the float model’s 8-bit GGUF in this http URL at a seventh of the memory. The frozen table wins by 1.7 to 2.3 nats on rare words, accepts 10,000 new words after training with a 3% cost where the learned table collapses by 43%, and all of it is released with the weights, the data and every evaluation answer.
[NLP-132] PRIME: Towards Verifiable Evidence-Grounded Reasoning in LLM Agents
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）代理在执行合规性检查任务时因推理路径不明确或证据链不完整而导致的误判问题，具体表现为：代理可能在关键检查未通过的情况下仍接受报告，或在存在更强证据路径时却选择较弱结论。其解决方案的关键在于提出一种名为PRIME（Precommitted Rules for Inference from Minimal Evidence）的框架，该框架在检查前预先定义一系列可被验证的证据路径（evidence routes），以明确每种结论的合法支持方式。通过一个独立的程序记录所有检查结果，并依据预设规则与审计日志（ledger）严格校验结论合法性，确保仅输出经规则授权的结论。此外，系统能主动识别并引导代理执行可提升结论强度的补充检查，或自动执行这些检查。实验表明，在480个公开记录任务中，未经PRIME约束的代理在26.7%的任务中报告了无依据的结论，而PRIME实现了零误报；同时，通过提供检查建议和执行额外验证，平均得分从30.1提升至满分59.2，显著提升了推理的可靠性与结论强度。
链接: https://arxiv.org/abs/2610.22426
作者: Hamed Khosravi,Xiaoming Huo
机构: H. Milton Stewart School of Industrial and Systems Engineering (乔治亚理工学院工业与系统工程学院); Georgia Institute of Technology (乔治亚理工学院)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:LLM agents are used to check records against an organization’s requirements, for example whether a vulnerability report is ready for investigation. Requirements often allow several ways to be met, such as a severity rating with a curator-reviewed advisory or a published advisory link. Agents fail by accepting a report after a required check has failed, or by settling for a weaker conclusion without trying another way that could still succeed. We introduce PRIME (Precommitted Rules for Inference from Minimal Evidence). Before checking, it fixes evidence routes specifying how each conclusion may be justified. A separate program records each check result and refuses any conclusion the results do not justify. It also points the agent to checks that could justify a stronger conclusion, or runs them itself. PRIME reports only conclusions authorized by its rules and ledger. With correct requirements, exact checks at known locations, and sufficient budget at stopping, completion reaches the strongest supported conclusion. On 480 public-record tasks, an agent shown the requirements reported unjustified conclusions on 26.7% of tasks, while PRIME reported none. Its directions and added checks raised the average score for justified conclusions from 30.1, with refusal alone, to the maximum of 59.2 out of 100.
[NLP-133] Eliciting Specifications: When Does Interactive Clarification Help? NEURIPS2026
【速读】： 该论文旨在解决大语言模型（LLM）在将任务描述自动形式化为Lean定理证明器可验证的规范时，因输入描述存在不完整或模糊性而难以生成准确规范的问题。其核心挑战在于：自然语言任务描述常遗漏验证器关注的关键边缘情况，导致生成的正式规范质量受限。为此，论文提出一种基于参考知识的模拟开发者（Elicitation Agent, EA）机制作为解决方案的关键——通过在形式化前主动识别任务中的未决决策，以受控数量的澄清问题（bounded clarification questions）向模拟开发者提问，并利用回答结果重构更精确的自然语言规范，再交由LLM转化为Lean代码。实验在189个Verina Lean 4规格生成任务上验证了该方法，在跨模型配置（如Claude Sonnet 4.6 → GPT-5.6）下，尤其当允许最多10次提问（k=10）时，严格通过率（strict pass@1）最高提升至65.6%（+6.3个百分点），且多数条件下编译条件通过率亦有显著改善。研究进一步揭示后置条件（postconditions）是主要退化来源，表明未来自动形式化系统应将经澄清的自然语言规范视为中间产物，需协同优化编译输出的语义正确性与澄清信息向Lean代码的可靠映射。
链接: https://arxiv.org/abs/2610.22382
作者: Mowafak Allaham,Aditya Vempaty,Ashish Jagmohan,Prasenjit Dey,Ravi Kokku
机构: Emergence AI( Emergence AI)
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Accepted to NeurIPs 2026 Workshop on AI for Verifiable Coding
Abstract:LLMs can generate Lean specifications from task descriptions, but those descriptions are often underspecified or ambiguous and may omit verifier-relevant edge cases, yielding imperfect formal specifications. We study whether an elicitation agent (EA) can improve verifier-checked Lean specifications by seeking clarification from a reference-grounded simulated developer before formalization. The agent decomposes the task, identifies unresolved decisions, asks bounded clarification questions, and uses the answers to write a refined natural-language specification that an LLM formalizer then translates into Lean. On 189 Verina Lean 4 specification-generation tasks, we compare single-shot generation with elicitation under question budgets of k=3 and k=10, allowing at most three or ten EA questions per task, across same-model and cross-model configurations. The same-model Claude Sonnet 4.6 pipeline has negative strict pass@1 estimates at both budgets, showing elicitation is not a uniform accuracy booster. In contrast, the Sonnet 4.6-to-GPT-5.6 and Sonnet 4.6-to-Gemini 3.8 Flash pipelines have positive strict estimates. The largest strict result reaches 65.6%, a +6.3-point change (+12 tasks), in the Sonnet 4.6-to-GPT-5.6 pipeline. Six of eight conditions have positive compile-conditional estimates, with a maximum cross-model difference of +4.9 points among compiling outputs. Treatment estimates also differ across the basic and advanced splits; the largest advanced-task change occurs in the Sonnet 4.6-to-GPT-5.6 configuration at k=10, from 42.0% to 50.6%. Failure analysis identifies postconditions as the largest source of degradation. These findings suggest future autoformalization systems should treat elicited natural-language specifications as intermediate artifacts and optimize jointly for semantic correctness among compiling outputs and reliable translation of clarified information into Lean.
[NLP-134] Binding-Specific Relearning Savings for Test-Time Continual Learning NEURIPS2026
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在持续学习过程中，当模型对某一事实的检索难度增加时，究竟是知识丢失还是仅变得不可访问的问题。其核心关切在于：当前基于输出准确率的评估方法是否足以反映先前经验对后续学习过程的深层影响。解决方案的关键在于设计了一种受控实验范式，通过在Qwen2.5-1.5B模型中对比三种匹配且反平衡的学习历史——已正确学习目标事实（LEARNED）、仅暴露但未学习该事实（CONTROL）以及接收到同一槽位中竞争性有效答案（WRONG）——在无重放的持续微调流后，考察目标事实的再学习速度。结果表明，相较于CONTROL，LEARNED组在部分遗忘的检查点处表现出显著更快的再获取速度，而WRONG组与CONTROL无显著差异，但与LEARNED组的对比提供了槽位匹配的验证证据，尽管存在竞争干扰可能。经前期行为表现调整后，该优势仍保持正向。此外，在非配对从未见过的样本池中的二次描述性分析显示，先前学习使恢复所需优化器更新量减少约18%。这表明，当前的行为准确率可能低估了先前习得经验的持久影响，揭示出模型内部表征中存在超越表面准确性的隐含知识留存机制。
链接: https://arxiv.org/abs/2610.22381
作者: Zhe Zhou
机构: University of Washington(华盛顿大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: Accepted at the NeurIPS 2026 Workshop on Towards Test-Time Continual Learning Agents (TTCL). 15 pages, 2 figures. Code: this https URL
Abstract:When a fact becomes harder to elicit from a language model, has the model lost it, or has it merely become less accessible? Parameter-updating test-time continual learning agents are usually evaluated by current output accuracy; we ask whether that measurement captures how prior experience shapes later updates. We test whether learning a specific factual binding produces a later relearning advantage beyond familiarity with its entity, answer vocabulary, format, or query slot. In Qwen2.5-1.5B, we compare three matched, counterbalanced histories: LEARNED receives the correct binding, CONTROL receives matched exposure without it, and WRONG receives a competing relation-valid answer in the same slot. After an identical rehearsal-free continual fine-tuning stream, LEARNED reacquires the target faster than CONTROL at a prespecified checkpoint of partial forgetting. WRONG shows no positive advantage over CONTROL, while its comparison with LEARNED provides slot-matched corroboration, subject to possible competitor interference. The advantage remains positive after adjustment for measured pre-relearning behavior. In a secondary descriptive conversion using an unpaired never-seen pool, prior learning reduces the optimizer updates required for restoration by about 18%. Current behavioral accuracy can therefore underestimate the lasting effect of previously acquired experience. Code is available at this https URL
[NLP-135] Revealing the Impact of Visual Text Style on Attribute-based Descriptions Produced by Large Visual Language Models ICMR2026
【速读】： 该论文旨在解决生成式视觉语言模型（Large Visual Language Model, LVLM）在处理图像中文字时，其视觉样式（如字体、颜色、大小等）是否以及如何影响模型对所指概念的语义描述问题。尽管人类阅读时能够忽略文本样式而聚焦于语义，但研究表明，即使LVLM正确识别了图像中文本所代表的概念，其属性层面的描述仍会受到文本样式的显著影响。关键在于，研究揭示了文本样式信息存在非平凡的“风格泄露”（style leakage），即装饰性样式（如彩色手写体）与功能性样式（如黑色无衬线体）会分别诱导模型生成具有不同属性倾向的描述，从而导致语义推理偏差。这一发现凸显了在基于LVLM的多媒体系统中引入风格感知评估与风格干扰缓解机制的重要性，以提升模型输出的语义一致性与鲁棒性。
链接: https://arxiv.org/abs/2604.27553
作者: Xiaomeng Wang,Martha Larson,Zhengyu Zhao
机构: Radboud University (奈梅亨大学); Xi’an Jiaotong University (西安交通大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
备注: Accepted by ICMR 2026. Code is available at this https URL
Abstract:When the visual style of text is considered, a wide variety can be observed in font, color, and size. However, when a word is read, its meaning is independent of the style in which it has been written or rendered. In this paper, we investigate whether, and how, the style in which a word is visualized in an image impacts the description that a Large Visual Language Model (LVLM) provides for the concept to which that word refers. Specifically, we investigate how functional text styles (readability-oriented, e.g., black sans-serif) versus decorative styles (display-oriented, e.g., colored cursive/script) affect LVLMs’ descriptions of a concept in terms of the attributes of that concept. Our experiments study the situation in which the LVLM is able to correctly identify the concept referred to by a visual text, i.e., by a word or words rendered as an image, and in which the visual text style should not influence the attribute-based description that the LVLM produces. Our experimental results reveal that even when the concept is correctly identified, text style influences the model’s attribute-based descriptions of the concept. Our findings demonstrate non-trivial style leakage from text style into semantic inference and motivate style-aware evaluation and mitigation for LVLM-based multimedia systems.
[NLP-136] EmbodiedMidtrain: Bridging the Gap between Vision-Language Models and Vision-Language-Action Models via Mid-training
【速读】： 该论文旨在解决视觉-语言-动作模型（Vision-Language-Action Models, VLAs）在实际应用中性能受限的问题，其根源在于当前大多数VLAs基于未经领域适配的通用视觉-语言模型（Vision-Language Models, VLMs），而这些VLMs的数据分布与具身任务（embodied domain）存在显著差异。解决方案的关键在于提出一种名为EmbodiedMidtrain的中段训练数据构建方法：通过分析发现，VLA数据集中在与VLM整体分布相分离的紧凑区域，且不同数据源间对齐程度差异显著；为此，研究设计了一个轻量级可学习的邻近度估计器（proximity estimator），从大规模VLM数据池中筛选出最符合具身任务特性的样本，构建了包含100万条样本的EmbodiedMidtrain-1M数据集，并利用该数据集对VLM进行中段训练（mid-training），从而获得更优的初始化状态。实验表明，该方法在仿真机器人操作和真实世界任务中均显著提升下游VLAs性能，且增益在未参与数据选择的下游任务上依然有效，证明了该方法具备良好的泛化能力。进一步分析揭示，中段训练能从微调初期即带来性能增益并持续扩大，且所提数据引擎同时捕捉数据集级与样本级的对齐信号，偏好空间推理类任务，同时保持原始VLM数据的多样性。
链接: https://arxiv.org/abs/2604.20012
作者: Yiyang Du,Zhanqiu Guo,Haochen Zhang,Annabella Macaluso,Xin Ye,Liu Ren,Yonatan Bisk,Chenyan Xiong
机构: Carnegie Mellon University (卡内基梅隆大学); Bosch Research North America (博世北美研究院); Bosch Center for Artificial Intelligence (博世人工智能中心)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Robotics (cs.RO)
备注:
Abstract:Vision-Language-Action Models (VLAs) inherit their visual and linguistic capabilities from Vision-Language Models (VLMs), yet most VLAs are built from off-the-shelf VLMs that are not adapted to the embodied domain, limiting their downstream performance. In this work, we propose EmbodiedMidtrain to bridge the gap between VLMs and VLAs. We first characterize the data distribution gap between them, showing that VLA data occupy compact regions that are largely separated from the broader VLM distribution, while the degree of alignment varies substantially both across and within VLM data sources. Then, we build a mid-training data engine that leverages a lightweight learnable proximity estimator to select the most VLA-aligned candidates from a large VLM pool, and produce EmbodiedMidtrain-1M, a VLM dataset that we further use to mid-train the VLM before downstream VLA fine-tuning. Experiments on both simulated robot manipulation benchmarks and real-world tasks show consistent VLA performance gains over off-the-shelf checkpoints across VLM backbones and VLA action heads. These gains extend to downstream tasks not used to guide data selection. Further analysis reveals that mid-training provides a stronger initialization for VLA fine-tuning, with gains emerging from the earliest steps and widening throughout training. Moreover, the data engine captures both dataset-level and sample-level alignment signals, favoring spatial reasoning over text-centric tasks while preserving the diversity of the VLM data. We will release all code, data and models for future research.
[NLP-137] Reading the News Without Hindsight: NoLBERT Topics and the Cross-Section of Stock Returns
【速读】： 该论文旨在解决金融新闻语言模型在回测中因训练与测试数据时间重叠而产生的“事后洞察”（hindsight）偏差问题，即模型可能通过学习未来已知的信息导致绩效评估失真。其解决方案的关键在于构建一个基于NoLBERT的新闻主题模型——该模型在1976至1995年间仅使用无标签文本进行预训练，并通过对比学习（contrastive fine-tuning）将其适配为句子编码器，避免引入任何与市场回报相关的标签或目标变量，从而确保训练过程不包含未来信息。研究进一步将不同语言模型（包括LDA、MiniLM和FinBERT）生成的主题转化为注意力冲击（attention shocks）和股票暴露度，并在2012–2024年共126万篇金融新闻上进行多模型、多频率回测。结果显示，尽管单一模型表现不佳，但所有读者与模型的等权组合实现了月度净夏普比率1.12，在未调整搜索偏差前显著高于零，但不及仅基于价格的基准策略。此外，将注意力冲击的时间窗口从5个交易日扩展至63个交易日可显著提升各模型的测试阶段信息系数（IC），表明长期动态捕捉对预测能力至关重要。最终，NoLBERT展现出最稳定的话题结构，且预测性能与传统LDA相当，验证了其在去偏新闻建模中的有效性。
链接: https://arxiv.org/abs/2610.22745
作者: Kevin Foley,Jonathan Hartadi,Shivesh Prakash,Charishma Takkallapalli,Swapnil Vatsal
机构: 未知
类目: atistical Finance (q-fin.ST); Computation and Language (cs.CL); Machine Learning (cs.LG); Portfolio Management (q-fin.PM)
备注:
Abstract:Language models that read financial news are usually trained on text from the same years used to test them, so a backtest can reward hindsight. We build news topics from NoLBERT, an encoder pretrained only on text published between 1976 and 1995, adapted into a sentence encoder by contrastive fine-tuning on passages of the same article, with no labels or returns. We compare it with LDA, MiniLM and FinBERT readers on 1.26 million articles from 2012 to 2024, convert each reader’s topics into attention shocks and stock exposures, and forecast the returns of the 1,000 largest US stocks with eight models at three rebalancing frequencies. Models are fitted on 2013-2019, choices are made on 2020-2021, and 2022-2024 is held out. The experiment with the best validation score loses money at every frequency. Measuring attention shocks against the past 63 trading days instead of 5 raises the test rank IC of every reader. NoLBERT gives the most stable topics and forecasts about as well as LDA. An equal-weight combination of all readers and models earns a monthly net Sharpe ratio of 1.12, significant against zero before adjusting for the search but not against a price-only benchmark.
信息检索
[IR-0] ool Recommendation for Multi-Step Agents : Finding Every Tool a Task Needs
链接: https://arxiv.org/abs/2610.23707
作者: Sam Partee,River Stevens,Thierry Damiba,Valerie Fanelle
类目: Machine Learning (cs.LG); Information Retrieval (cs.IR)
备注: 11 pages, 3 figures, 5 tables
Abstract:A multi-step agent task fails when a single tool it needs is missing, yet the tools an agent may need are too many to list: the roughly 8,200 tools on the platform we study have definitions that together exceed the context window of any model we evaluate. We present Tool Recommendation, a deployed system that gives an agent all of them, hundreds of times more than a typical MCP server exposes, through a fixed interface of three tools. It runs at the gateway between agent and tools, executing every call under the end user’s own credentials and checking it against the user’s consent for each service and the tools the agent may use. We evaluate it on both a public retrieval benchmark and billed end-to-end session cost. On ToolRet, Tool Recommendation places every tool a query needs in the top ten 57% more often than BM25, a lexical ranking that commercial tool search offers, and compares favorably with published systems up to nine times its size. In paired multi-step workflows, its first recommendation covered every required tool in 93% of Codex sessions, and against a fixed list of 100 tools it cut input tokens by 17% with Codex and 40% with Claude Code and the total bill in all 14 runs, by up to 40%, with no change in task completion.
[IR-1] Humanitys Last Choice: A Common Evaluation Standard for Generative Engine Optimization
链接: https://arxiv.org/abs/2610.23541
作者: Qianfeng Wen,Yifan Simon Liu,Difan Jiao,Junda Wu,Di Bai,Jintao Liu,Xidong Wu,Zhenwei Tang
类目: Information Retrieval (cs.IR)
备注: 19 pages, 2 figures. Project website: this https URL
Abstract:Humanity’s Last Choice describes the idea that the last choice people need to make is which AI system they entrust with the decisions that follow. These systems rely on external sources that content providers can edit through generative engine optimization (GEO) to increase source visibility and influence recommendations. Choosing among these systems therefore requires comparable evidence of how well they carry out user requests when exposed to GEO. Benchmarks have made progress in standardizing tasks, models, and metrics, but they often define their own spaces of GEO strategies and use different editing constraints. These differences can conflate gains from better optimization algorithms with those from greater freedom to edit sources, while making evaluation coverage difficult to compare across benchmarks. We therefore call for a common evaluation standard and propose a meta-taxonomy that describes where sources are edited, how they are edited, and what signals those edits target. Drawing on representative literature, we map existing strategies into a searchable space and outline a protocol for controlled comparisons with explicit coverage reporting. This provides a foundation for common leaderboards that can inform which systems people choose to rely on.
[IR-2] LSNR: Learned Sparse News Recommendations with Attentive User Modeling SIGIR
链接: https://arxiv.org/abs/2610.23510
作者: David Vos,Andrew Yates,Maarten de Rijke
类目: Information Retrieval (cs.IR)
备注: 10 pages, 3 figures, 9 tables. Accepted at SIGIR-AP 2026
Abstract:Current learned sparse approaches for recommender systems, such as STAR, represent a user by concatenating their interaction history into a single query string. For news, this creates two bottlenecks: concatenation collapses a user’s distinct reading interests into one flat query, and, because news articles are long-form text, concatenating several news articles quickly overruns the context window of the language model used as backbone and forces quality-degrading truncation. To address both bottlenecks, we introduce learned sparse news recommendations (LSNR). LSNR encodes each article independently and constructs a sparse user vector by combining learned, attention-based weighting of the per-article sparse representations with a dense-projected expansion at the user level. Experiments on the MIND and EBNeRD datasets show that LSNR significantly outperforms STAR, with these gains holding even in short-history settings where truncation is not a factor. Compared to state-of-the-art dense recommenders, LSNR is competitive in standard settings and it outperforms them when the user history is short. An ablation of the user encoder confirms that learned attention-based aggregation outperforms simpler pooling alternatives. Beyond effectiveness, an additional advantage of LSNR is that it serves from an index an order of magnitude smaller than an uncompressed dense baseline. Comments: 10 pages, 3 figures, 9 tables. Accepted at SIGIR-AP 2026 Subjects: Information Retrieval (cs.IR) ACMclasses: H.3.3; H.3.1 Cite as: arXiv:2610.23510 [cs.IR] (or arXiv:2610.23510v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2610.23510 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[IR-3] QD-Rec: Open-Ended Evolution of Recommendations via Quality-Diversity
链接: https://arxiv.org/abs/2610.23470
作者: Takashi Ishida,Taro Makino,Issa Sugiura
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
备注:
Abstract:Open-ended search combined with foundation models has produced advances such as faster algorithms, new results on open mathematical problems, and self-improving agents. We extend this paradigm to recommendation for decision-making, focusing on decisions where the set of possible choices is fixed but the reasons for choosing them are not. In high-stakes decisions, e.g., which company to acquire or which customer to target, comparing diverse, well-reasoned options can be more useful than a single ranked list. We formalize open-ended recommendation as building a diverse, high-quality set of recommendations, each an item together with a hypothesis, a natural-language reason why the item fits the needs of a decision maker. We propose QD-Rec, a quality–diversity method that selects items with a contextual bandit and generates hypotheses with a foundation model from earlier ones kept in an archive as stepping stones. A new recommendation enters the archive if it is novel or better than its nearest neighbor. Across MA, customer targeting, and paper and book recommendation, QD-Rec gives the highest mean quality and builds the largest archive in most settings.
[IR-4] Jev-as-a-Judge for Brazilian Portuguese Information Retrieval: Relevance Assessment and Reranking
链接: https://arxiv.org/abs/2610.23346
作者: Leandro Carísio Fernandes,Jayr Pereira
类目: Information Retrieval (cs.IR)
备注:
Abstract:This study investigates Jev, a System One model designed for fast, structured decision-making, in information retrieval. We use Jev-as-a-judge to generate relevance judgments for four Brazilian Portuguese IR collections covering different domains and compare them with the official relevance judgments of each collection. We also evaluate Jev as a reranker. For relevance assessment, exact agreement ranged from 52.9% to 64.9%, while quadratic weighted Cohen’s kappa ranged from 0.52 to 0.82. Most disagreements involved adjacent relevance levels, with at least 94.5% of annotations matching exactly or differing by only one relevance level. Despite differences at the individual query–document level, Jev-generated qrels produced similar retrieval system rankings, with Kendall’s tau ranging from 0.90 to 0.93 for NormasTCU and Ulysses-RFCorpus across the two evaluation metrics. For reranking, Jev reordered the top 1,000 BM25 candidates for each of the 50 Quati test queries, increasing nDCG@10 from 0.3329 to 0.6331 (+90.2%) and MRR from 0.6852 to 0.9033 (+31.8%). These findings support Jev as a relevance assessor and show improved ranking effectiveness over BM25 in the evaluated reranking setting.
[IR-5] Code Skill Server: A Large-Scale Search Engine for Code Generation Skills
链接: https://arxiv.org/abs/2610.23186
作者: Xiaoyong Li,Haolong Li,Dongliang Xu,Haoyu Li,Yifan Wang,Jingyuan Shi,Yanghua Wang,Yanbin Sun,Shidong Pan,Yue Yao
类目: oftware Engineering (cs.SE); Information Retrieval (cs.IR)
备注:
Abstract:Agent skills have proven to be an effective technique for improving agent ability on specific tasks, including software engineering, natural science, and mathematics. During such a process, using agent skills often involves manually discovering and organizing skills from heterogeneous sources, while existing skill retrieval methods mainly focus on selecting relevant skills from established skill libraries. However, with the rapidly increasing number and variety of reusable agent skills, manually searching for relevant skills across heterogeneous sources is becoming impractical at scale. In this paper, we introduce the Code Skill Server (CSS), a large-scale search engine for finding the most useful code generation skills for a target task. CSS consists of a skill server that incrementally collects and indexes publicly available code generation skills from multiple sources, with 993 skills included in the corpus used in this work. Extensive experiments across different retrieval methods and code generation benchmarks show that the best CSS configuration improves Pass@1 by up to 13.91% and consistently improves all evaluated models by at least 4.21% in the cross-model evaluation. These results demonstrate that CSS can effectively search for and serve relevant skills, improving code generation quality and enhancing models’ code generation capabilities. Code and the CSS-993 corpus are available at this https URL.
[IR-6] Reasoning Reranking with Semantic IDs as the Sole Item Representation
链接: https://arxiv.org/abs/2610.22797
作者: Zhenwei Tang,Jintao Liu,Di Bai,Xidong Wu,Xinwu Cheng
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 30 pages, 3 figures. Code: this https URL ; data: this https URL
Abstract:Reasoning rerankers write a rationale before ranking a candidate slate, and existing systems read item metadata to do so, even though generative recommenders increasingly represent items as compact Semantic IDs (SIDs). Ranking from SIDs alone has two appeals: SID prompts are far shorter than the matched item text, and although item codes are fixed from content, the backbone learns their token embeddings in part from user-item interaction sequences, so the tokens can carry behavioral signal that item text lacks. That training targets next-item generation, and whether it supports ranking a supplied slate from SIDs alone is untested. We build ReRankID, a controlled testbed in which models reason and rank from a history and slate shown only as SIDs, varying item rendering, history length, and slate size one at a time. On this testbed, such models often copy the input order instead of reranking, and readable metadata does not repair the failure. We compare supervised fine-tuning, rejection-sampled self-training, on-policy distillation, and verifier-based reinforcement learning as post-training methods and find that they elicit the missing behavior. The strongest post-trained model, the 8B model trained with a graded ranking reward, gives the best SID-only ranking and matches strong text-reading LLMs whose prompts are about 20x longer. It ranks above random with no history at all, where a caption-reading reference does not, and it keeps its gains at long histories, where text prompts pass 50k tokens and the caption-reading reference stops improving. In a paired rationale audit, this model argues trade-offs better than that reference but describes items less accurately. The common post-training methods we test thus elicit ranking from SIDs but not faithful item descriptions, which calls for post-training that improves the reasoning itself, not only the ranking it produces.
[IR-7] Improving OCR Accuracy Through Multi-Classifier Verification of Visually Similar Characters
链接: https://arxiv.org/abs/2610.22726
作者: Nuzhat Khan,Ab Al-Hadi Ab Rahman,Mohamad Anuar Kamaruddin,Muhammad Paend Bakht,Sana Fatima,Farhad Khodadadi,Ibrahim Yousef Alshareef
类目: Computer Vision and Pattern Recognition (cs.CV); Information Retrieval (cs.IR)
备注: 32 pages, 8 figures, 2 tables. Submitted to Computer Vision and Image Understanding (CVIU). Dataset available at this https URL
Abstract:Distinguishing visually similar characters is a persistent challenge for optical character recognition (OCR) systems, causing errors in critical applications of text digitization. While preprocessing techniques improve image quality and word-level post-processing corrects spelling errors, neither addresses character-level visual ambiguity, an intrinsic property of certain character pairs that limits even state-of-the-art OCR models. We present the Character Verification Model (CVM), a character-level post-processing framework that verifies and corrects ambiguous characters in OCR outputs using specialized binary classifiers trained on visually confusable pairs such as ‘O’/‘0’, ‘Z’/‘2’, and ‘S’/‘5’. CVM operates independently of dictionary validation, making it suitable for non-lexical data including serial numbers, alphanumeric codes, and license plates. Experiments on three OCR engines (Tesseract, EasyOCR, TrOCR) and ten fonts with low to high complexity demonstrate accuracy improvements of 3.95-14.81 percentage points, with an 87.3% correction rate for visually similar character errors and a false-correction rate of only 2.3%. The modular architecture integrates with existing OCR pipelines without modification to underlying recognition models.
[IR-8] NoveltyScope: A Multi-Agent Research Assistant for Evidence-Grounded Novelty Assessment of Scientific Papers
链接: https://arxiv.org/abs/2610.22703
作者: Yueran Tian,Huizhi Liang
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
备注: 13 pages, 1 figure
Abstract:NoveltyScope is a prototype that compares a scientific paper with retrieved prior work and produces a structured report and a five-level novelty score. The system extracts the core task and at most two author-claimed contributions, then searches a closed local corpus using BM25 and dense retrieval with reciprocal rank fusion. Contribution comparisons request paired passages for potential refutations, and quotations are checked against cached source text. Four scoring roles (Prosecutor, Defender, Judge, and Critic) use the resulting evidence, with at most one revision of the Judge’s score. Evaluation uses 50 class-balanced RINoBench samples whose idea-level labels are retained for full-paper input. Mean absolute error is 1.48 and 1.26 for single-LLM and multi-agent scoring with Gemini, respectively, and 1.22 and 1.16 with Qwen3. The vector-similarity baseline has an error of 1.68. A HiCLIP case study illustrates the contribution comparisons and paired evidence passages. Overall, the results demonstrate the feasibility of evidence-grounded novelty assessment and show that the multi-agent configurations achieve lower scoring errors than their single-LLM counterparts across both evaluated models.
[IR-9] AuditKIT: A Unified Evaluation Spine for Benchmarking Judging and Comparing Large Language Models and Agent ic Systems
链接: https://arxiv.org/abs/2610.22511
作者: Shreeyans Arora,Utsav Avaiya,Zera Lyngkhoi,Hem Gosalia,Ram Mohan Rao Kadiyala,Vinay Kumar Sankarapu,Pratinav Seth
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Software Engineering (cs.SE)
备注: 25 pages, 2 figures, 13 tables. Code: this https URL . Shreeyans Arora and Utsav Avaiya contributed equally; work done while at Lexsi Labs
Abstract:Evaluating language models and agents requires combining benchmark accuracy, judge-based assessment, runtime performance, tool-use behavior, and retrieval quality. Differences in interfaces, metric semantics, and trace formats complicate these comparisons and can hide failures in agentic workflows. We present AuditKIT, a source-available evaluation library that records these measurements in one result representation, with direction-aware metric contracts, configuration fingerprints, and extensible metrics and backends. For agents, it provides turn-aware tool-call matching, separates parallel from sequential execution, and distinguishes verified task outcomes from trajectory-level diagnostics. Missing or unobservable evidence is marked unscored rather than counted as failure. For retrieval-augmented generation, it combines ranked retrieval metrics, judged grounding, and deterministic evidence checks. Seven case studies spanning model comparison, serving runtimes, retrieval, tool calling, and interoperability show how answer extraction, runtime behavior, and metric selection change evaluation outcomes.
[IR-10] Star Light Star Bright: Can Multimodal Models Parse an Astronomy Visual Question Answering Dataset Right? AACL
链接: https://arxiv.org/abs/2610.22474
作者: J. P. Naiman
类目: Computer Vision and Pattern Recognition (cs.CV); Instrumentation and Methods for Astrophysics (astro-ph.IM); Digital Libraries (cs.DL); Information Retrieval (cs.IR)
备注: 5 main pages, 12 pages total, accepted to The 4th Annual Workshop for Artificial Intelligence for Scientific Publications (WASP) 2026, International Joint Conference on Natural Language Processing Asia-Pacific Chapter of the Association for Computational Linguistics (IJCNLP-AACL)
Abstract:Figures are essential for scientific communication, and the field of Visual Question Answering (VQA) aims to test the figure understanding capabilities of large multimodal models (LMMs). Here, we investigate the accuracy of several low-cost LMMs on domain-specific VQA tasks for astronomical images of the sky. As instrument for this task we build a domain-specific VQA dataset comprising 2001 partially-synthetic figures divided evenly between contour plots, real telescope images drawn from NASA’s SkyView, and “fake” sky images which share the axes and angular scales of the real images but draw their colors from a gaussian mixture model, alongside 47357 questions. Each figure is released as pristine and with realistic page-aging degradations mimicking the historical astronomical corpus. Testing three low-cost proprietary LMMs, we find models frequently fail to answer domain-specific questions at high accuracy. An initial comparison with 104 human-verified, astrometrically tagged figures from historical journal publications shows similar median errors in angular extent. All code is available on GitHub: this https URL and dataset is available on HuggingFace: this https URL.
人机交互
[HC-0] Bad Diagnosis: Analyzing Consent and Data Collection Practices of Mobile Medical Apps
链接: https://arxiv.org/abs/2610.23851
作者: Ethan Myers,Lorenzo De Carli,Shantanu Sharma,Ritwik Banerjee,Joel Reardon,Indrakshi Ray
类目: Human-Computer Interaction (cs.HC); Cryptography and Security (cs.CR); Computers and Society (cs.CY)
备注:
Abstract:Mobile health (mHealth) apps collect personally-identifiable information while handling uniquely sensitive medical data, ranging from mental health records to real-time glucose or heart rate monitoring – yet they remain poorly understood at scale. We present the first comprehensive analysis of consent and data collection practices across 10,000+ mHealth apps on the Google Play Store, combining static analysis and privacy policy examination. We develop a rigorous operational definition to distinguish health apps from general wellness programs, and design a measurement framework to detect misuse of permissions and violations of consent. Our analyses reveal a troubling ecosystem: 69.15% of apps request at least one permission unrelated to their stated purpose, with over 52.9% apps requesting location data, and 50.5% and 34.9% apps requesting camera and microphone access, respectively – despite no functional justification based on app descriptions and stated purposes. Most alarmingly, nearly 93% of apps fail to provide users with a consent form, directly violating regulatory requirements such as GDPR; and even among the remaining 7%, 1 in 14 demonstrate a gap between the obtained consent and collection of personally identifiable information. These findings demonstrate an urgent need for stronger enforcement of privacy regulations in mHealth apps. We plan to release our dataset and measurement tools for continued monitoring of this critical space.
[HC-1] Sensory Comfort Across Multisensory XR Prototypes for Autistic Adults
链接: https://arxiv.org/abs/2610.23833
作者: Yibo Meng,Xiujun Chen,Bingyi Liu,Xiaolan Ding
类目: Human-Computer Interaction (cs.HC)
备注: ACM ISS 2026 PACMHCI full paper; revised manuscript; 4 figures; accepted paper; Article ISS030
Abstract:Multisensory XR may support informal learning for autistic adults, but sensory-friendly design often emphasizes reducing input. We conducted a mixed-methods, counterbalanced within-subjects study with 20 Chinese autistic adults who evaluated five complete XR configurations presenting the same popular-science content. Comfort differed substantially across the configurations (Friedman chi-square(4) = 46.39, p .001, Kendall’s W = .58). L3 received the lowest mean rating (2.97/5) and was tied for lowest for 16/20 participants, whereas L5 received the highest mean rating (4.43/5) and was tied for highest for 17/20. Participants described the static configurations in terms of clarity and self-pacing, L3 in terms of abruptness, simultaneous demands, insufficient warning, and limited disengagement, and L4/L5 in terms of gradual transitions, opportunities to disengage, and coordinated narrative cues. The study provides a configuration-level account of sensory comfort and identifies participant-described qualities that can inform future controlled studies of multisensory XR.
[HC-2] “To Boldly Go Where No One Has Gone Before” What Has AI Let Developers Unlock?
链接: https://arxiv.org/abs/2610.23708
作者: Rudrajit Choudhuri,Jenna L. Butler,Christian Bird,Carmen Badea,Brian Houck,Anita Sarma
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Measuring AI by how much faster developers work is like judging electricity by how much faster it spun a steam-era factory. A fuller measure asks what new work becomes possible. We surveyed 331 software developers at Microsoft about what AI lets them do that they could not do before and analyzed their answers using reflexive thematic analysis with a human-in-the-loop, multi-model AI council. Developers described doing familiar work faster (Faster, 45%), undertaking work previously beyond their reach (Further, 36.9%), and taking on new kinds of work centered on building for and directing AI agents (Afresh, 12.4%). Across these accounts, learning and scaffolding ran in both directions: developers learned new skills with AI while externalizing their tacit knowledge as the context and standards that guide AI’s work. Drawing on Vygotsky’s zone of proximal development, we frame developers and AI as joint learners whose zones meet at the skill-judgment nexus: AI can supply skill faster than it can reliably judge its own work, and developers must continuously bridge that gap. Whether they do sets them on one of two paths. On one, developers’ judgment grows alongside AI’s skill, and they codify what they learn so the loop keeps turning. On the other, they become hollow orchestrators, approving work they can no longer judge and losing the identity and meaning they once drew from it. Based on our findings, we offer a two-way scaffolding playbook to help developers, teams, and tool builders take the first path.
[HC-3] PresenceBench: An Exploratory Evaluation of Large Language Model Adaptation to User State
链接: https://arxiv.org/abs/2610.23625
作者: Peña Arroyo Gallego,Mauro Ruiz Moreno,Alejandro Acien,Sergio Romero-Tapiador,Teresa Arroyo-Gallego
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 33 pages, 6 figures, 13 tables
Abstract:Presence-aware intelligence is the capacity of an AI system’s intelligence layer to adapt appropriately to the functional state of the person it serves (here readiness, flow, and stability) while preserving task quality and safety. We introduce PresenceBench, an instrument for measuring it. PresenceBench evaluates how the intelligence layer uses a supplied state, not how that state is sensed, and breaks the capability into measurable properties. Two are tested here. Receptivity: does the model take up the state and give a better response than with no signal? Specificity: does any improvement depend on the correct state, rather than on extra structured context? A third, adaptation across states, is a required endpoint not measured in this release. A five-arm causal design answers both questions, against each model’s own baseline and an equally structured mismatched-state control, scored by a correctness-gated human paired comparison. The exploratory evaluation covered four models across eighteen single-turn, six fixed three-turn and six simulated natural-conversation scenarios. Supplying a synthetic state, raw or translated into a response policy, gave human-preference point estimates above 0.5 for all four models (Claude Sonnet 5, DeepSeek V4 Pro, Gemini 3.6 Flash, GPT-5.6 Terra), with three of eight 95% intervals excluding chance. Specificity was not resolved: no per-model interval excluded chance, and the design cannot yet separate the state values from a generic instruction to adapt. Policy delivery reduced priced cost and output tokens against each model’s baseline; these are resource-use changes, not efficiency gains, because correctness non-inferiority was not established. Safety-floor failures were rare (13 of 360 cases), with a non-significant excess under policy delivery. We report the instrument, the frozen results, and the controls a confirmatory study needs.
[HC-4] mmPencil: Toward Writing-Style-Independent In-Air Handwriting Recognition via mmWave Radar and Large Vision-Language Model
链接: https://arxiv.org/abs/2610.23419
作者: Yifan Guo,Zhu Wang,Qian Qin,Yangqian Lei,Qiwen Gan,Zhuo Sun,Chao Chen,Bin Guo,Zhiwen Yu
类目: Human-Computer Interaction (cs.HC); Computer Vision and Pattern Recognition (cs.CV)
备注: Published in ACM IMWUT, Vol. 9, No. 3, Article 81, 2025; presented at ACM UbiComp 2025. 30 pages
Abstract:In-air handwriting recognition based on radio frequency (RF) signals offers a promising solution for human-computer interaction, yet existing methods are often limited by predefined writing planes, constrained writing styles, and small recognition vocabularies. To address these limitations, we propose m^2VLMs, a modality mapping architecture that bridges millimeter-wave (mmWave) radar sensing and large vision-language models (VLMs), and develop mmPencil, a 3D in-air handwriting word recognition system. mmPencil reconstructs 3D handwriting trajectories from mmWave signals using a multi-stage spatial trajectory reconstruction algorithm and maps them into visual representations through a spatial-to-visual mapping algorithm, enabling VLMs to recognize handwriting while reducing sensitivity to writing styles and environmental factors. We collect and release a 3D mmWave handwriting dataset containing 7,664 samples of 200 words from 12 users across 22 writing scenarios. Extensive experiments demonstrate that mmPencil achieves robust word recognition under variations in writing position, range, angle, speed, size, direction, and user movement. It achieves 97.60% accuracy on 200 word classes for seen users and 92.50% zero-shot accuracy on 50 word classes for unseen users, outperforming state-of-the-art baselines. Comments: Published in ACM IMWUT, Vol. 9, No. 3, Article 81, 2025; presented at ACM UbiComp 2025. 30 pages Subjects: Human-Computer Interaction (cs.HC); Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2610.23419 [cs.HC] (or arXiv:2610.23419v1 [cs.HC] for this version) https://doi.org/10.48550/arXiv.2610.23419 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Journalreference: Proc. ACM Interact. Mob. Wearable Ubiquitous Technol. 9, 3, Article 81 (September 2025), 30 pages Related DOI: https://doi.org/10.1145/3749504 Focus to learn more DOI(s) linking to related resources
[HC-5] Wait for It: A Component-Level Analysis of Latency Feedback Mechanisms for LLM Agents in VR
链接: https://arxiv.org/abs/2610.23336
作者: Pratham Arora,Abhinav Lodha,Siddharth Siddharth
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Delays in AI responses pose a significant challenge to immersive Virtual Reality experiences. While previous research demonstrates that feedback can reduce the perception of waiting, we lack understanding of which types work best and why. We conducted a within-subjects study with 18 participants to systematically compare three isolated feedback approaches during VR agent interactions: verbal fillers (e.g., um…, hmm…), visual indicators (loading spinners), and embodied gestures (thinking animations). Our results revealed that embodied gestures significantly outperformed visual loading indicators across multiple dimensions, making agents feel more lifelike and socially present. Interestingly, explicit visual feedback like loading spinners actually degraded perceived humanness and broke immersion by forcing users to shift from social to technical mental models. These findings suggest that natural, behavior-based feedback maintains the illusion of conversing with a thinking entity, while technical indicators expose the computational substrate. We present empirically-grounded design guidelines for creating more engaging AI-powered VR agents.
[HC-6] Designing for Imperfect Prompts: A Simulation-Based Evaluation of an LLM -Based Enterprise Data Assistant Using Knowledge Graphs
链接: https://arxiv.org/abs/2610.23304
作者: Niklas Wagner,Alexander Mädche
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Databases (cs.DB)
备注: Published in the Proceedings of the 34th European Conference on Information Systems (ECIS 2026)
Abstract:Enterprise data assistants using large language models (LLMs) promise democratized data access in organizations. Yet in our real-world deployment within the CFO unit’s data lake at an industry partner, we observed that business users frequently pose imperfect prompts that include domain-specific vocabulary and lack schema references or structural information. This Design Science Research study examines automatic resolution using a semantic mediation layer. We design and evaluate two design variants: schema-based mediation via semantic similarity, and knowledge graph-based mediation leveraging an automatically constructed knowledge graph from the data lake. Our simulation-based evaluation using real-world data, supported by LLM-as-a-Judge assessments and human validation, shows that knowledge graph-based mediation outperforms the schema-based approach in Precision, Recall, F1, and preserving user intent. We propose a design principle for knowledge graph-based mediation, demonstrating that structured knowledge representation bridges user intent and database structure and guides the design of enterprise data assistants across skill levels.
[HC-7] “Can Computing Systems Eventually Disappear?”: A Long-Term Tracking Study on the Use of Diabetes Computing Systems by Patients with Diabetes
链接: https://arxiv.org/abs/2610.23276
作者: Yibo Meng,Jinyuan Zhan,Yao Lyu,Zhuoran Lu,Chengxi Zang
类目: Human-Computer Interaction (cs.HC)
备注: Accepted to Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), Vol. 10, No. 4, 2026
Abstract:Ubiquitous computing has long pursued Weiser’s vision of technologies that “disappear” into everyday life. As wearable health systems advance, they increasingly substitute for rather than scaffold users’ judgment, yet the long-term consequences remain underexplored. We report an eight-year, five-wave longitudinal interview study from 2017 to 2025 with 37 patients with diabetes, complemented by interviews with 15 clinicians, designers, and vendors. Across three technological generations including glucometers, continuous glucose monitoring, and predictive and automated insulin delivery systems, we identify a novel user state, continued use with cognitive disengagement. In this state, devices remain in use but active interpretation and questioning recede, leaving users to respond only to alarms. This state emerges from functional expansion, delegation of interpretive authority, and information fatigue, and is reinforced by clinical prioritization, information compression, and automation. It results in less frequent independent health judgment and a shift from understanding to responding. We argue that future diabetes computing systems should reduce unnecessary cognitive work while preserving users’ opportunities to interpret, contextualize, and question system outputs.
[HC-8] When Ubiquitous Health Data Stops Helping: Patients Experience of Breakdowns in the Data-Interpretation-Action Loop at the End of Life
链接: https://arxiv.org/abs/2610.23259
作者: Yibo Meng,Xinru Wang,Zhuoyan Li,Chengxi Zang,Zhuoran Lu,Zhicong Lu
类目: Human-Computer Interaction (cs.HC)
备注: Accepted to Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), Vol. 10, No. 4, 2026
Abstract:Ubiquitous computing systems in healthcare often assume that greater data visibility improves awareness, interpretation, and action. In end-of-life care, however, this assumption warrants scrutiny. Through semi-structured interviews with 25 terminal-stage patients as primary informants and 8 clinicians as supplementary informants, we examine patients’ descriptions of how they encountered, interpreted, acted on, and disengaged from state signals, trajectories, algorithmic outputs, and behavioral audits. Our analysis identifies three recurrent patterns in these descriptions. First, participants described repeated and unresolved sensemaking when visible data lacked explanatory pathways. Second, they described outputs they could understand but whose suggested actions did not fit their changing bodily capacities, care goals, or decision-making roles. Third, they described difficulty disengaging when data remained embedded across care environments. Rather than rejecting data altogether, patients called for less data exposure and a reorientation toward existential support. Based on this analysis, we situate data exposure as a central interaction problem in end-of-life ubiquitous computing and derive design implications for interpretive stability, care-context alignment, and existential support.
[HC-9] QuAD: Quantitative Assessment via Distribution-Alignment for Qualitative Data Quality Enabled by a Content-Aware Multi-Agent System
链接: https://arxiv.org/abs/2610.23153
作者: Wenshuo Zhang,Leixian Shen,Qiqi Zhou,Shuchang Xu,Feiyang Hao,Huamin Qu,Haobo Li,Zhao Chen,Lin-Ping Yuan
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:
Abstract:Assessing whether qualitative responses collected from semi-structured questions are ready for a particular analysis goal is a persistent challenge. Manual qualitative analysis is interpretive and labor-intensive, while existing automated methods often lack a dataset-level perspective for deciding whether a response set supports a researcher-selected analytical lens. We introduce QuAD, a framework for quantitative pre-analysis assessment of qualitative response datasets. QuAD compares a reconstructed intrinsic cluster distribution reconstructed from responses with an extrinsic, researcher-defined analytical lens derived from metadata. QuAD is structured into a four-quadrant assessment whose metrics are read jointly, from lens-level relations to within-cluster variation, to provide fitness-for-purpose signals for revealing general patterns and unique insights. Because this assessment relies on lens-conditioned inter-response dissimilarity, we propose a content-aware multi-agent system with two-stage instruction generation and adaptive anchors to improve scoring consistency. We validate QuAD through controlled experiments on synthetic data, including robustness and baseline comparisons, and a real-world social survey case study. The case study shows how QuAD quantitatively assesses whether a selected analysis axis is promising and also surfaces representative or unusual responses for human interpretation.
[HC-10] Can AI interview? Comparing AI-led to Human-led Chat Interviews
链接: https://arxiv.org/abs/2610.23149
作者: Semra Guven,Tobias Gårdhus,Andreas Bjerre-Nielsen,Anna Rogers,Hjalmar Bang Carlsen
类目: Computers and Society (cs.CY); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC)
备注: 37 pages, 3 figures, 6 tables. Submitted to Socius. Code and data: this https URL
Abstract:Large language model chatbots have been proposed as AI-interviewers for collecting qualitative interview data at scale, potentially bridging social science’s longstanding divide between the breadth of surveys and the depth of qualitative interviews. This paper provides the first direct empirical comparison of data quality in AI-led and human-led interviews. In a randomized experiment (N = 41), university students were interviewed via synchronous text chat by either trained non-expert interviewers or a locally hosted AI system; we evaluate exposure, specificity, and relevance. Median interview duration was similar across conditions, but AI-interviewers asked more than twice as many questions per interview and elicited roughly 65 percent more respondent text per interview. Per-question responses were longer with human interviewers, though not after multiple-testing correction; only AI-condition participants skipped questions. Responses were almost uniformly on-topic in both conditions; for specificity we detect no difference, but the design cannot exclude one of up to three quarters of a point. These results were obtained with a small, locally hosted open-weight model (7B parameters), which addresses concerns about data security, GDPR compliance, and scientific reproducibility. Our findings support the viability of AI-interviewing for large-N qualitative data collection.
[HC-11] From GAN-Based Models to GPT -5: A Taxonomy-Guided Study of Human Performance in Detecting Deepfake Face Images
链接: https://arxiv.org/abs/2610.23144
作者: Zhiqing Guo,Lixin Jia,Luning Sun,Xiaowei Chen,Guofeng Yu,Liejun Wang,Gaobo Yang,David Stillwell,Thorsten Holz,Shujun Li
类目: Computer Vision and Pattern Recognition (cs.CV); Cryptography and Security (cs.CR); Human-Computer Interaction (cs.HC)
备注:
Abstract:Generative models for deepfakes have rapidly evolved from GAN-based architectures to diffusion models and multimodal large language models (MLLMs). However, empirical research on human performance in detecting deepfake face images has not kept pace. It remains unclear if (and how) detection performance varies across generative models and whether model-specific training improves detection more effectively than general-purpose training. In this paper, we propose the \emphDeepfake Face Artifact Taxonomy (DFAT), a cross-model taxonomy of human-recognizable visual artifacts in deepfake face images, based on which we build both general-purpose and model-specific training materials. Through an online experiment ( n = 524 ), we examined human performance in detecting deepfake face images and evaluated training efficacy across five representative generative models: three GAN-based, one diffusion-based, and one MLLM (GPT-5). Our results show that participants’ detection accuracy was strongly model-dependent and that images generated by more recent and advanced models were not necessarily harder to detect. After training, participants achieved significant performance gains across most conditions, and their improvement varied by model. Notably, model-specific training generally yielded greater numerical gains than general-purpose training. In a five-month follow-up test ( n = 342 ), we observed no significant decline in detection accuracy and no significant increase in perceived stress. Overall, our findings provide actionable insights for designing effective, sustainable, and targeted training programs to counter the growing threat posed by deepfakes.
[HC-12] he Ethics of Prompt Engineering
链接: https://arxiv.org/abs/2610.23082
作者: Federico Germani,Giovanni Spitale
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Prompt engineering is commonly understood as a technical practice aimed at eliciting useful, accurate, and appropriately structured outputs from large language models (LLMs). Yet prompts do more than optimize model performance: they shape what information is elicited, which assumptions are introduced, and which perspectives are privileged. In this paper, we argue that prompting should therefore be understood as an epistemic and normative act. We distinguish between two conceptions of prompting: knowledge generation, in which the model is implicitly treated as an epistemic authority, and knowledge injection, in which users provide relevant knowledge, evidence, and context that the model can organize, articulate, and extend. We contend that the latter provides a more appropriate model for responsible LLM use, since prompting cannot substitute for subject-matter expertise or the critical evaluation of model outputs. At the same time, prompts inevitably inject values, intentions, and assumptions into the generation process. Building on these premises, we propose five principles for responsible prompting: epistemic humility, explicit engagement with uncertainty and epistemic underdetermination, source sensitivity, avoidance of deceptive framing, and recognition of model limitations and appropriate task allocation. We further argue that responsibility for AI-generated outputs is distributed across users, developers, institutions, and societies, and that some applications cannot be rendered ethical through better prompting alone.
[HC-13] PersonaCUABench: Benchmarking Computer-Use Agents Serving Diverse Personas
链接: https://arxiv.org/abs/2610.22990
作者: Haoming Li,Defu Cao,Xu Hu,Yingtao Luo,Yiyang Wang,Xianghe Xu,Jiashu He,Wanting Jiang,Yifan Zhang,Yunhan Zhao,Siyang Wu,Chen Yang,Haixin Wang,Zheyuan Deng,Taoran Li,Jingwen Zhang,Zhejian Peng,Shushu Liang,Yue Zhao,Bingzhe Li,Feng Chen,Yifan Wang,Jintao Huang,Yijun Wang,Yuexing Hao,Xiaomin Li
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Machine Learning (cs.LG)
备注: 31 pages, 13 figures, 24 tables
Abstract:Computer-use agents (CUAs) are commonly evaluated against fixed task specifications, but users may omit requirements, add constraints, or require approval before consequential actions. Completing the benchmark instruction does not guarantee serving the person who asked. We introduce PersonaCUABench, which holds 74 tasks across 11 domains fixed and varies only who asks. A persona agent, playing one of 44 personas in four groups, rewrites each instruction in that user’s own words and answers the CUA’s questions, while the CUA sees only the dialogue, never the profile or the original instruction. Across nine backbones (3,330 runs), mean task completion falls by 36% relative to the original requests. The losses have two sources: in 59.9% of cells that lose points, the agent engages with what the user asked for and still gets it wrong, and 11.2% are stops the user requested that the checker scores as failures. Even when a persona request is fully completed, the user gives the top satisfaction rating in only 62.7% of cells, against 84.1% for the original requests. CUAs should therefore be evaluated not only on task completion but also on whether they meet the requirements and constraints users express. Code: this https URL. Project page: this https URL
[HC-14] Designing Human-AI Collaborative Science: The 9A Framework
链接: https://arxiv.org/abs/2610.22948
作者: Zihan Zheng,Linyue Pan,Xin Lan
类目: Human-Computer Interaction (cs.HC)
备注: 19 pages, 2 figures, 2 tables
Abstract:Artificial intelligence (AI) for Science is often described as a progression from human-directed assistance toward increasingly autonomous AI scientists. Human-AI collaboration offers a further mode in which human and AI contributions substantively shape scientific reasoning, judgments, and the course of inquiry. Existing research addresses work allocation, human-AI coordination, and system evaluation, but how these aspects interact to shape collaborative scientific systems remains insufficiently articulated. We develop the 9A Framework for Human-AI Collaborative Science (HACS) to connect these considerations across three layers. The organization layer specifies what scientific activities are required (Activity), who performs them (Actor), and who holds authority over consequential decisions (Authority). The coordination layer addresses how participants access relevant research information (Awareness), how scientific work advances through iterative proposal, execution, examination, revision, and resolution (Accountability), and when collaborative arrangements should be adjusted (Adaptability). The evaluation layer establishes what constitutes effective system performance (Achievement), how comparisons can be designed to identify the effects of organizational and coordination choices (Attribution), and what evidence supports the resulting conclusions (Assurance). Together, these layers frame HACS as a system-design problem, providing a basis for designing collaborative scientific workflows and for empirically comparing their effectiveness.
[HC-15] SwitchTube: A Proof-of-Concept System Introducing “Adaptable Commitment Interfaces” as a Tool for Digital Wellbeing
链接: https://arxiv.org/abs/2610.22947
作者: Kai Lukoff,Ulrik Lyngs,Karina Shirokova,Raveena Rao,Larry Tian,Himanshu Zade,Sean A. Munson,Alexis Hiniker
类目: Human-Computer Interaction (cs.HC)
备注: 22 pages. Published in Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems (CHI '23)
Abstract:YouTube has many features, such as homepage recommendations, that encourage users to explore its vast library of videos. However, when users visit YouTube with a specific intention, e.g., learning how to program in Python, these features to encourage exploration are often distracting. Prior work has innovated ‘commitment interfaces’ that restrict social media but finds that they often indiscriminately block needed content. In this paper, we describe the design, development, and evaluation of an ‘adaptable commitment interface,’ the SwitchTube mobile app, in which users can toggle between two interfaces when watching YouTube videos: Focus Mode (search-first) and Explore Mode (recommendations-first). In a three-week field deployment with 46 US participants, we evaluate how the ability to switch between interfaces affects user experience, finding that it provides users with a greater sense of agency, satisfaction, and goal alignment. We conclude with design implications for how adaptable commitment interfaces can support digital wellbeing.
[HC-16] Immersive FocusContext Techniques to Assist in Interpretation of High Density Forest Point Clouds
链接: https://arxiv.org/abs/2610.22878
作者: Spencer O’Keeffe,Jan Rombouts,Bruce H Thomas,Michelle Balasso,Jim O’Hehir,Andrew Cunningham
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Virtual Reality (VR) is changing how we interact with complex spatial datasets. VR has the potential within forestry to help analyse remotely sensed spatial data for inventory management. However, challenges of scale and legibility of dense forest point clouds must be addressed. To do so, we present three VR Focus+Context (F+C) techniques: Wedgelight, Remote Lantern, and Ray Lantern. These techniques present point cloud data using a constrained highdetail focus within a broader abstracted context representation, manipulated using egocentric and exocentric interactions. We conducted two user studies. The first informed the design of the F+C techniques. The results from the first study indicate that VR significantly outperforms desktop displays in basic analysis of forestry point clouds, with improvements to both time and accuracy. The second study evaluated the performance of the resulting F+C techniques in a task analogous to real-world forestry operations. Our F+C techniques significantly outperformed current approaches, confirming their viability for facilitating detailed and effective forest assessments. These findings indicate that VR has significant advantages for spatial data analysis tasks in forestry, making a strong argument for its adoption in future analytical frameworks. The F+C visualisation techniques we introduced show considerable promise for analysis of high density point clouds, opening new avenues for development in immersive analytics. Supplemental materials (code, scans, results) are available at this https URL.
[HC-17] ElectroCalm: Investigating User Experience and Physiological Responses to Multi-Node Transcutaneous Electrical Stimulation for Stress Modulation
链接: https://arxiv.org/abs/2610.22748
作者: Yoon Kyung Shon(1),Pablo E. Paredes(1) ((1) University of Maryland)
类目: Human-Computer Interaction (cs.HC)
备注: 25 pages, 6 figures, to appear in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, Vol. 10, No. 4, December 2026
Abstract:Transcutaneous Electrical Stimulation (TES) for stress modulation has largely focused on vagus nerve stimulation at the neck, yet two questions remain underexplored: how users experience electrotactile sensations at this sensitive site, and whether electrotactile feedback itself can function as a haptic stress intervention. To investigate this gap, we applied an 8-node TES array to two body sites during a cognitively demanding 2-back task: the neck, a site used in prior vagal TES research for its anatomical proximity to the cervical vagus nerve, and the arm, the most common haptic wearable site, which primarily provides cutaneous electrotactile sensation without vagal access. In a usability study, both sites were rated as comparably comfortable without disrupting task performance, though the same stimulation was interpreted differently across sites. A second study assessing the physiological effects of active versus inactive stimulation under identical electrode attachment revealed that only neck TES produced cortical and autonomic changes consistent with stress modulation, while arm TES did not differ from inactive conditions. These findings suggest that neck TES can be acceptable despite the site’s sensitivity, and that both the physiological effects and the sensory experience of TES depend on placement.
[HC-18] A misinformation playbook sharpens judgement without breeding cynicism
链接: https://arxiv.org/abs/2610.22724
作者: Tong Guo,Teng Ye,Yutong Xie,Qiaozhu Mei
类目: Computers and Society (cs.CY); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC); Social and Information Networks (cs.SI)
备注:
Abstract:Misinformation harms public health and democracy, and generative artificial intelligence now writes falsehoods as persuasive as people’s. Defences range from fact-checking individual claims to training people to resist falsehoods. Yet many claims are too new to check, and interventions that teach doubt can reduce belief in truth too. We propose a defence that targets not the facts of each claim but the persuasion strategies that recur across them. Here we show that learning misinformation’s persuasion playbook, mapped from 214,646 fact-checked claims in 51 languages, can sharpen human and machine judgement of claims they know little about, without making either more cynical about true claims. The playbook spans social influence, information manipulation and emotional appeals, combined around a core of fear, conspiracy and fabrication. Briefed with it, plus a reminder that persuasion is not proof of falsehood, a language model better distinguished true from machine-generated false claims about events after its training data. In a pre-registered experiment, the same briefing improved people’s discernment of claims they reported as unfamiliar and left their belief in true claims unchanged. Averaged over all claims, the gain was not statistically significant. Because it targets what recurs, this defence need not wait for claim-specific fact-checks.
[HC-19] Risk and Refusal: Mechanisms of LLM Responses to Suicidal Ideation
链接: https://arxiv.org/abs/2610.22715
作者: Drishti Goel,Soorya Ram Shimgekar,Jeongah Lee,Violeta J. Rodriguez,Daniel S. Brown,Ravi Karkar,Koustuv Saha
类目: Human-Computer Interaction (cs.HC); Computation and Language (cs.CL); Computers and Society (cs.CY); Social and Information Networks (cs.SI)
备注:
Abstract:Refusal is a common safety mechanism in large language models (LLMs), yet in contexts of suicidal ideation (SI), refusing to engage can itself constitute a safety failure. We evaluate how eight instruction-tuned LLMs respond to SI and self-harm disclosures across three datasets, varying whether models are informed that a disclosure expresses SI and whether they receive a clinician-developed response protocol, and complement this with mechanistic analyses in six open-weight models. Our results reveal a nuanced relationship between refusal and response-level risk. At baseline, elevated risk in model responses is driven primarily by inattention to the disclosure, alongside refusal to engage. Mechanistic analyses suggest that models may refuse SI disclosures through the same direction that mediates their refusals of harmful requests. However, informing models of the SI context, or suppressing refusal directly, increases engagement without reliably lowering response risk, and may raise it. In contrast, the clinician-guided protocol consistently reduces response-level risk while also reducing refusal. Together, these analyses and ablations motivate the need for models that refuse only when warranted and engage in interactionally safe ways when users disclose SI.
[HC-20] Author-in-the-loop Agent ic Reproducibility Auditing for a Machine Learning Conference
链接: https://arxiv.org/abs/2610.22465
作者: Yukun Dong,Linjun Zhang,Tian Zheng
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注: 23 pages, 8 figures, 4 tables
Abstract:Reproducibility checks during peer review require a shared understanding of what authors’ code is expected to reproduce. We present an author-in-the-loop collaborative audit that makes author declarations explicit inputs to an AI agent’s investigation and assessment. A structured checklist records reproduction instructions and limitations; a two-stage workflow on GitHub connects agent checks to author clarification, revision, and agent reassessment. The agent separately assesses whether results have been reproduced and whether authors need to provide clarification or make repairs, while human reviewers approve requests and decide whether concerns are resolved. We deployed the system across 36 conference papers and 133 experimental tables and figures, raising 20 issues after human review. Further checks supported closing all the issues, although full reproduction remained unverified in some cases. We further compared repeated audits with complete and blank checklists to examine how author declarations shaped the agent’s investigation and assessment. Our work contributes a collaborative workflow and agent design that use author knowledge to guide reproducibility auditing. The conference deployment provides empirical evidence that this approach can help identify and address reproducibility problems through collaboration among authors, agents, and reviewers.
[HC-21] MultiPASE: Shape-Aware Comparative 3D Genome Analysis
链接: https://arxiv.org/abs/2610.22786
作者: N. Nipu,G. Johnson,C. Roth,D. Rogers,S. R. Starkenburg,C. R. Steadman,G. E. Marai
类目: Quantitative Methods (q-bio.QM); Human-Computer Interaction (cs.HC)
备注:
Abstract:3D genome structures reconstructed from Hi-C data provide valuable insights into chromatin organization, but comparing them across conditions and time points remains challenging because of noise, variability, and arbitrary orientation. State-of-the-art comparison approaches often rely on contact-matrix-based summaries, which do not preserve the spatial geometry of the reconstructed structures and make integration with gene expression and chromatin accessibility difficult. There is therefore a need for robust structural alignment and spatial abstraction methods that support integrated analysis across conditions and time points. We present MultiPASE, a computational pipeline for comparative analysis of reconstructed 3D genome structures represented as gene-level spatial point clouds. MultiPASE aligns point clouds into a shared coordinate frame using coarse-to-fine rigid registration with PCA-based axis disambiguation and robust ICP refinement, then summarizes spatial organization using two complementary shape abstraction methods that produce stable, nested, multi-scale representations from dense core regions to broader structural extent. Using Hi-C-derived reconstructions from viral infection experiments, we show how the approach reveals condition-dependent structural changes that are difficult to interpret from raw 3D structures alone. By integrating spatial structure with gene expression and chromatin accessibility, MultiPASE supports interactive analysis of spatial organization and gene relocation patterns. A quantitative validation, reported in the Supplementary Materials, shows that the proposed shape abstractions preserve non-convex features, remain stable across samples, and that the alignment pipeline is necessary for meaningful comparison. A usage case and a usability evaluation with eight participants demonstrate the pipeline’s effectiveness for comparative structural analysis.
[HC-22] Harness-Driven Agent ic Operation of Optical Pooled Screen Analysis NEURIPS2026
链接: https://arxiv.org/abs/2610.22617
作者: Matteo Di Bernardo,Claire Peterson,Ege Topkoc,Matt Might,Iain M. Cheeseman
类目: Genomics (q-bio.GN); Human-Computer Interaction (cs.HC); Quantitative Methods (q-bio.QM)
备注: Accepted at the NeurIPS 2026 Workshop on Agentic Life Sciences
Abstract:Optical pooled screens connect genetic perturbations to changes in cell morphology and behavior, but turning the microscopy data into interpretable results takes complex, multi-day analyses. Established pipelines process this data, yet researchers still select parameters, inspect outputs, and resolve failures for each experiment. Here, we present Brieflow-auto, a pipeline-specific harness implemented as a Claude Code plugin that lets a coding large language model (LLM) operate Brieflow, an open-source end-to-end pipeline for optical pooled screening data. A researcher supplies experimental facts and approves the launch. An agent tunes parameters in an interactive notebook, and the harness validates the configuration, monitors execution, recovers from common failures, and checks output quality. For pipeline defects, the agent opens upstream issues proposing backward-compatible fixes for the broader community. In five previously analyzed screens, harness runs yielded more cells with a confident single-gene assignment than the original hand analyses. On a screen the pipeline was not designed for, the harness turned each failure into a reviewed upstream fix. More broadly, we propose that wrapping an established pipeline in a harness can substantially accelerate its use and the long-term reliability of its codebase.
计算机视觉
[CV-0] Learning to Correct Geometry in Generated Videos NEURIPS2026
链接: https://arxiv.org/abs/2610.23856
作者: Weijie Lyu,Tiancheng Shen,Jason Li,Yujing Wang,Xiaoyuan Wang,Haobo Yuan,Yizhou Zhao,Ming-Hsuan Yang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by NeurIPS 2026. Project page: this https URL
Abstract:Generated videos often lose geometric consistency under camera motion: structures warp under viewpoint change, parallax becomes inconsistent, and smooth camera movement may collapse into shot-like transitions. We argue that this failure is largely a data problem. In web-scale video corpora, most visible motion comes from dynamic objects captured by static or weakly moving cameras, leaving only sparse supervision for viewpoint-induced appearance changes. To address this, we propose Video Geometry Corrector (VGC), a post-hoc video-to-video model trained on synthetic paired supervision. Starting from real videos with diverse camera motion and pose annotations, we construct VG-Pairs with a synthetic paired-supervision data generation pipeline. Each video clip is treated as a geometry-consistent target and paired with a distorted counterpart synthesized by a pretrained image-to-video inpainting model using trajectory-adaptive keyframes. VGC learns to map the distorted input to the geometry-consistent target without modifying the source generator, and uses optical flow as an auxiliary cue to help disentangle camera motion from object motion. Experiments on pose-conditioned and prompt-driven benchmarks show that VGC improves geometric consistency across multiple source generators, and scaling to a larger backbone further extends the framework toward more general video refinement.
[CV-1] Rethinking the design space of Vanilla Pixel Diffusion Transformers
链接: https://arxiv.org/abs/2610.23854
作者: Soumik Mukhopadhyay,Long Mai,Abhinav Shrivastava,Aniruddha Mahapatra
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Project webpage: this https URL . 33 pages, 23 figures, 15 tables. Submission under review
Abstract:Pixel-space diffusion models have recently re-emerged as a compelling alternative to latent diffusion, avoiding the lossy latent bottleneck and separately trained autoencoder. JiT shows that a vanilla diffusion transformer can generate directly in pixel space by replacing the commonly used velocity prediction ( \boldsymbolu -prediction) with clean image prediction ( \boldsymbolx_0 -prediction). In this work, we revisit this choice and introduce MFlow, an adaptive parameterization that learns to interpolate between \boldsymbolx_0 - and \boldsymbolu -prediction across principal directions of the data, learned end-to-end with diffusion training. We find that the preferred parameterization varies systematically with the data representation and model capacity – notably, increasing model capacity shifts preference toward velocity prediction. We evaluate MFlow across ImageNet, UCF-101 video, and large-scale text-to-image generation, demonstrating competitive or improved generative performance across diverse settings. In addition, we find that vanilla pixel diffusion transformers, irrespective of parameterization, retain a persistent high-frequency deficit relative to real images. To address this, we introduce ConvRefiner, a lightweight convolutional module trained post-hoc with an adversarial loss while keeping the diffusion backbone frozen. With only a small amount of additional training, ConvRefiner restores the high-frequency statistics of generated images while preserving FID. Project webpage: this https URL.
[CV-2] StopWatch: Goal Grounding for Object-goal Navigation ATC
链接: https://arxiv.org/abs/2610.23841
作者: Xinhao Liu,Yunsheng Tian,Yuheng Qiu,Pieter Abbeel,Xi Chen,Guanya Shi,Chen Feng
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: Project website this https URL
Abstract:Pretrained VLMs bring rich world knowledge and semantic priors to object-goal navigation, and fine-tuning a VLM into a navigation policy has become a common recipe. Despite this progress, success rates remain low, and it is unclear which capability these fine-tuned policies lack. Our diagnosis identifies goal grounding as the dominant failure. These policies explore well, yet often fail to notice that the goal is already in view or to estimate how far away it is. To address this failure, we ground the goal explicitly. Explicit grounding substantially improves success and reaches state-of-the-art results on six benchmarks across three scene collections. We also correct annotation defects in these benchmarks, so that future methods are evaluated more faithfully. Our analysis on the clean benchmarks shows that, with ground-truth grounding, our policy nearly matches the expert that generated its demonstrations, leaving the headroom in perception rather than policy learning. Our work points to goal grounding as a missing piece of object-goal navigation and offers a simple and effective solution to the task.
[CV-3] Persistent Planar Memory for Video World Models NEURIPS2026
链接: https://arxiv.org/abs/2610.23795
作者: Yuze He,Bin Tan,Zelin Gao,Yujun Shen,Yong-Jin Liu,Nan Xue,Yinghao Xu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 20 pages, 9 figures. Accepted by NeurIPS 2026
Abstract:Video world models that generate and explore environments at scale need a persistent 3D memory of what they have produced. Plane primitives offer a natural candidate: a compact set of oriented planes captures the dominant structural elements (e.g., walls, floors, facades) of indoor and outdoor scenes, while requiring only 7 parameters per primitive. We present PPMem, a video world model that adopts plane primitives as its persistent 3D memory. PPMem augments a video diffusion transformer with a planar prediction head and a Depth Token Fusion (DTF) module: each autoregressive chunk jointly produces RGB video and planar geometry, which merges into a persistent plane memory that conditions subsequent chunks via rendered depth. The resulting memory is three orders of magnitude more compact than per-pixel point clouds (~100K primitives, 5MB for a full scene), geometrically explicit (metric depth, normals, exportable surfaces), cross-view consistent (jointly predicted by the shared backbone), and adaptive (refined by the merging procedure as generation proceeds). On DL3DV and RealEstate10K, PPMem outperforms prior 3D-grounded world models in video quality while jointly producing metric-scale depth, surface normals, and exportable planar geometry, sustaining coherence across hundreds of frames under large viewpoint changes. Project page: this https URL.
[CV-4] he Collection of a Real-World Large-Scale Dataset for Automated Skill Evaluation in Central Line Dressing Change
链接: https://arxiv.org/abs/2610.23789
作者: Ummey Tanin,Shannan K. Hamlin,Hsin-Mei Chen,Peizhu(Pam)Qian
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 9 pages
Abstract:Central line-associated bloodstream infections (CLABSIs) remain a major source of preventable morbidity and mortality in healthcare settings. Proper execution of the central line dressing change (CLDC) procedure is essential for infection prevention, yet current approaches to competency assessment rely primarily on manual ob- servation by clinical educators, making evaluation time-intensive, difficult to scale, and limited in consistency and granularity. These challenges highlight the need for AI-assisted skill evaluation methods that can support more objective, reproducible, and scalable assessment of clinical performance. However, progress in this area has been constrained by the lack of broad-scale, clinically grounded datasets for developing and evaluating such methods. In this work, we present the collection process of a large-scale video dataset for CLDC skill evaluation, comprising 365 recordings of registered nurses collected during the 2026 annual competency exam at Houston Methodist Hospital. Each video is paired with subtask-level temporal annotations and clinical performance assessments spanning 17 subtasks of the procedure. We describe the data collection and curation pipeline, including pilot testing of record- ing configurations, annotation protocol design, and post-collection processing. We further establish the feasibility of video-based skill assessment through benchmark experiments with spatiotemporal deep learning models. This dataset provides a foundation for future research on automated clinical skill assessment, feedback generation, and AI-assisted training in nursing education.
[CV-5] Video Diffusion Avatars: Taming Video Diffusion Models into Controllable 3D Avatars
链接: https://arxiv.org/abs/2610.23772
作者: Simon Giebenhain,Tobias Kirschstein,Liam Schoneveld,Davide Davoli,Zhe Chen,Matthias Nießner
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project webiste: this https URL
Abstract:Video diffusion models (VDMs) trained on internet-scale data encode a rich prior on appearance, motion, and physics that is difficult to replicate with explicit 3D rendering and costly 3D data capture. We introduce Video Diffusion Avatars (VDAs), a method that creates photo-realistic, multi-view consistent avatars from a single image by fine-tuning a VDM for precise control over camera viewpoint, facial expression, and upper-body pose. Departing from 3D morphable models (3DMMs), VDAs use a learned expression space that lifts the expressiveness ceiling of 3DMMs, e.g., for the tongue and mouth interior. However, learned expression codes exhibit heavy viewpoint bias, as our DISTRACT benchmark reveals. We show that this bias can be removed through disentanglement, enabling their use as driving signals for multi-view consistent avatars. Camera and upper-body pose are controlled jointly via proxy-mesh renderings, giving independent control over expression, pose, and viewpoint. A camera-triplet training strategy makes the approach efficient: our models train for 34 hours on a single A100 using only public multi-view data. VDAs outperform state-of-the-art single-image 3D avatars on novel-view reenactment, reducing face LPIPS by 29% and improving perceptual quality (JOD) by 66% on BecomingLit. A static 3DGS reconstruction probe further confirms that our generations are the most multi-view consistent. Built on a video prior, VDAs produce effects beyond the reach of explicit renderers, including physically plausible hair dynamics, shadows, reflections, and refractions.
[CV-6] A Paired RGB-Sonar Dataset for Fish Monitoring with Consumer Sonar
链接: https://arxiv.org/abs/2610.23759
作者: Mourad Derbedrosian,Michael Hobley,Pietro Perona
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Imaging sonar enables underwater observation in conditions where optical imaging is ineffective due to darkness and turbid water. However, imaging-sonar systems previously used for ecological monitoring have typically been expensive, limiting their accessibility. Consumer forward-looking sonar has the potential to make fish monitoring more accessible, but its noisy imagery makes annotation and the development of automated methods challenging. We introduce the Huntington and Caltech Koi RGB and Sonar dataset (HaCKRS), which pairs Garmin LiveScope imaging-sonar recordings with top-down RGB video of koi carp. The dataset contains more than 105,000 temporally matched frame pairs, spanning approximately 3.3 hours of recordings across four range and orientation configurations. We provide frame-wise metadata, sonar geometry and coordinate information, synchronization data for all four configurations, and planar registration for the three horizontal configurations. The release focuses on paired observations and calibration rather than task-specific fish annotations. We intend HaCKRS to support annotation and the development of fish detection, tracking, and counting methods, and investigations into the potential of consumer sonar for ecological monitoring.
[CV-7] MTEdit: Packed Trajectories Beat Rollouts for Multi-Turn Image Editing
链接: https://arxiv.org/abs/2610.23751
作者: Houyuan Chen,Fanzhou Wang,Xiangyu Fan,Songlin Yang,Zhitao Yang,Lvmin Zhang,Pei Yang,Hao Zhao,Lei Yang,Anyi Rao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 41 pages, 22 figures, 8 tables
Abstract:Multi-turn image editing aims to modify an image according to a sequence of instructions, making it better aligned with human-AI collaborative creation than single-turn editing. However, constructing high-quality multi-turn training data remains challenging, as conventional sequential rollout accumulates errors and introduces visual-quality drift across editing turns. To address this challenge, we introduce MTEdit, a multi-turn image editing dataset built through a pipeline centered on packed trajectory canvas (PTC) generation. By jointly generating multiple editing states within a unified visual context, this pipeline reduces visual-quality drift across turns. Covering both step-wise and context-aware editing scenarios, MTEdit comprises 500K multi-turn image editing trajectories, with up to 8 turns and an average of 5.38 turns per trajectory, making it one of the most extensive datasets for multi-turn image editing. We further introduce MTEdit-Bench, a benchmark designed to evaluate both scenarios by assessing models’ ability to follow editing instructions, maintain content consistency, resist visual-quality drift, and effectively use historical context across turns. Using MTEdit, we fine-tune Qwen-Image-Edit-Causal to obtain Qwen-Image-MTEdit. On MTEdit-Bench, we evaluate a broad range of closed-source and open-source image editing models, including Qwen-Image-MTEdit. Experimental results show that Qwen-Image-MTEdit achieves the best performance among the evaluated open-source models, demonstrating the effectiveness of MTEdit in improving models’ multi-turn image editing capabilities.
[CV-8] Beyond Multi-Stage Pipelines: A Label-Correlation Ensemble for Multi-Label Retinal Fundus Disease Classification
链接: https://arxiv.org/abs/2610.23747
作者: Md Asif Uddin,Jannatul Ferdous,Asif Islam,Zobaer Mahmud Zisan,Promit Ghosh,Shah Ahafik Arman,Niloy Farhan,Rafeed Rahman
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Retinal fundus images frequently carry more than one disease, and public fundus datasets are heterogeneous and severely long-tailed. Clinical triage suggests a staged design, in which a screening gate, a disease-group router and group-specific classifiers operate in sequence, but it is unclear whether this hierarchy earns its complexity. We present a controlled comparison of three architectures on a unified 39-target label space built from RFMiD 1.0, RFMiD 2.0 and ODIR-5K (12,652 images): (i) a three-stage gated cascade of eight models, (ii) a flat model with 39 independent sigmoid outputs, and (iii) a label-correlation model that replaces the flat linear head with 39 learned label queries and a two-layer transformer decoder. All three share one Swin-V2-B backbone, one preprocessing pipeline and one training protocol, so differences are attributable to the architecture. Staging offers no consistent advantage: tracing every missed disease instance to the stage that lost it shows that the two coarse gates discard 34.2% of true instances while the fine-grained classifiers lose only 6.7%, and the gates’ hard zeros depress the cascade’s macro-AUC to 0.747. Label-correlation modeling gives the largest single gain, raising the detection rate from 0.621 to 0.682 (0.689 with test-time augmentation). Blending the flat and label-correlation models under four-view test-time augmentation is the best configuration on six of eight metrics (exact match 0.694, micro-F1 0.678, macro-F1 0.456, macro-AUC 0.972). On definition-matched class subsets it matches or exceeds four recent studies on macro- and micro-F1, most clearly on low-support pathologies. We also show that ensembling a gated cascade with continuous-output models is harmful, and that per-source negative sampling is needed to remove a dataset shortcut that otherwise yields a spurious AUC of 1.000.
[CV-9] Material and Defocus Cues for Occluded Object Localization in Single Hyperspectral Images: An Exploratory Study
链接: https://arxiv.org/abs/2610.23733
作者: Ali Zia,Jun Zhou
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Partial occlusion makes object detection difficult when visible regions have weak texture or similar appearance. Hyperspectral imaging provides material-dependent spectra and wavelength-dependent optical blur, which offer complementary evidence for separating objects. We investigate a pipeline that combines nonnegative-matrix-factorization-based material grouping, divergence-based optical cues, and graph Laplacian propagation. The propagated scores support region extraction and bounding-box localization without requiring a separate depth sensor at inference. A collection of 30 indoor and six outdoor hyperspectral scenes acquired using two camera systems provides qualitative examples spanning different materials, depth arrangements, and partial occlusions. The examples illustrate how material information can support separation when optical cues are weak, and how optical cues can distinguish some surfaces with similar material properties. Objects sharing both material and depth remain challenging. This exploratory study establishes the sensing rationale and mathematical formulation, while identifying the quantitative evaluation and implementation details needed to assess detection performance systematically.
[CV-10] ESC: Escaping 2D Diffusion Ambiguity with Geometric Constraints for Consistent Novel View Synthesis
链接: https://arxiv.org/abs/2610.23732
作者: Ge Ya Luo,Daniel Severo,Brian Karrer,Christopher Pal,Adriana Romero-Soriano,Michal Drozdzal
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Generating 3D-consistent novel views from 2D diffusion models remains a fundamental challenge in computer vision. In this work, we introduce a framework that enhances geometric alignment for monocular view synthesis under drastic camera changes by jointly generating a target view and its corresponding depth map. By conditioning the diffusion pipeline on the single source image, source depth, and target camera rays, our approach forces the model to inherently reason in 3D space even under extreme viewpoint shifts. To guarantee spatial fidelity, we propose a novel per-pixel depth consistency constraint, which verifies and enforces 3D consistency across generated views. Our framework establishes a robust pipeline for geometrically consistent view synthesis, yielding superior visual accuracy and spatial alignment compared to current state-of-the-art methods. With this work, we pave the road towards robust 3D world models capable of understanding complex environments from a single view image.
[CV-11] Synthetic Image Oversampling for Rare-Class Representation Learning in Deep Facial Phenotyping
链接: https://arxiv.org/abs/2610.23730
作者: Miles Liao,Zhanliang Wang,Hongzhuo Chen,Kai Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: BIBM 2026 ML4BMI Workshop
Abstract:Recent deep facial phenotyping methods can support the recognition of genetic syndromes, but patient images are often scarce and unevenly distributed across disorders. Synthetic facial images offer a way to supplement the training data without collecting additional patient photographs. Their usefulness, however, depends on whether they improve recognition of real patients: generated images may introduce artifacts or repeat features already represented in the training set. It is therefore necessary to examine both how much synthetic augmentation to use and which disorders benefit. We investigate disease-specific synthetic oversampling using images from the Phenotypic Disease Image Database to augment a subset of the GestaltMatcher Database. We compare increasing augmentation levels in an image-based model, GestaltMatcher-Arc, and a multimodal model, GestaltMML, while keeping the real training and patient-disjoint evaluation partitions fixed. Repeated training runs assess overall and disease-specific accuracy, and an exploratory embedding analysis examines changes in class compactness and separation. Augmentation improves mean recognition accuracy in both training setups, although gains remain modest in GestaltMatcher-Arc and continue across the tested levels in GestaltMML. These aggregate improvements are accompanied by uneven disease-level responses, including reduced accuracy for some disorders. In the embedding analysis, disorders with fewer real training images tend to gain more class separation, but geometric changes do not consistently track classification performance. Together, the findings support synthetic oversampling for limited-data facial phenotyping and emphasize the need to assess its amount and benefit for individual disorders within each training setup.
[CV-12] FedRAPP: Federated Test-Time Prompt Tuning for Vision-Language Models with Reliability-Aware Prompt Personalization
链接: https://arxiv.org/abs/2610.23705
作者: Syed Md. Ahnaf Hasan,Md Akil Raihan Iftee,Sajib Mistry,Amin Ahsan Ali,AKM Mahbubur Rahman
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 15 pages, 5 figures and 10 tables
Abstract:Federated prompt learning enables vision-language models like CLIP to collaboratively adapt lightweight soft prompts across distributed clients without sharing raw data. However, existing federated prompt learning methods are designed primarily for supervised settings with labeled client data and assumed relatively stable training distributions. At test time, source-free federated prompt tuning remains largely unexplored. Clients may receive unlabeled target samples with heterogeneous, often non-IID distributions that differ substantially across clients and from the source training distribution, causing locally adapted prompts to drift in different directions and reducing the effectiveness of federated aggregation. In this work, we explore Source-Free Federated Test-Time Prompt Tuning (SF-FTTPT), where distributed clients collaboratively adapt soft prompts on unlabeled test streams without access to source data. Our study reveals that conventional federated aggregation methods struggle under heterogeneous client distributions, as locally adapted prompts become increasingly specialized and difficult to combine effectively. To address this challenge, we propose FedRAPP, a personalized aggregation framework that combines reliability-aware client assessment, server-side exponential moving-average prompt memory, and similarity-guided peer aggregation. FedRAPP selectively incorporates information from reliable and compatible clients while preserving local specialization, enabling more robust adaptation under distribution shift. Extensive experiments across nine CLIP benchmark datasets and ImageNet variants demonstrate that FedRAPP consistently outperforms standard federated aggregation baselines under heterogeneous client distributions while maintaining the efficiency advantages of prompt-only adaptation.
[CV-13] Agent ic Physics-Grounded Video Generation via Memory-Augmented In-Context Reasoning
链接: https://arxiv.org/abs/2610.23677
作者: Xikang Guan,Shaocong Yang,Yanqi Bao,Can Wang,Zheng Gu,Zhengqing Chen,Kai Chen,Jing Huo,Yang Gao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 12 pages, 5 figures, 1 table
Abstract:Physical plausibility is essential for realistic video generation. Existing methods impose physical guidance before synthesis via simulations or refined prompts, but this can limit dynamic diversity, miss errors that emerge only after generation, and require repeated regeneration when inconsistencies remain. We therefore study a generate-diagnose-repair paradigm that preserves the original generation capability and selectively repairs physical inconsistencies afterward, raising two key questions: what to repair and how to repair? To address them, we introduce PhysRepair, an agentic video generation framework that combines causal reasoning over the observed physical failure with prior repair experience to enable localized and adaptive correction. Specifically, Spatiotemporal Causal Constraints identify the first invalid physical state and its causally affected states, producing a repair specification that guides physically plausible local regeneration while preserving unaffected content. Experience-augmented In-context Learning combines the ongoing repair context with relevant experience from previous repairs, allowing the agent to adapt its repair decisions across diverse physical failures and accumulate completed trajectories for future cases. Experiments on Physics-IQ Verified across solid mechanics, fluid dynamics, optics, thermodynamics, and magnetism demonstrate that PhysRepair consistently improves the physical plausibility of generated videos over existing methods. Our project page is available at this https URL.
[CV-14] PCDFusion: Proposal-Context-Detail Bayesian Rendering for Infrared-Visible Image Fusion NIPS2026
链接: https://arxiv.org/abs/2610.23663
作者: Rui Liu,Lin Gu,Hesong Li,Ying Fu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by NIPS 2026
Abstract:Infrared-visible image fusion (IVIF) aims to combine visible and thermal infrared observations for human perception and downstream vision tasks. Visible and thermal infrared observations are physically complementary but unreliable in different ways. Existing deep IVIF methods mainly improve feature interaction and reconstruction, but often lack an explicit local rule for deciding when visible appearance should dominate and when thermal responses should contribute. To address this issue, we reformulate IVIF as local Bayesian posterior rendering, where fusion is controlled by posterior action weights rather than direct modality mixing. In this paper, we propose Proposal-Context-Detail Fusion (PCDFusion), a visible-anchored framework that implements this formulation through three rendering stages. PCDFusion preserves potentially useful thermal responses, updates exposure-degraded visible regions, and renders reliable band-limited infrared structures into the final luminance. Extensive experiments on public IVIF benchmarks show that PCDFusion improves fusion quality, target recovery, and downstream detection and segmentation performance under challenging illumination. The code is available at this https URL.
[CV-15] An Efficient Approach to Beetle Damage Detection of Banana Using Deep Learning
链接: https://arxiv.org/abs/2610.23661
作者: Md. Mehedi Hasan Naeem,Ishtiyak Ahmmad Araf,Md. Arefin Haque Mahir,Tushar Kanti Saha
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 6 pages. Accepted at the 12th IEEE International Conference on Sustainable Technologies and Engineering (IEEE iCOSTE 2026), Cox’s Bazar, Bangladesh, November 1-3, 2026
Abstract:Visible surface damage of banana fruits by Basilepta subcostata (the banana fruit scarring beetle) can reduce their market value. Here, manual inspection of Banana fruits is subjective and difficult to scale. This study presents a lightweight framework based on YOLOv8n deep learning model for locating individual bananas and classifying their visible surface condition as healthy or unhealthy. The term beetle damage in the paper title is used operationally for the locally observed scar-like surface condition; the model does not identify a specific beetle species or biological cause. A custom dataset of 2,473 images containing 5,827 annotated banana instances was prepared from local-market imagery in Bangladesh. The reported experiment used an 80/20 protocol with 1,982 training images and 491 test images. A pretrained YOLOv8n model was trained at 640x640 resolution for 50 epochs with a batch size of 4. On the test partition, it achieved 81.36% precision, 74.30% recall, 77.67% F1-score, 87.04% mAP@0.50 and 59.33% mAP@0.50:0.95. A YOLOv8s comparison gave slightly lower F1 and mAP while using a substantially larger checkpoint. The YOLOv8n checkpoint occupied 5.96 MB and required 6.54 ms/image for inference on an NVIDIA Tesla T4. The results support YOLOv8n as a compact baseline for instance-level banana surface inspection within the studied acquisition conditions.
[CV-16] FlexPart: Text-Controlled Semantics-Aware 3D Part Segmentation at Flexible Granularity
链接: https://arxiv.org/abs/2610.23659
作者: Yiqun Zhao,Jiachen Xu,Yihao Zhi,Yunfei Zhao,Zibo Zhao,Xintong Han,Chunchao Guo,Shenghua Gao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages main, 8 figures
Abstract:A 3D asset admits multiple valid part decompositions, while semantics-aware segmentation should reflect both the desired granularity and the parts a user wants to extract. Ideally, a practical segmentation model should strive for a simple principle: what you ask for is what you get. However, realizing this goal is hindered by two core obstacles: the lack of multi-granularity training data and the absence of a holistic control mechanism. We propose FLEXPART, a unified framework for text-controlled 3D mesh segmentation. To establish flexible control, we introduce a holistic control prompt comprising semantic part names with location and shape clauses. To support this mechanism, we utilize an automated Vision-Language Model (VLM) pipeline to construct a large-scale dataset of multi-granularity partitions and their corresponding prompts. Driven by this data, FLEXPART generates a view-independent part-identity field in native 3D space, directly assigning discrete part IDs to mesh faces. To compel the model to respect semantic conditions rather than rely on geometric priors, we train the model using randomized background dropping. Furthermore, an inference-time VLM planner can automatically complete descriptive clauses from 2D renderings, allowing users to simply name the desired parts in ordinary words. FLEXPART with planner-completed prompts outperforms the strongest evaluated baseline by 15% in mIoU, even when that baseline receives ground-truth guidance. On our held-out FLEXPARTBENCH, FLEXPART leads at each granularity and improves average mIoU by 22% over the strongest baseline. With only the requested part names supplied by the user, it also improves IoU by 35% over the strongest click-driven baseline on selective segmentation. Comments: 8 pages main, 8 figures Subjects: Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2610.23659 [cs.CV] (or arXiv:2610.23659v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2610.23659 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Yiqun Zhao [view email] [v1] Fri, 16 Oct 2026 15:54:52 UTC (29,237 KB) Full-text links: Access Paper: View a PDF of the paper titled FlexPart: Text-Controlled Semantics-Aware 3D Part Segmentation at Flexible Granularity, by Yiqun Zhao and 7 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-17] A Language-Informed Loss for Visual Generation
链接: https://arxiv.org/abs/2610.23653
作者: Suchir Agarwal,Keshigeyan Chandrasegaran,Ziyu Chen,Kyle Sargent,Juan Carlos Niebles,Li Fei-Fei,Jiajun Wu
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 29 pages; Project website: this https URL
Abstract:End-to-end visual generation with pixel-space loss has recently achieved competitive performance without the need for learned image tokenizers. In this work, we study whether language-informed supervision can provide a richer training signal than pixel loss for visual generation. We introduce LILIE (pronounced “lily”), a Language-Informed and Lossless Image Encoder. LILIE is an invertible image transform trained such that a vision-language model can recover captions from noised features. The encoder uses a normalizing flow transformer trained jointly with a vision-language model via next token prediction. We then use the encoder to extract a simple language-informed loss for text-to-image generation. We train JiT models with LILIE loss, reducing gFD by 14 points (66.84 \to 53.01) against a pixel-only baseline on the GPIC dataset. Our code, encoders, and generative models are available at this https URL.
[CV-18] Diversity Without Labels: Task-Agnostic Coreset Selection Across Perception and Action
链接: https://arxiv.org/abs/2610.23642
作者: Julia Hindel,Benjamin Ramtoula,Kyuhwan Yeon,Nick Heppert,Paul Newman,Daniele De Martini,Abhinav Valada
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:While data collection for model training is relatively easy, curating high-quality datasets from large collections is inherently challenging. Coreset estimation methods address this problem by identifying the best data subsets to train models on. However, existing approaches often require proxy model training, extensive annotations, or task-specific tailoring. Unsupervised alternatives typically exploit geometric properties in foundation model feature spaces, but are limited to pairwise image-to-image comparisons or comparisons against a fixed subset. We introduce CoreDNA, an unsupervised, task-agnostic coreset estimation approach that pools neuron activations from vision foundation models into per-image and per-dataset distributions. By comparing each image’s distribution to the dataset-level distribution, CoreDNA can capture both low-level visual structure and high-level semantic information across layers. We find that high-quality samples concentrate in the tails of the last layer’s activation statistics, which we validate by relating them to training-based coreset methods. Building on these findings, we develop a computationally efficient selection mechanism to construct task-agnostic, high-quality coresets for selective annotation and model training. Across image classification, semantic segmentation, depth estimation, and imitation learning for end-to-end driving and mobile manipulation, CoreDNA matches or outperforms the state of the art, demonstrating unprecedented generalization across perception and action tasks. These results highlight the effectiveness of compact activation-based representations for data selection.
[CV-19] Stereo LWIR-RGB Automatic Labeling for Low-Light Camera-Based VRU Detection
链接: https://arxiv.org/abs/2610.23621
作者: Dimitrios Bouzoulas,Pyry Jäppinen,Pejman Habibiroudkenar,Eerik Alamikkotervo,Risto Ojala
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:VRU detection in low-light conditions is a key component in ensuring VRU safety during the most collision prone time of the day. Currently, advances in low-light detection are limited due to the lack of large scale datasets and automated labeling methods. Most autolabeling methods use RGB detections as their labels source, which cannot be replicated in low-light conditions. To fill this gap in this paper we present an automated pipeline for low-light VRU labeling. We use LWIR-RGB image pairs and source our labels from LWIR as VRUs differentiate from the background due to their body temperature. Unlike our previous work that relied on a complex hardware setup to align the two modalities, we perform this experiment using a LWIR-RGB stereo setup, lowering hardware complexity and cost, making the pipeline easier to reproduce. We evaluate method end-to-end on a dataset collected in the capital region of Finland. Models trained on our generated labels beat both off-the-shelf baselines and models fine tuned on a hand labeled subset on F1, mAP@50 and LAMR metrics when tested on a manually labeled dataset, and the result holds for two different LWIR detectors. This shows that low-light VRU autolabeling can be made both accessible and effective without specialized alignment hardware.
[CV-20] ADA 2025: A Multimodal Dataset for Water-Deficit Stress Detection in Sugar Beets
链接: https://arxiv.org/abs/2610.23609
作者: Piper Powell,Robert Franz,Amos Smith,Isaak Ihorst,Joachim Hertzberg,Martin Atzmueller
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 6 pages, 4 figures, 1 table. Accepted at the 2026 IEEE International Workshop on Metrology for Agriculture and Forestry (MetroAgriFor). Dataset: this https URL ; code: this https URL
Abstract:Research and development into artificial intelligence (AI) and machine learning (ML) applications in crop management are hindered by a lack of suitable public datasets. Whereas applications in disease and pest detection benefit from a comparatively higher amount of available data, applications like water-deficit stress detection often have very few if any public datasets available. Moreover, most datasets that are available are single-modality and frequently RGB-only, leaving the early detection and enhanced performance offered by other methods, and particularly by multimodal approaches, out of reach. Datasets with accurate metadata, higher numbers of plants, and smaller intervals between data points are also scarce. To address this, we publicly release ADA 2025, a 500 GB comprehensive multimodal dataset of young sugar beet plants under healthy and water-deficit conditions combining thermal, hyperspectral, RGB, and depth imagery along with relevant environmental and metadata and both relative water content (RWC) and leaf thickness as ground truth sources. Exemplary results showcase the dataset’s validity in terms of differences detectable between stressed and healthy plants.
[CV-21] Learning to Discriminate Scene Structures Makes Self-Supervised Depth Learning Scalable NEURIPS2026
链接: https://arxiv.org/abs/2610.23600
作者: Mengtan Zhang,Yuchuan Cui,Yu Ma,Zizhan Guo,Wei Ye,Rui Fan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to NeurIPS 2026
Abstract:Self-supervised monocular depth estimation is appealing for its potential to scale with unlabeled data, yet existing methods often degrade when trained on diverse scenes. This study finds that, under self-supervision, models tend to encode dataset-specific structural biases rather than transferable scene geometry, leading to cross-domain optimization conflicts. To address this issue, this study proposes the MaSS framework, aiming to make self-supervised depth learning scalable to diverse data. MaSS introduces a dedicated structural representation pathway explicitly decoupled from contextual features, which is guided by a novel scene-structure discrimination objective to learn generalized, layout-aware structural representations for depth decoding. Trained and evaluated on four diverse datasets, both individually and jointly, MaSS substantially outperforms prior self-supervised methods and is the first to consistently benefit from mixed-domain training. Zero-shot evaluation on eight additional datasets further demonstrates its remarkable generalizability, particularly on those entirely out-of-distribution scenes. The source code is publicly available at this https URL.
[CV-22] SemPurify: Semantics-Aware Data Purification for Black-box Backdoor Defense WACV2027
链接: https://arxiv.org/abs/2610.23589
作者: Jiawei Yu,Syed Irfan Ali Meerza,Yi Wu,Amir Sadovnik,Jian Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV 2027)
Abstract:Backdoor attacks have emerged as a critical security threat in machine learning, where adversaries embed triggers within poisoned training samples to manipulate a model’s predictions on targeted inputs while leaving its general behavior unaffected. This stealthy nature makes backdoor attacks difficult to detect, as the model functions normally until activated by a specific malicious trigger. Current purification-based defenses use image transformations (e.g., blurring) to disrupt trigger patterns but often degrade image quality. Even with advanced restoration techniques like diffusion models, these approaches can still alter visual or semantic information, reducing utility for downstream tasks. To address this, we propose the first semantics-aware data purification framework for black-box backdoor defense, SemPurify, which explicitly incorporates multi-level semantics (i.e., low-level semantic embeddings and high-level image tags) within a text-to-image (T2I) framework to guide image restoration during purification. Using this semantic information, the T2I model subsequently refines the transformed samples, helping preserve visual details and semantic information during restoration. Extensive experiments spanning multiple backdoor scenarios, i.e., classification models, CLIP-style encoders, and generative models, demonstrate that SemPurify reliably removes backdoors while effectively preserving semantic information for downstream tasks.
[CV-23] CLEAR: Cooperative Latent Flow for Visual Counterfactual Explanations with Accurate Region Preservation WACV2027
链接: https://arxiv.org/abs/2610.23577
作者: Yan Zeng,Changlu Guo,Oskar Kristoffersen,Anders Nymark Christensen,Morten Rieger Hannemose,Anders Bjorholm Dahl
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to WACV 2027
Abstract:Visual counterfactual explanations (VCEs) should alter a classifier’s prediction while remaining realistic and preserving decision-irrelevant content. Image-space methods provide high-quality counterfactuals but are computationally costly, whereas latent-space methods are more efficient but often less faithful. We identify two sources responsible for this gap. First, latent generation recovers a natural sample from a perturbed representation, whereas classifier guidance pushes it toward a target class, so their update directions may oppose. Second, VAE reconstruction can alter regions that should remain unchanged. We introduce CLEAR, a latent Flow Matching framework for VCEs. CLEAR uses Flow Matching’s explicit instantaneous velocity as a reference for the generative direction to decompose classifier guidance into cooperative and conflicting components. It preserves the cooperative component and adaptively attenuates only the conflicting one according to conflict strength. We further analyze conflict frequency and strength throughout generation to reveal how classifier guidance interacts with the generative process. CLEAR also dynamically identifies editable regions, constrains latent modifications, and directly preserves factual content outside them to reduce VAE-induced changes. Across five bidirectional tasks on two datasets, CLEAR retains latent-space efficiency while achieving counterfactual quality comparable to or better than strong image-space baselines, with a 2.5x speedup over the strongest baseline. Project page: this https URL
[CV-24] From Adult Renal Masses to Pediatric Wilms Tumors: Evaluating Cross-Domain Generalization of Kidney Lesion Segmentation on MRI
链接: https://arxiv.org/abs/2610.23568
作者: Sarah de Boer,Myrthe A. D. Buser,Hartmut Häntze,Jawed Nawabi,Lisa C. Adams,Alida F. W. van der Steeg,Marry M. van den Heuvel-Eibrink,Annemieke S. Littooij,Keno K. Bressem,Mathias Prokop,Alessa Hering
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Deep learning segmentation models for pediatric oncology are often trained on small, single-center cohorts, whereas much larger datasets exist for adult diseases, such as kidney cancer. Whether models trained on adult renal lesions generalize to pediatric Wilms tumors remains unclear. We trained a segmentation model on single-sequence T1- or T2-weighted adult kidney MRI and evaluated it on a held-out adult test set and a pediatric Wilms tumor cohort, where it was compared with a pediatric-specific nnU-Net. Segmentation performance was quantified using the Dice similarity coefficient (DSC) and 95th-percentile Hausdorff distance. On the adult test set, the model achieved a median DSC of 0.86 for renal lesions. On the pediatric cohort, the adult-trained model achieved a median DSC of 0.85 compared with 0.90 for the pediatric-specific model. Despite differences in age, anatomy, and tumor type, the adult-trained model showed promising performance on pediatric Wilms tumors, with only a modest difference in median DSC compared with the pediatric-specific model. However, non-inferiority could not be shown, highlighting the value of pediatric training data. In future work, a fine-tuning strategy could be explored.
[CV-25] LiveSync: Iterative Self-Training for Lip-Sync Generation
链接: https://arxiv.org/abs/2610.23558
作者: Huaishao Luo,Huan Zhang,Guanbo Bao,Yinglu Liu,Youzheng Wu,Xiaodong He
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Audio-driven lip synchronization aims to match target speech through localized mouth and jaw articulation while preserving subject identity, head pose, and surrounding facial context. Naturally recorded videos provide synchronized reconstruction targets but rarely offer pairs that are matched in identity and pose yet differ primarily in speech-related articulation. Mask-based training enables reconstruction from these videos, but random-reference inputs differ from the aligned self-reference used at inference. We present LiveSync, an iterative self-training framework in which a single mask-based model generates pseudo pairs and is refined for deployment. The model is first trained on a curated real-video corpus augmented with externally generated clips covering under-represented visual conditions. A frozen checkpoint then uses independently sampled audio to generate mouth-edited references from real-video frames, copying pixels outside the edit mask from each source frame. After confidence filtering, these references are paired with aligned masked inputs and original real-frame targets. The retained pseudo pairs are mixed with real training data to refine the same model from its current weights, updating training supervision without adding inference stages. Although we instantiate the framework with a single-pass GAN-based model, the self-training strategy could in principle also be adapted to diffusion-based models. Experimental results show that LiveSync achieves leading synchronization performance on HDTF and competitive performance compared with the baselines on the AIGC-LipSync subset, while maintaining low inference latency.
[CV-26] PanoBench: A Panoptic Occupancy Prediction Benchmark for Autonomous Driving
链接: https://arxiv.org/abs/2610.23537
作者: Yi Feng,Zizhan Guo,Junwu E,Yu Ma,Mengtan Zhang,Jin Wu,Wei Ye,Hanli Wang,Qijun Chen,Danwei Wang,Rui Fan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Panoptic occupancy unifies 3D geometry, semantics, and instance identities, but its development is limited by scarce instance-aware annotations. Existing benchmarks mainly provide semantic labels reconstructed from incomplete sensor observations, while panoptic datasets have limited domain coverage or annotation completeness. We introduce PanoBench, a camera-centric panoptic occupancy prediction benchmark spanning synthetic and real-world driving scenes. It combines PanoBench-CarlaOcc, PanoBench-nuScenes, and PanoBench-KITTI360 through canonical data interfaces and unified evaluation protocols, providing multi-resolution annotations with persistent instance identities. The synthetic branch derives complete panoptic geometry from simulated meshes. For real-world data, a new pipeline reconstructs backgrounds using geometric priors and topology constraints, generates foreground mesh assets from multi-frame multi-view images, and fuses them into sparse panoptic voxels. A unified valid-mask protocol evaluates complete thing instances supported by temporal observations while scoring stuff regions at the voxel level. Analyses of observation fidelity, geometric completeness, spatial continuity, and temporal consistency demonstrate the benefits of the resulting annotations over existing occupancy benchmarks. We benchmark representative methods on semantic occupancy prediction, panoptic occupancy prediction, and 4D occupancy forecasting, and study voxel resolution, synthetic-to-real transfer, and storage efficiency. These results reveal complementary strengths of volumetric and ray-based metrics and persistent challenges in real-world instance-aware prediction. PanoBench is available at this https URL.
[CV-27] PreciseEdit: Strong Yet Stable Semantic Transport for One-Step Image Editing
链接: https://arxiv.org/abs/2610.23535
作者: Jingyi Wang,Xiao Lu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 9 pages, 7 figures, and 5 tables
Abstract:Recent one-step text-to-image (T2I) models enable efficient image synthesis, opening up new possibilities for interactive Web-based image editing. However, existing one-step editing methods often construct a global transport field through simple vector operations between conditional predictions. While such strong transport can effectively drive the image toward the target semantics, it may be released along unreliable directions or over non-edited regions, leading to structural distortions within the editing region and background drift elsewhere. More importantly, stabilizing the editing process through global smoothing or uniformly weakening the transport field often makes large semantic changes overly conservative. In this work, we revisit one-step image editing from the perspective of the tension between semantic strength and transport stability, and identify a key insight: strong semantic transport is not inherently unstable. What matters is whether it is applied along reliable directions and at reliable spatial locations. Motivated by this insight, we introduce PreciseEdit, a model-agnostic, training-free and inversion-free one-step editing framework that strengthens target-semantic alignment while preserving background fidelity and structural consistency. We further propose a dual-noise-scale role decomposition strategy: a higher-noise scale produces the strong target-directed transport used for actual editing, whereas a lower-noise scale jointly serves as a probe for directional agreement and spatial structure. Extensive experiments on PIE-Bench demonstrate that PreciseEdit achieves state-of-the-art performance and effectiveness on various editing tasks.
[CV-28] ReVision: Reprogramming Image Generators for Dense Visual Perception
链接: https://arxiv.org/abs/2610.23533
作者: Yitong Yang,Yinglin Wang,Xuexin Liu,Wei Chen,Zhaoyang Xu,Jiaxin Lin,Rui Chen,Shuting He,Kai Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Recent image generators exhibit rich visual priors that can support dense visual perception. However, existing approaches typically encode multiple perception tasks into a shared parameterization, making joint training sensitive to the mixture and weighting of heterogeneous data and prone to cross-task interference. To address this issue, we introduce ReVision, an instruction-conditioned parameter generation framework built on a frozen multimodal understanding model and a frozen image generator. Given an image and a free-form instruction, the understanding model jointly encodes them into a multimodal image–instruction representation, which a trainable parameter generator maps to transient low-rank updates for the generator. Each image–instruction pair induces a distinct set of transient prediction parameters, enabling instance-specific adaptation while both pretrained backbones remain unchanged. ReVision requires no task-specific heads, explicit task identifiers, or stored adapters. A shared RGB output interface unifies segmentation, depth estimation, and surface-normal prediction. Across diverse dense prediction benchmarks, ReVision achieves competitive performance with state-of-the-art methods while supporting flexible task specification through natural-language instructions. Our analysis further reveals substantial gradient disagreement across tasks, highlighting the conflict introduced when heterogeneous objectives share a single parameterization. Meanwhile, the generated updates exhibit structured task- and input-dependent organization in weight space, further validating instance-level parameter generation for multi-task perception.
[CV-29] PONDER: Learning to Generate in the Right Order
链接: https://arxiv.org/abs/2610.23531
作者: Julius Erbach,Dominik Narnhofer,Bernt Schiele,Jan Eric Lenssen,Konrad Schindler
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:We introduce PONDER, a method that learns in which order an image generator should fill in content and which of its earlier predictions to revise. Masked generative models construct a sample by repeatedly filling in subsets of missing variables. The order of these partial predictions determines the context available in subsequent steps and therefore affects the quality of the final sample. Common strategies are to choose the order randomly or based on local confidence estimates. Conditioned on the partially generated sample and the generator’s features, our policy chooses which positions to fill next and/or which previously filled positions to mask for later regeneration, a mechanism we call rethinking. In this way, the generation process is able to adapt to the evolving context and also to revise earlier predictions. We optimize the policy using group-relative policy gradients, with rewards that measure constraint satisfaction, human preference, or distribution quality. For the latter, we introduce FIDgrad, an image-level reward derived from the (first-order) change in Fréchet Inception Distance caused by adding a generated sample to the current feature distribution. With the generators frozen, learned ordering reduces MAR-B ImageNet FID from 4.09 to 2.78 and raises the accuracy on hard visual Sudokus from 3.9% to 56.6%, surpassing uncertainty-based greedy ordering (38.7%). Rethinking further improves these results to 2.70 FID and 96.7% accuracy, while increasing the rethinking budget of a fixed policy yields continued improvements at inference time.
[CV-30] AN: A Size- and Form-Adaptive Masked-Convolution Encoder for Tissue Graphs
链接: https://arxiv.org/abs/2610.23475
作者: Jan Herschel,Anselm von Gladiss
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Tissue graphs represent histology images as graphs over tissue-conforming superpixels, which keeps homogeneous tissue regions intact instead of cutting them into rectangular patches. To embed superpixels, current encoders crop them to bounding boxes and then pad, resize, or mask them, so background or fill values can leak into the node features. This matters in practice, and especially in medical imaging, where a prediction should depend on the tissue and not on an arbitrary preprocessing choice: on BACH, changing only the fill of the masked region shifts the accuracy of frozen encoders by up to 6 percentage points, larger than the differences among the four best encoders under mean fill, and the preferred fill differs between datasets, so it cannot be fixed once and for all. We present TAN (Tissue-Aware Network), a compact convolutional encoder (0.41M parameters) that processes each superpixel at its native shape and extent and excludes background from every operation, which makes its embeddings exactly invariant to the fill by construction. Under identical masks and an identical graph classifier, TAN’s accuracy on BACH is not significantly different from that of the 303M-parameter foundation model UNI (81.1% vs. 80.7-82.6%, depending on the fill), with ~740x fewer parameters, ~1,190x fewer FLOPs, and no pretraining. On BRACS, TAN is close to a ResNet18 trained under the same protocol and trails the histopathology foundation models. An ablation with TAN’s architecture but standard processing of fill-padded crops shows that masking keeps accuracy similar (nominally higher on BACH, lower on BRACS, within noise) at less than half the FLOPs.
[CV-31] When Evidence Vanishes: Distilling Visual-Evidence Dependency in Medical Vision-Language Models
链接: https://arxiv.org/abs/2610.23469
作者: Qizhen Lan,Yiling Ma,Ruiqi Chen,Mengchen Fan,Xi Xiao,Lin Zhao,Liang He,Baocheng Geng,Pulin Che,Tianyang Wang,Xiaoqian Jiang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Compact medical vision-language models (VLMs) can answer clinical questions correctly without reliably using the image, relying instead on question wording or learned priors. Standard knowledge distillation copies a teacher’s answer preference on the original image, but it leaves unchecked whether that preference is conditional on the visual evidence required by the question. We study this missing target as medical visual evidence dependency: support for the correct answer should decrease when the matched evidence is absent. We introduce Evidence-Conditioned Dependency Distillation (ECDD), which pairs each training image with a control matched to the evidence level and distills the teacher’s drop in answer margin across the pair. A lower bound objective transfers positive dependency by penalizing only shortfalls, while original-image supervision preserves task accuracy. The same formulation applies when the evidence is a queried clinical region, a surgical instrument, or whole-image visual content. Across three medical VQA settings, ECDD keeps original-image accuracy competitive with distillation and alignment baselines while making answer support more sensitive to evidence removal. Under the DME headline objective, for example, the positive-case margin response to target rather than random masking rises from 0.786 with KD to 2.023 with ECDD, while original-image accuracy differs by 0.11 points. These results suggest that medical VLM distillation should transfer both the teacher’s answer preference and its dependence on visual evidence.
[CV-32] riINR: Unsupervised Atmospheric Turbulence Mitigation with Triple Implicit Neural Representations WACV2027
链接: https://arxiv.org/abs/2610.23454
作者: Yuka Ogino,Rikuto Otsuka,Yuho Shoji,Takuya Ogawa,Takahiro Toizumi,Atsushi Ito
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to WACV2027
Abstract:Atmospheric turbulence degrades image quality in long-range imaging. The degradation is generally modeled with two spatio-temporally varying components, tilt and blur. Recent unsupervised turbulence mitigation methods have demonstrated the effectiveness of representing a sharp image as an implicit neural representation (INR). However, their tilt and blur representations remain frame-dependent, and the number of optimization parameters grows with the number of input frames. To address this limitation, we propose TriINR, an unsupervised turbulence mitigation method that recovers a sharp image from multiple degraded frames. TriINR represents the sharp image, tilt, and blur as three implicit neural representations. By modeling tilt and blur as spatio-temporal INRs, TriINR achieves frame-count-independent degradation representation and memory-efficient optimization. We further incorporate physical and statistical properties of atmospheric turbulence as constraints on the INRs to improve restoration accuracy. Experiments on synthesized turbulence data with three degradation strengths show that TriINR restores images more robustly across varying turbulence strengths than prior TM methods and achieves the highest accuracy among the evaluated unsupervised methods. Our method also demonstrates robustness on real long-range scenes containing additional non-ideal motion.
[CV-33] ShadowCast: On-the-Fly Context-Aware Shadow Augmentation for Remote Sensing Image Segmentation
链接: https://arxiv.org/abs/2610.23440
作者: Yechan Kim,Dongho Yoon,Jongmin Joo,JongHyun Park,Moongu Jeon
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Shadows cast by elevated objects alter the spectral appearance and obscure the boundaries of objects in remote sensing images, thereby degrading semantic segmentation performance. Most existing approaches handle shadows as degradations to be detected and removed, requiring additional costs from dedicated networks and shadow-specific annotations. We propose ShadowCast, which takes the opposite approach: it synthesizes shadows and injects them during training for remote sensing segmentation. It extracts masks of elevated objects, such as buildings and trees, directly from the segmentation labels and estimates an image-specific cast-shadow direction cue by probing the brightness around them. It then projects synthetic shadows along the inferred cast-shadow direction, requiring no learned network and no acquisition metadata. Experiments on three widely used remote sensing benchmarks, Vaihingen, Potsdam, and LoveDA, show that ShadowCast consistently outperforms the baseline across diverse state-of-the-art segmentation architectures. Moreover, it brings additional performance gains when combined with standard augmentation and regularization techniques, indicating that stochastic shadow injection serves as a simple and complementary augmentation strategy for remote sensing segmentation.
[CV-34] Momentum: Persistent Dynamics for Video World Models
链接: https://arxiv.org/abs/2610.23423
作者: Nedko Savov,Stefan Maria Ailuro,Danda Pani Paudel,Yilun Du,Xi Wang,Luc Van Gool
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Video world models must capture how a scene evolves beyond the current camera view to generate consistent observations when its content reappears. Existing approaches rely on observation history, static point clouds, or image caches constructed from past frames. These representations are often expensive and, more importantly, cannot update dynamic scene content outside the field of view: generated vehicles and pedestrians may be forgotten after leaving the field of view or reappear in physically inaccurate states. We represent an evolving scene through synchronized videos from reference-camera rigs, each comprising eight cameras that jointly cover the surrounding scene. We connect these rigs in a graph and propagate video generation between neighboring rigs to coordinate scene dynamics across locations and time. We introduce Momentum, a two-stage pipeline: WorldUpdate jointly generates reference-camera videos as a shared memory of scene dynamics, and WorldInteract grounds camera-controlled video generation in this evolving memory, allowing moving content to remain represented outside the target view and reappear consistently. To support training and evaluation, we develop a CARLA pipeline that captures synchronized videos from reference camera rigs and a moving target camera, together with an evaluation framework measuring consistency within the generated world state and the target this http URL on generated street scenes show that grounding target videos in scene dynamics generated by Momentum improves dynamic scene consistency by 126.2% in unseen scenes over static image conditioning, while also improving visual fidelity. Our models and source code will be made publicly available.
[CV-35] GeoReason : Bridging Logical Reasoning and Spatial Fidelity in Remote Sensing Segmentation
链接: https://arxiv.org/abs/2610.23416
作者: Cheng Fan,Han Wu,Tingzhang Luo,Minjing Dong,Yehui Tang,Jianyuan Guo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal Large Language Models (MLLMs) have significantly advanced instruction-driven perception through the “Embedding-as-Mask” paradigm. However, extending this capability to Remote Sensing (RS) reasoning segmentation remains exceptionally challenging under cluttered geographical contexts and extreme scale variations, often leading to Spatial Fidelity Degradation and Target Misidentification. We identify a critical yet underexplored bottleneck: the sequential alienation of the segmentation token [SEG] from its initial visual grounding within the autoregressive generation paradigm. Produced only at the end of long reasoning chains, [SEG] relies on visual evidence that has been progressively diluted by linguistic abstraction and autoregressive noise. This information decay is particularly catastrophic for RS targets with minuscule pixel footprints, where even marginal fidelity loss leads to localization failure. To address these bottlenecks, we propose GeoReason, a novel framework designed to fortify logical deduction and preserve spatial integrity. First, we introduce an Anticipatory Prior mechanism that injects the [SEG] token directly into the initial user query, shifting it from a terminal output to a primary condition. This paradigm shift transforms the token into a persistent spatial anchor, preventing information decay during long-form linguistic inference. Second, we enhance visual reasoning via a Location-aware Chain-of-Thought (CoT) that enforces coarse-to-fine localization using bounding boxes, coupled with a Latent Patch-to-Token Alignment (PTA) Loss that explicitly aligns the latent segmentation token with physical RS textures. Extensive experiments demonstrate that GeoReason consistently outperforms state-of-the-art methods on the Earthreason and RRSIS-D benchmarks. Code is available at this https URL.
[CV-36] Can We Use Saliency Maps for the Evaluation of Generative Image Models?
链接: https://arxiv.org/abs/2610.23411
作者: Eleonora Ristori,Luca Bindini,Marco Lippi,Paolo Frasconi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Evaluating generative image models remains challenging, as commonly used metrics depend on fixed pretrained feature extractors optimized for different tasks that do not always align with human realism perception across datasets and model families. To address this limitation, we investigate a human-inspired evaluation framework that avoids representations learned for unrelated tasks. Specifically, we train a generator-specific classifier to distinguish real images from samples produced by a given model and analyze its decisions using saliency maps. Across diverse benchmarks, we observe that these classifiers learn generator-specific artifacts rather than a universal notion of “fakeness”, exhibiting strong accuracy and limited cross-generator transfer. This makes saliency analysis particularly informative: the resulting attribution maps highlight visual cues and structural irregularities characteristic of each model’s outputs. By combining classification accuracy and saliency information, we obtain a composite score aligned with human judgments, without relying on large pretrained backbones and using only real and generated samples from the evaluated model.
[CV-37] LithoWorld: An Endpoint World Model for Lithography Process-Window Recovery
链接: https://arxiv.org/abs/2610.23408
作者: Botong Zhao,Fang Yu,Xinyuan Chen,Yue Lu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Lithography process drift can move a fabricated pattern outside its process window, producing unqualified scanning electron microscopy (SEM) morphology. We present LithoWorld, an endpoint world model for SEM-guided parameter recovery. Interior representative targets jointly determine qualified reference appearance and correction supervision. A causal predictive branch aligns currentimage features with endpoint representations, while a conditional flow policy generates a two-dimensional correction as an action chunk. The model learns from 173 groups of measured SEMs organized as focus-exposure maps, preserving unmeasured cells as unknown. On 200 test pairs from the 73-group subset, LithoWorld achieves 88.5% qualified-region success and 46.0% interior success, exceeding the strongest comparison method by 2.5 and 6.0 percentage points, respectively. Ablations separate the effects of target selection, endpoint alignment, and output parameterization. Qualitative SEM retrieval illustrates recovery to qualified interiors without requiring exact reference-coordinate matching.
[CV-38] FloodVLM-Pix: Language-Guided Pixel-Level Flood Understanding
链接: https://arxiv.org/abs/2610.23377
作者: Chengyang Zhang,Zhihao Wang,Zhiyu Long,Xueqian Wang,Quanming Yao,Wei Wu,Gang Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:As climate change intensifies, floods are becoming more frequent and destructive, making accurate flood understanding an urgent priority. However, existing vision-language models perform poorly on flood-related tasks, suffering from attentional bias and inaccurate descriptions of flood magnitude. To bridge this gap, we construct the first large-scale remote sensing flood multi-modal benchmark that supports language-level flood understanding and pixel-level flood segmentation. The benchmark contains 131,200 flood image patches and corresponding 443,100 textual descriptions for flood severity and evolution. Based on this benchmark, we propose FloodVLM-Pix, a language-guided pixel-level flood understanding model that includes a flood prior module and a multi-temporal flood pixel decoder to achieve accurate flood understanding. The flood prior module automatically distills sub-region flood spatial priors and inputs priors into FloodVLM-Pix, solving the attentional bias problem in flood language understanding. The flood pixel decoder introduces a novel general-specific feature fusion mechanism to address performance degradation caused by interference (e.g., clouds or mountain shadows) during flood segmentation. Experiments on FloodVL-Pix demonstrate that FloodVLM-Pix improves both flood description quality and pixel-level flood segmentation performance compared with state-of-the-art VLMs and flood segmentation models. Our code and data will be released at this https URL.
[CV-39] MOSS-Image Technical Report
链接: https://arxiv.org/abs/2610.23374
作者: MOSI.AI
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 35 pages. GitHub: this https URL
Abstract:Recent closed-source image generation systems can produce diverse visual content from simple compositions to information-dense graphics, yet open-weight text-to-image models still struggle with dense text and intricate layouts. To narrow this gap, we introduce MOSS-Image, an open generation system that combines high-fidelity visual synthesis with intent-aware composition. First, we pretrain a 5B single-stream diffusion transformer (DiT) from scratch with a carefully designed recipe, enabling bilingual long-form text rendering and precise layout control. Second, we develop an intention-aware agentic prompt enhancement (PE) module that translates diverse user requests into detailed content and layout specifications. Together, they enable MOSS-Image to generate text-rich and layout-coherent images. We further introduce LongLongTextBench (LLTB) for evaluating long-form English and Chinese text rendering, and RealLayoutArena (RLA) for end-to-end layout-rich generation. MOSS-Image achieves state-of-the-art performance among open-weight models on LLTB, with an average text coverage of 96.3%. On RLA, MOSS-Image ranks second in the end-to-end system comparison, while its backbone ranks first among generators evaluated with shared enhanced prompts. We will release the model, training recipe, and evaluation framework to support further research and promote the text-to-image community.
[CV-40] UniPolar: Joint Image and Polarimetric Restoration under Compound Degradations via Physical Consistency Learning
链接: https://arxiv.org/abs/2610.23360
作者: Yang Jin,Guangyu Guo,Linghao Shen,Binglu Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV)
备注: 13 pages, 11 figures
Abstract:Polarimetric image restoration aims to recover polarization angle images and reliable polarimetric quantities from degraded observations. Beyond visual fidelity, it also requires preserving the physical relationships underlying reliable polarimetric estimation. However, existing methods typically address either device- or environment-induced degradations in isolation, limiting their applicability to color polarization filter array (CPFA) imaging, where both often coexist. Such compound degradations jointly perturb image intensities and cross angle polarization differences. Consequently, improved image fidelity alone does not necessarily ensure accurate Stokes, DoLP, and AoLP recovery. To address this challenge, we propose UniPolar, a coupled Image Stokes framework for joint image polarimetric restoration from compound-degraded CPFA observations. Specifically, UniPolar employs two complementary streams to reconstruct four polarization-angle images and directly predict Stokes difference components. To coordinate these streams, Degradation Conditioned Cross Domain Interaction combines Dual Domain Frequency Interaction for frequency selective evidence exchange with Degradation Conditioned Modulation for condition adaptive feature correction. Furthermore, at the output level, Image Stokes Physical Consistency Learning aligns image derived and directly predicted Stokes components while regularizing angular fidelity and the physical feasibility of DoLP. On the compound-degradation benchmark, UniPolar achieves average image and DoLP PSNR of 31.522 dB and 24.097 dB, respectively, with an average AoLP error of 16.833 degree. These results underscore the importance of jointly considering visual fidelity and polarimetric reliability, offering a physically grounded perspective on learning based restoration under compound degradations.
[CV-41] STC-MGNet: Spatial-Temporal-Channel Attention-Enhanced Multimodal Graph Network for Pedestrian Intention Prediction
链接: https://arxiv.org/abs/2610.23356
作者: Yuanzhe Li,Meng Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Accurate prediction of pedestrian crossing intentions plays a critical role in enhancing the safety of pedestrian-vehicle interactions in autonomous driving. However, effectively capturing subtle spatiotemporal patterns in pedestrian pose sequences and integrating complementary visual and ego-vehicle motion information remain challenging. This paper proposes a spatial-temporal-channel attention-enhanced multimodal graph network (STC-MGNet) for pedestrian crossing intention prediction. STC-MGNet introduces a spatial-temporal-channel (STC) attention-enhanced pose graph encoder to learn discriminative representations from pedestrian pose sequences. Three parallel attention branches selectively emphasize informative skeletal joints, observation frames, and feature channels, with their outputs integrated through learnable fusion weights. To leverage complementary visual and ego-vehicle motion information, a multi-stage cross-modal channel modulation mechanism is introduced to adaptively refine pose representations through modality-specific channel gating at successive graph encoding stages. Experimental results demonstrate that STC-MGNet achieves accuracy rates of 70%, 87%, and 88% on the JAADbeh, JAADall, and PIE datasets, respectively, outperforming several representative baseline methods. Ablation experiments further highlight the complementary contributions of visual and ego-vehicle motion information to prediction performance.
[CV-42] CatchUp: Depth-Aware Re-Admission of Visual Tokens for Efficient Large Vision-Language Models WACV2027 ATC
链接: https://arxiv.org/abs/2610.23324
作者: Eunju Park,Uiwon Hwang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: WACV 2027. Code: this https URL
Abstract:Large Vision-Language Models (LVLMs) repeatedly process hundreds of visual tokens across decoder layers, incurring substantial computation and memory costs. Existing inference-time visual token pruning methods often irreversibly remove low-ranked tokens, although the set of visual tokens receiving high query-conditioned attention can change across decoder depth. Recoverable routing preserves skipped tokens for later reconsideration, but re-admits them with stale representations after bypassing intermediate transformations. We further find that replacing stale re-admitted states with their current-depth oracle states from a full-token reference trajectory substantially improves grounding performance. We therefore propose CatchUp, a training-free framework for depth-aware visual-token re-admission. CatchUp repeatedly re-evaluates query-conditioned visual relevance, processes only selected visual tokens through each decoder interval, and preserves deferred tokens for later re-admission. Upon re-admission, CatchUp approximates missed representation evolution using reference-to-current transition signals from semantically and spatially related visual tokens. Across three referring-expression grounding datasets and six multimodal benchmarks, CatchUp achieves the highest average relative retention in both benchmark groups at 64-, 128-, and 192-token budgets. At a 64-token budget with 88.9% fewer active visual tokens, CatchUp retains 92.17% of the full model’s performance on average across six multimodal benchmarks while reducing end-to-end latency by 8.0%. Code is available at this https URL
[CV-43] Look Reason Draw a Region: Recursive Refinement for Spatial Context Prediction
链接: https://arxiv.org/abs/2610.23302
作者: Wajeeha Tahir,Anna Tegon,Lionnus Kesting,Gamze İslamoğlu,Kristóf Karacs,Luca Benini,Thorir Mar Ingolfsson
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 17 pages, 6 figures, 3 tables
Abstract:Spatial context prediction requires visual grounding and spatial reasoning to identify vacant regions relative to objects in a scene. Spatial relationships and physical constraints remain challenging for large vision-language models (VLMs). Recursive reasoning has demonstrated strong performance on discrete and abstract reasoning tasks, motivating its use for improving spatial context prediction in natural images. Weight sharing enables higher accuracy with a small trainable parameter count. We introduce Spatial-TRM, a compact recursive model with frozen vision-language features, a convex region decoder, and just 11.2M trainable parameters. This weight-tied reasoner supports six spatial relationships, three reference frames, and open-vocabulary anchor objects. On the RoboSpatial-Home benchmark, Spatial-TRM achieves 29.51% context accuracy, a gain of 9.02 percentage points (44.0% relative) over a separately trained single-step model with the same parameter count. Including the frozen encoders, the model has 877.8M total parameters, approximately 15\times fewer than the reported 13B size of RoboPoint. These results show that repeated refinement improves spatial context prediction from pretrained vision-language features at a fixed trainable parameter count on this benchmark.
[CV-44] FlashReFocus: Interactive Image Refocusing in Seconds
链接: https://arxiv.org/abs/2610.23297
作者: Ching-Heng Cheng,Chia-Ming Lee,Ming-Ching Chang,Xin Li,Yu-Lun Liu,Chih-Chung Hsu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Image and Video Processing (eess.IV)
备注: 38 pages (8 main + appendix), 47 figures, 16 tables. Project page: this https URL Code: this https URL
Abstract:Single-image refocusing changes the focal plane and depth of field of a photograph after capture. It is inherently interactive: users try many focal points and blur strengths on one image, so each edit should be fast and change only the blur. Recent generative methods, however, rerun large diffusion models for every edit. We observe that the two sub-problems differ in kind: recovering the all-in-focus image is ill-posed and needs a generative prior, whereas rendering blur is largely determined by depth and the focal plane. We therefore propose FlashReFocus, which runs diffusion once per photograph and renders every edit with a lightweight network. To address the ill-posed restoration, a one-step diffusion deblurrer with a decoder refiner recovers fine detail efficiently while balancing fidelity and perceptual quality. For precise control, the renderer is conditioned on an explicit defocus map and uses defocus-bias attention, which lets pixels at similar defocus share information without interference from pixels at other depths; in-focus content stays unchanged, and editing the map yields tilt-shift and two focal planes. To further test generalization to unseen cameras, we collect 3CReal, a three-camera benchmark of 89 pairs. FlashReFocus achieves the best perceptual fidelity for zero-shot bokeh rendering on EBB! and 3CReal, refocuses 148.3x faster than GenRefocus, and takes about half a second per further edit, 462x faster.
[CV-45] he Optimal Ranking Tradeoffs Between Two Ranking Scores in Two-Class Classification
链接: https://arxiv.org/abs/2610.23294
作者: Sébastien Piérard,Adrien Deliège,Marc Van Droogenbroeck
类目: Performance (cs.PF); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:The accuracy, true negative rate (negative class recall), true positive rate (positive class recall), negative predictive value (negative class precision), positive predictive value (positive class precision), class-specific intersections over unions, class-specific F-scores, Cohen’s kappa, and balanced accuracy are scores that one intuitively wants to maximize when developing machine learning methods. However, they lead to different rankings of such methods. So, how can we establish a ranking that is an optimal tradeoff between any pair of these scores? It turns out that all these scores are special cases of ranking scores: a continuum of scores parameterized by the relative importance values assigned to the true negative, false positive, false negative, and true positive outcomes. Since all these scores are known to induce meaningful rankings, we seek the optimal tradeoffs among them. We consider the tradeoff between two scores to be optimal when it minimizes the Fréchet variance regarding the Kendall distance (also known as the bubble-sort distance) between rankings. For any given set of classifiers to rank, the optimal tradeoff lies at the center of the shortest (geodesic) path on the manifold of all rankings induced by the ranking scores. However, in the parameter space of the ranking scores, it is rarely achieved by averaging the importance values corresponding to the initial scores. In this paper, we present an exact analytical solution for determining a ranking score that is an optimal tradeoff for any given pair of initial scores, in the common case where the class priors are the same for all compared performances. We also experimentally investigate what happens when performance follows some distributions with fixed priors.
[CV-46] I2V-DETACH: Source Grounding Detachment for Unauthorized Image-to-Video Generation NEURIPS2026
链接: https://arxiv.org/abs/2610.23287
作者: Chanhui Lee,Yeonghwan Song,Yewon Kang,Jeany Son
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: NeurIPS 2026
Abstract:Recent diffusion models have advanced image-conditioned video generation, but their accessibility raises risks of unauthorized manipulation. Image immunization has emerged as a promising defense for Image-to-Video (I2V) protection, yet existing methods often rely on degrading visual quality or temporal coherence. In this work, we revisit I2V immunization as reducing source-content preservation, shifting the focus from output degradation to weakening source-conditioned content propagation. To this end, we propose I2V-DETACH, an image immunization method that prevents unauthorized I2V generation through source grounding detachment. I2V models preserve source content by propagating source-image cues into video latents during denoising. Accordingly, I2V-DETACH disrupts this process in two complementary ways: it suppresses source-video grounding to reduce source cue propagation, and strengthens intra-video interactions to shift denoising toward video-side dynamics with reduced reliance on the source image. To obtain informative optimization signals, we initialize auxiliary video latents with source-oriented states that remain within the source-conditioned generation regime while deviating from source-preserving trajectories. Extensive experiments across diverse diffusion architectures show that I2V-DETACH substantially reduces source-content preservation, improves over prior immunization methods, and exhibits strong black-box transferability.
[CV-47] Xiaomi-Lens-0 Technical Report
链接: https://arxiv.org/abs/2610.23279
作者: Pei Fu,Yike Liu,Yue Sun,Hanmo Wang,Xiuwen Xi,Ruoceng Zhang,Shaojie Zhang,Chao Jiang,Hang Li,Bingrui Liu,Zhenbo Luo,Heng Qu,Changjian Xu,Ming Xu,Jian Luan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal large language models (MLLMs) answer questions about whole images well, yet they still struggle with fine-grained region understanding. A key reason is that text-only supervision never specifies which regions support an answer. To close this gap, we present Xiaomi-Lens-0, a 4B MLLM built on Qwen3.5 that predicts masks from its own visual representations. Specifically, a lightweight decoder combines vision-encoder patch features with visual latents, the language model’s mask-token states, so mask supervision reaches the shared backbone without an additional segmentation encoder. Because each latent decodes to a mask, the model can interleave reasoning with region-specific latents and thus think with visual latents, while grounded reasoning traces supervise the regions it visits during chain-of-thought. As a result, across all eight RefCOCO, RefCOCO+, and RefCOCOg splits, Xiaomi-Lens-0 gains 5.6 points on average over Qwen3.5 in referring comprehension and achieves the highest segmentation cIoU among compared models. Moreover, it surpasses Qwen3.5 across all perception benchmarks. Homepage: \hrefthis https URLthis https URL
[CV-48] Agent ic Image Generation in the Wild
链接: https://arxiv.org/abs/2610.23269
作者: Yifan Li,Bowen Qin,Kun Zhou,Haochen Zhao,Zikang Liu,Zheng Li,Minsheng Hao,Yuhao Cheng,Haosheng Zou,Jeff Chen,Wayne Xin Zhao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Agentic image generation pairs an image generator with a search agent that retrieves references from the web, extending generation beyond the generator’s training data. Real-world requests, however, often concern entities that appeared only days ago. We call this setting Image Generation in the Wild. In this setting, the web offers limited visual evidence, search results are crowded with easily confused entities, and the target is rarely shown on its own. To study this setting, we build an automated pipeline that synthesizes prompts from recent headlines and keeps only those whose real search results are highly confusing, yielding requests that are hard to solve yet natural to ask. With it, we construct Savanna, a benchmark of 500 prompts across 16 subcategories. On Savanna, even harness frameworks built on the frontier models retrieve a correct reference for only about half of the prompts. We then propose Locate-Corroborate-Isolate, a paradigm in which the agent first pins down the target’s identity through text search, then verifies candidate images against multiple sources, and finally crops a clean reference for the generator. Trained on 15K trajectories that follow this paradigm, Ranger-8B achieves a search score of 67.2, outperforming harnesses equipped with frontier models. It also achieves the best generation quality with every generator we test, and surpasses prior agentic methods on other image generation benchmarks. We release the pipeline, benchmark, training data and model at this https URL.
[CV-49] SCALE: Improving Guidance-Free Visual Generation via Self-Contrastive Alignment for Likelihood Estimation NEURIPS2026
链接: https://arxiv.org/abs/2610.23263
作者: Jiwan Hur,Dong-Jae Lee,Dongyeun Lee,Gyojin Han,Junmo Kim
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to NeurIPS 2026. Project page: this https URL
Abstract:Classifier-free guidance (CFG) is essential for improving sample quality in visual generation, but it incurs substantial sampling overhead. Existing guidance-free approaches halve this cost by approximating the CFG with a single model inference, but their performance is bounded by the CFG and inherits its limitations. In this paper, we propose self-contrastive guidance (SCG), which contrasts the model’s prediction on real and self-generated samples to obtain a direct, model-based signal that pushes generation toward the true data manifold. Integrating SCG into the standard likelihood objective via reparameterization yields Self-Contrastive Alignment for Likelihood Estimation (SCALE), a guidance-free training framework that produces SCG-enhanced predictions in a single forward pass. Across class-conditional and text-to-image generation on diffusion and autoregressive models, SCALE consistently outperforms CFG-mimicking guidance-free baselines and improves further when combined with DPO, demonstrating that the self-contrastive signal yields gains complementary to both CFG and preference-based fine-tuning.
[CV-50] Expand-and-Prune: Test-Time Candidate-Space Adaptation for End-to-End Autonomous Driving WACV
链接: https://arxiv.org/abs/2610.23262
作者: Fabian Schmidt,Karol Fedurko,Markus Enzweiler,Abhinav Valada
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2027
Abstract:Candidate-based end-to-end driving planners reason over a finite set of trajectory hypotheses. However, for sparse planners, a limited candidate set may provide insufficient support for scene-relevant maneuvers, while fixed downstream candidate budgets do not account for prediction uncertainty. To address these challenges, we introduce Expand-and-Prune, a test-time framework that jointly adapts candidate support and capacity without updating learned parameters. The framework expands the native candidate set using semantic driving intent and temporally propagated predictions and then uses uncertainty in intermediate trajectory scores to adaptively prune the resulting candidates. Across four heterogeneous candidate-based planners and four benchmarks, Expand-and-Prune consistently improves planning performance. It achieves state-of-the-art results on LangAuto, increasing the Driving Score from 68.2 to 75.9, while providing further improvement on Bench2Drive, Fail2Drive, and NAVSIMv2, with particularly strong improvements in out-of-distribution scenarios. Moreover, in its full instantiation, 72.5% of selected trajectories originate from the newly introduced intent-guided or temporal candidates, confirming that candidate-space expansion provides useful hypotheses beyond those generated natively by the planner. The code is available at this https URL.
[CV-51] LonGaussian: Tracking Foundation Priors for Long-Context Feedforward Gaussian Reconstruction
链接: https://arxiv.org/abs/2610.23250
作者: Zhoufeng Xie,Guanjun Wu,Shuai Zhang,Xinggang Wang,Wenyu Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Reconstructing high-quality, renderable 3D scenes from long videos is essential for robotic perception and embodied intelligence. Many feed-forward Gaussian reconstruction methods rapidly predict 3D scenes from a small set of images, but extending them to longer sequences presents two bottlenecks: jointly processing all views increases GPU memory requirements, while accumulating pixel-aligned predictions redundantly represents the same surfaces and increases storage and rendering costs. We present LonGaussian, a streaming feed-forward framework that reconstructs compact, globally renderable Gaussian scenes from long image sequences with known intrinsics and unknown camera poses. Built on a pretrained streaming 3D reconstruction backbone, LonGaussian decodes long-context geometric features into dense Gaussians. Within each chunk, multi-frame point tracks link repeated observations of the same surface and guide a learned merger to consolidate the corresponding Gaussian predictions. Rendered-depth consistency filtering further suppresses Gaussians from other chunks that incorrectly occlude locally reconstructed surfaces. On DL3DV and RealEstate10K benchmarks, LonGaussian improves novel-view rendering quality while using substantially fewer Gaussians than the evaluated feed-forward baselines.
[CV-52] Can the Parts Fool the Test? Counterfactual Pairing Cycles for Relational OOD NEURIPS2026
链接: https://arxiv.org/abs/2610.23248
作者: Yibo Zhou,Yirong Xiang
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Accepted at Neurips 2026
Abstract:Relational out-of-distribution errors are easy to describe and hard to evaluate cleanly. An image may be natural, a caption may be fluent, and all mentioned objects may be common, yet the caption may describe the wrong relation in the image. Many evaluations intended to test this behavior can be solved without checking the pairing at all, because negatives differ in caption style, template, length, object frequency, or image statistics. We propose counterfactual pairing cycles, a blocked evaluation primitive that reuses the same two images and the same two captions on both sides of the comparison, changing only which image is paired with which caption. This makes image-only, text-only, and additive component shortcuts cancel exactly in finite samples. The same contrast also gives a training loss for relational outlier exposure and a diagnostic for bilinear vision-language scores. Across synthetic controls, GQA, and COCO, cycles separate coarse image-caption matching from directional role-binding failures, while also identifying geometric size relations that remain strongly visible in global features. In a shortcut-contaminated COCO outlier-exposure stress test, a model selected by a near-perfect shortcut validation AUROC retains substantial relational signal but remains below cycle training on exact cycles; cycle training improves CycleAcc from 0.882 to 0.927. The main lesson is simple: to test whether a model understands a pairing, the benchmark should keep the parts fixed, change the pairing, and ask only about coupling. The official implementation is available online.
[CV-53] Kingfisher: Fast Personalized Visual Autoregressive Generation via Scale-Aware Prompt-Conditioned Scale Decomposition
链接: https://arxiv.org/abs/2610.23238
作者: Dinh-Khoi Vo,Akihiro Sugimoto,Minh-Triet Tran,Trung-Nghia Le
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Personalized image generation has been extensively studied in diffusion models, often relying on subject-specific optimization of model weights, textual embeddings, LoRA modules, or adapters. In this work, we explore a different direction for fast personalization in visual autoregressive models. Our key observation is that a short prefix of early autoregressive scale codes is sufficient to strongly determine both object structure and global appearance, while later scales primarily refine the established generation trajectory. Based on this insight, we introduce Kingfisher, a model-tuning-free framework that decomposes early residual scale codes into content- and style-related components through prompt-conditioned scale-code subspaces. Given two real images, Kingfisher first decomposes their early scale codes, then recomposes the content components of one image with the style components of the other into a compact semantic prefix. The frozen generator subsequently predicts the remaining scales through bitwise self-correction for coherent refinement. Experiments on CSD-100 show that Kingfisher achieves strong content preservation and the highest style and text alignment among the evaluated methods, while reducing pair-specific preparation time by approximately 54x compared with diffusion-based methods and 7.5x compared with VAR-based alternatives. These results demonstrate that effective personalization can be achieved by directly manipulating the native scale-space representation of a frozen visual autoregressive generator.
[CV-54] X-Splat: Unified Gaussian Splatting for Joint RGB and X-ray 3D Reconstruction SIGGRAPH
链接: https://arxiv.org/abs/2610.23237
作者: Wangze Xu,Chu Zhou,Mae Yamaguchi,Linrui Dai,I-Chao Shen,Takashi Ijiri,Imari Sato
类目: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
备注: SIGGRAPH Asia 2026 Conference Papers. 19 pages (12 main paper + 7 supplementary), 13 figures. Project page: this https URL
Abstract:Jointly reconstructing RGB and X-ray provides new insights by connecting external appearance with internal structure, enabling a more complete understanding of real-world objects than single modality. Yet existing methods train separate models per modality and suffer from redundant geometry, alignment artifacts, and loose coupling that limits cross-modal insight discovery. We propose a unified representation in which all Gaussian primitives share a single set of geometric parameters, with only surface-proximal primitives additionally carrying appearance attributes. We perform cross-modal registration through clean multi-view stereo and X-ray geometric proxies instead of aligning two sets of optimized Gaussians, improving alignment stability. We then jointly optimize the unified representation via supervision from both modalities, complemented by a surface-depth consistency loss derived from X-ray geometry to guide RGB reconstruction. Experiments on the InsideOut dataset and our Flower Insect dataset show that our framework achieves more robust cross-modal alignment, higher RGB rendering quality without compromising X-ray reconstruction, and a more compact representation than maintaining two independent geometries.
[CV-55] RAP: A Dataset for Paper-to-Poster Generation via Declarative Markup from Human-Crafted Exemplars
链接: https://arxiv.org/abs/2610.23230
作者: Haonan Han,Yukang Lin,Minliang Lin,Haoxian Tan,Jun-Yan He,Rui Yang,Zhi-Qi Cheng,Jie Hu,Guanglu Wan,Xiu Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Making a conference poster is visual coding in miniature: an entire paper must be compressed onto a single page, which exercises summarization, visual design, and code implementation at once, using nothing beyond the paper itself. Yet benchmarks for this task in the context of visual coding remain scarce: existing resources are small and annotated only with layout boxes, so evaluation must fall back on VLM-as-judge or human checklists rather than objective render-and-compare, and the generation methods they support design layouts from scratch, leaving human-crafted posters unusable as either a retrievable design library or as verified supervision. We introduce RAP (Reconstruct A Poster), a dataset and benchmark for paper-to-poster generation as a visual coding task. RAP contains 25K+ verified paper-poster pairs from various machine learning conferences held between 2022 and 2026, annotated with a three-level content taxonomy, plus a held-out benchmark of 150 test papers each paired with its human poster. A fully automated derendering pipeline turns each poster into styled HTML/CSS of typed, refillable slots, so every sample is a paper-poster-markup triplet verifiable by rendering against the source poster. The corpus serves both inference and training: PosterRAG, a training-free multi-agent pipeline, retrieves an exemplar and refills its slots without touching layout or styling, while an SFT+RL recipe trains a dedicated model from the same triplets. An online demo is available at this https URL.
[CV-56] MLLM Frame Subset Ensembling for Audio-Visual Video QA and MLLM -based Reranking for Ad-hoc Video Search in TRECVID 2025
链接: https://arxiv.org/abs/2610.23214
作者: Andreas Goulas,Damianos Galanopoulos,Ioannis Patras,Vasileios Mezaris
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Proc. TREC 2025 Workshop, Dec. 2025
Abstract:This paper presents the overview of the runs related to the Ad-hoc Video Search (AVS) and Video Question Answering (VQA) tracks of TRECVID 2025 on behalf of the CERTH-ITI team. For the AVS track, we introduce a two-stage framework built on foundation models. In the first stage, multiple vision-language models (VLMs) encode both the input query, augmented through LLM-generated rephrasings, and the candidate video shots, producing weighted similarity scores for initial retrieval. In the second stage, we utilize a Multimodal-LLM(MLLM)-based reranking module that evaluates the semantic alignment between each shot among the top-N highest-ranked ones and the original query, generating updated relevance scores for reordering these shots. This MLLM-driven reranking significantly improves contextual matching and produces more accurate final rankings without requiring any model training. Regarding the VQA track, we fine-tune an audio-visual MLLM model on the provided TRECVID training dataset and we implement an inference-time scaling technique to enhance the multimodal understanding capabilities of the MLLM. For the open-ended Answer Generation (AG) task, we aggregate multiple model responses per question via a majority vote. The responses are generated with greedy sampling from different random frame subsets of the video and they are ranked based on the number of votes. For the Multiple-Choice (MC) task, instead of voting, we use mean pooling on the logits assigned by the fine-tuned model to each candidate response. Through the combination of fine-tuning and frame subset ensembling we achieve the highest score across 3 metrics in the VQA AG task and the second highest in the VQA MC task.
[CV-57] Knowledge-Guided Distributional Prompt Learning with Visual Hierarchy for Zero-Shot Anomaly Detection
链接: https://arxiv.org/abs/2610.23172
作者: Jiyul Ham,Seongho Yoon,Yonggon Jung,Sangjae Lee,Jun-Geol Baek
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 22 pages, 12 figures
Abstract:Zero-shot anomaly detection (ZSAD) aims to detect anomalies in previously unseen classes using auxiliary images, with applications such as industrial visual inspection and medical imaging. Recent vision-language-model-based approaches improve ZSAD by aligning textual and visual representations, while distributional prompting further models prompt variability for unseen categories. However, existing methods do not explicitly combine normality-referenced distributional modeling with hierarchy-specific prompt specialization, limiting their ability to model anomalous semantics relative to normality while addressing both image-level detection and fine-grained localization. We propose Hierarchical Knowledge-Guided Distributional Prompt Learning (HiDist), a normality-referenced hierarchical distributional prompt framework for ZSAD. HiDist models each prompt token with a diagonal Gaussian distribution and generates stochastic prompt variations through Monte Carlo sampling. Rather than learning normal and anomalous distributions independently, anomalous prompt distributions are parameterized as deviations from learned normal references, explicitly encoding their dependency. The resulting distributions are specialized into global and local representations for image-level detection and pixel-level localization, respectively, while an adapted global visual representation guides hierarchical alignment. Experiments on seven industrial and medical datasets demonstrate strong cross-dataset performance across anomaly detection and localization tasks. Extensive ablation and representation analyses further support the effectiveness of the proposed distributional modeling and hierarchy-specific representation learning.
[CV-58] Reliability-Consistent Appearance Modeling for RGB-Infrared Multi-Object Tracking
链接: https://arxiv.org/abs/2610.23125
作者: Bingyang Wang,Jiazuo Yu,Lu Zhang,Lihe Zhang,Huchuan Lu,You He
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 6 figures. Submitted to IEEE Transactions on Multimedia
Abstract:Recent RGB-infrared multi-object tracking (MOT) methods have substantially improved cross-modal representation learning and multimodal information integration. However, most existing methods compress RGB and infrared evidence into a unified descriptor. When one modality is temporarily degraded, the tracker can no longer regulate its contribution separately; unreliable evidence may distort association and contaminate appearance memory. We therefore propose RCA-MOT, a reliability-consistent appearance modeling framework that preserves cross-modal complementarity while retaining post-representation evidence controllability. RGB-specific, infrared-specific, and fused identity representations are maintained as jointly accessible appearance states. For each detection, a simple training-free target-scene estimator produces branch-wise appearance reliability used across the two temporal roles of appearance: relative reliability regulates different identity branches, absolute reliability calibrates borderline candidates near the ReID acceptance boundary, and the same branch-wise reliability controls how strongly matched descriptors update historical appearance memory. The individual components are intentionally simple, focusing evaluation on reliability-controllable appearance organization rather than elaborate fusion or learned reliability-estimation modules. RCA-MOT sets new SOTA results on VT-MOT and UniRTL, improving the previous best HOTA by 5.402/6.075 points and IDF1 by 11.549/12.789 points, respectively, while post-hoc analyses further validate reliability as an informative control signal. Code will be released upon acceptance.
[CV-59] Scene-Designer: Learning Design Intelligence for Agent ic 3D Scene Generation
链接: https://arxiv.org/abs/2610.23121
作者: Xuanyu Liu,Hongyan Feng,Miao Pan,Sunlai Chen,Lizhou Liang,Bing Sun,Wenqi Zhang,Xuhong Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Generating complete, coherent, and usable 3D indoor scenes from natural language requires strong \textbfdesign intelligence: the ability to design high-quality scenes and critique their quality. However, existing approaches either rely on general-purpose models lacking domain-specific design experience or learn from static scenes. In this work, we introduce \textbfScene-Designer, an agentic framework that learns design intelligence through complementary design and critique experiences. We construct two datasets: \textbfScene-OTA-35K, which provides Observation–Thought–Action supervision for learning scene design and efficient error correction, and \textbfScene-OTC-15K, which provides Observation–Thought–Critic supervision for learning scene quality critique across physical validity, functional correctness, and layout coherence. Based on these experiences, we propose \textbfScene-GRPO, a Critic-guided GRPO framework that improves design decisions and reduces unnecessary trial-and-error corrections. Extensive experiments demonstrate that Scene-Designer achieves superior performance in physical validity, functional correctness, and layout coherence, while the learned Scene-Critic significantly outperforms general-purpose models in scene critique. Moreover, test-time Critic guidance further improves scene generation, showing that the Critic can provide useful guidance beyond training. Additional analysis suggests a trend of mutual enhancement between scene design and critique abilities.
[CV-60] Domain-specific language alignment of multimodal embeddings towards explainable predictions
链接: https://arxiv.org/abs/2610.23110
作者: Gabrielė Tijūnaitytė,Thijs L. van der Plas,Mohamed-Bachir Belaid,Rob Knapen,Maciej J. Soja,Ioannis N. Athanasiadis
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal data are commonly fused into latent representations that improve downstream task performance but are not interpretable, limiting their application in domains that require explainable model decisions. Vision-language alignment has made image embeddings interpretable through language, but requires vast image-caption training data. However, captioned data are typically not available for multimodal embeddings, as relevant features cannot directly be observed and annotated from embeddings alone. Here, we propose a framework to language-align any embedding space that can be linked to interpretable auxiliary data. Captioning functions convert auxiliary data into partial captions; a soft contrastive loss weights pairs by the similarity of their properties that each caption describes; and a rank-based concept retrieval task evaluates alignment against the auxiliary data rather than the captions. We apply this framework to multimodal Earth embeddings to facilitate explainable predictions for two species distribution modelling tasks: S2BMS and SatBird. AlphaEarth embeddings beat previous state-of-the-art models on 6/7 metrics, while our framework language-aligns domain-specific concepts of the embedding space, outperforming previous Earth observation vision-language models on average concept retrieval for both datasets. Finally, aligned concepts recover species-habitat associations ( \rho =0.44) and relate habitat descriptions to species predictions, showing that abstract embeddings can be interrogated in the language of domain experts.
[CV-61] riFold: Folding irregular 3D Gaussians into structured latents for 3D Scene Generation
链接: https://arxiv.org/abs/2610.23050
作者: Yun Wang,Chiyu Wang,Fan Jiang,Ziyu Zhu,Zhaoxu Sun,Guangchi Fang,Mu Xu,Dapeng Wu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 30pages, Project page: this https URL
Abstract:Recent progress in 3D generation has motivated growing interest in compact generative representations of complex 3D scenes. For native 3D Gaussian scenes, however, learning such representations remains challenging due to widely varying primitive counts, nonuniform spatial density, and differences in geometry and appearance even among spatially neighboring primitives. In this work, we introduce TriFold, a native 3D Gaussian VAE that encodes Gaussian scenes into a fixed-layout triplane latent and jointly reconstructs their geometry and appearance. The key insight is to impose structural regularity on the latent space while preserving the spatial adaptivity of the underlying Gaussians. During encoding, we use cell-wise learned queries to aggregate projected Gaussians into multiple slots per cell, reducing the information loss caused by single-feature pooling. During decoding, occupancy prediction identifies active regions, where query-based set prediction and candidate selection reconstruct local Gaussian sets with spatially varying primitive counts. Building on this representation, we train a latent flow-matching model for conditional scene-level 3DGS generation, enabling point-cloud-conditioned synthesis and single-image-conditioned amodal generation. Experiments demonstrate that TriFold achieves high-fidelity reconstruction and generation from a single image that provides only a partial observation of the scene, validating its effectiveness for native 3D Gaussian representation learning. this https URL
[CV-62] SpikeQ: Spike-Guided Post-Training Quantization for Spiking Neural Networks
链接: https://arxiv.org/abs/2610.23046
作者: Kai Sun,Peibo Duan,Zhanglu Yan,Guowei Zhang,Yongsheng Huang,Levin Kuhlmann
类目: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:On neuromorphic chips, spiking neural networks (SNNs) consume far less energy than artificial neural networks (ANNs), but the limited memory of these chips requires low-bit synaptic weights and membrane potentials. Existing post-training quantization (PTQ) methods provide them without retraining and, like those for ANNs, minimize numerical errors such as the mean squared error (MSE). In an SNN, however, a quantization error reaches later layers only through the spikes it flips. We find that errors of equal MSE can cause very different accuracy losses, whereas errors that flip the same fraction of spikes cause the same loss. We therefore propose SpikeQ, a spike-guided PTQ framework. For weights, we select the level of each weight by simulating the downstream neurons and minimizing their spike errors. For membrane potentials, most existing methods quantize the state after the reset, whereas neuromorphic chips compare the threshold with the membrane potential stored at a fixed bit width. We target this pre-fire quantization and build a reset-aware grid that places the threshold on a rounding boundary, so rounding never flips the spike at the current time step. We evaluate SpikeQ on seven benchmarks of image classification, semantic segmentation, and object detection. With 2-bit pre-fire membrane potentials, every existing grid collapses, whereas SpikeQ retains most of the full-precision accuracy. With both at 4 bits, the models stay within 2.59 points of full precision, shrink their memory footprint 7.8–7.9 times, and use 6.4–6.9 times less energy.
[CV-63] Frame-Consistent Feed-Forward 3D Gaussian Splatting from Sparse Unposed Images
链接: https://arxiv.org/abs/2610.23039
作者: Viruthshaan Vinasirajan,Ranga Rodrigo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Feed-forward Gaussian models have made novel view synthesis almost instantaneous. Feed-forward methods predict Gaussians directly, either in two stages or in a single merged pass. Using two separately trained models is attractive because each can specialize in its own objective: one for accurate pose estimation and the other for high-quality reconstruction, rather than forcing a single jointly trained network to compromise between the two. A natural instantiation pairs a frozen pose estimator, VGGT, with a frozen reconstruction backbone, TokenGS. Prior work, such as VGGT-X and 3R-GS, also adopts a two-stage design, but attributes the resulting quality degradation to noisy pose estimates and introduces learned refinement. We show that this is not the primary source of the problem. Through per-camera analysis across 107 scenes from two datasets, DL3DV and MipNeRF360, using COLMAP poses as the reference, we find that the rotation error after alignment is small (median 1.6 degrees), whereas the translation scale is offset by a large, dataset-consistent factor. Controlled ablations over 140 DL3DV scenes show that the primary obstacles are two integration failures rather than pose error: a coordinate-frame inconsistency and a translation-scale mismatch, both of which can be corrected deterministically. With these fixes, a frozen VGGT driving a frozen TokenGS achieves 22.47 PSNR, within 0.51 dB of the COLMAP reference, with no additional training, when poses are estimated from all scene views. In the sparse setting, where reconstruction uses only four context images under the standard pose-free protocol, the same pipeline matches this all-view quality, yielding a fully COLMAP-free pipeline without any refinement step.
[CV-64] Semantic-Critical Regions for Video-to-Audio Generation
链接: https://arxiv.org/abs/2610.23021
作者: Yun Fang,Wei Guo,Zhuonan Liang,Heng Wang,Weidong Cai
类目: Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM); Sound (cs.SD); Audio and Speech Processing (eess.AS)
备注: 11 pages, 7 figures, 7 tables, accepted by ACM Multimedia Asia 2026
Abstract:In real-world videos, sound events are often associated with localized and fine-grained visual cues, such as object motion, object-part dynamics, contact events, and interactions between objects. Existing video-to-audio (V2A) methods employ implicit conditioning based on global visual representations or explicit control derived from temporal and spatial cues. However, subtle sound-producing and sound-indicative interactions remain insufficiently captured, leaving room for more precise visual conditioning of sound-event semantics and temporal dynamics. Motivated by this observation, we refer to regions containing such sound-relevant visual evidence as Semantic-Critical Regions (SCRs) and introduce SCR-V2A, a V2A generation framework that explicitly localizes and encodes SCRs as fine-grained visual guidance for audio generation. Specifically, we first identify SCRs through sound-related visual localization. We then transform the localized regions into a control representation through adapter-based dual visual encoding, which jointly captures global visual context and fine-grained sound-relevant evidence. The representation is used to condition a downstream audio generation model. Experiments on both in-domain and out-of-domain benchmarks demonstrate that SCR-V2A achieves strong performance in distribution matching and audio quality while remaining competitive in cross-modal alignment and requiring few trainable parameters. These results highlight the value of prioritizing fine-grained, sound-relevant visual regions for V2A generation.
[CV-65] BRICE: Brain-inspired Recurrent Illusory Contour Encoder for Point Cloud Completion
链接: https://arxiv.org/abs/2610.23001
作者: Yunlei Wang,Dawei Li,Xue-song Tang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Mainstream point cloud completion methods are unanimously based on deep learning architectures. On one hand, these methods tend to repetitively stacking attention modules to enhance network performance, which drifts away from the neat design of unified and interpret-able networks; on the other hand, nearly all existing architectures rely solely on feed-forward connections to propagate information, neglecting feedback from deeper layers. The recurrent neural architecture, formed by feedforward-feedback loops, have shown its ability to generate Illusory Contours (ICs) in the V1 brain region of mammals; and ICs are essentially a completion mechanism of the brain for visual information. Motivated by this, we design a Brain-inspired Recurrent Illusory Contour Encoder (BRICE) to “imagine” what is missing in the incomplete 3D point clouds. Building upon the foundation model of Predictive Coding (PC), the plug-and-play nature enables BRICE to be seamlessly integrated into mainstream point cloud completion baselines. Experimental results show that BRICE can enhance various completion baselines on different 3D datasets such as MVP2K, PCN, and PlantPCom. Our code: this https URL.
[CV-66] A100: A Hybrid Vision Foundation Model for Agricultural Plant Image Analysis
链接: https://arxiv.org/abs/2610.22991
作者: Mayur Pratim Das,Ajai Dubey,Aksshay Mathew P,Sana Afreen,Rishi Kumar Soni,Deepak Meena,Pratibha Kumari,Mukesh Saini
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 11 pages, 5 figures, 5 tables
Abstract:The use of accurate vision models for crop monitoring, disease diagnosis, and plant phenotyping in agriculture is hindered by the lack of expertly annotated data, especially in developing countries. We propose A100, an agricultural vision foundation model that is pre-trained using self-supervised learning on 3 million unlabeled images from fields covering more than 100 crop species. A100 uses a hybrid CNN-Transformer backbone that fuses MBConv stages for local feature extraction with Transformer stages for global contextual representation. A100 uses pre-training with the integration of DINO self-distillation, iBOT masked patch reconstruction, and KoLeo uniformity regularization, resulting in features that are better than ImageNet-12K supervised representations for agricultural applications without any fine-tuning. By utilizing a frozen backbone, the accuracy for crop classification of A100 is 95.37%, which is +2.56% better than ImageNet-12K, and reduces the leaf count error by 10.4% on the GroMo25 phenotyping benchmark, proving that the same backbone can be used for classification and regression tasks. After full fine-tuning, the performance of A100 is 99.90% on 32-class crop identification and 99.02% on 44-class disease detection. On the difficult PlantDoc dataset, A100 outperforms all baselines, such as CvT-13 (64.89% against 63.56%). Remarkably, A100 obtains 98.01% accuracy on plant classification using just 1% of the labelled data (about 284 images per category), showing great label efficiency for use cases with limited labeling. Under an identical protocol, A100 also outperforms the publicly released agricultural foundation models FoMo4Wheat and LeafVision by more than 14 points at 1% labels. We plan to publicly release our pretrained model to allow agricultural professionals to adapt A100 for various plant analysis applications using limited annotations.
[CV-67] UniField: A Unified Agent ic Framework with Loop Reasoning for MRI Field-Strength Enhancement
链接: https://arxiv.org/abs/2610.22982
作者: Yiyang Lin,Chenhui Wang,Tianliang Yao,Shuaibo Li,Zijun Sun,Yiming Deng,Yixuan Yuan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 6 figures, 8 tables
Abstract:MRI field-strength enhancement seeks to combine the accessibility of low-field scanners with the image quality of high-field scanners. Yet, existing methods remain prone to anatomical hallucinations, as uniform processing leaves difficult brain regions insufficiently examined and scarce paired data limit their learning of how tissue contrast and anatomical visibility change across field strengths. To address these challenges, we propose UniField++, a two-agent framework that unifies enhancement tasks across multiple field-strength transitions, including 64mT-to-3T and 3T-to-7T. Unified learning allows both the VLM-based planner and the flow-matching-based enhancer to draw on field-dependent image characteristics observed across tasks, rather than relying on a single transition. Trained with radiologists’ annotations of hallucination-prone regions and corresponding correction plans, the planner predicts regional risk and formulates targeted restoration strategies. Guided by these plans, the enhancer performs loop reasoning along the flow-matching trajectory, repeatedly processing high-risk regions for additional refinement. Together, cross-transition learning and risk-guided refinement support anatomically faithful enhancement across field strengths. Across the evaluated tasks, UniField++ improves PSNR by an average of 2.74 dB and SSIM by 14.05 percentage points over comparison methods, with corresponding gains of 3.97 dB and 13.92 percentage points over the ablation baseline. It also maintains enhancement quality on an external cohort with an unseen field strength and improves downstream whole-brain segmentation Macro Dice by an average of 0.1457 over the original low-field images. Code is available at this https URL.
[CV-68] How Murky Before a VLM Knows It Cannot See? Accuracy Confidence and Abstention on a Physical Visibility Axis
链接: https://arxiv.org/abs/2610.22981
作者: Mochamad Rizal Fauzan,Irgi Surya
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 16 pages, 9 figures, 10 tables
Abstract:Robustness studies of vision-language models (VLMs) grade degradation on nominal severity scales (“level 1-5”) that carry no physical units, and they rarely ask whether a model’s confidence follows the loss of visibility. We measure accuracy, internal (token-probability) confidence, verbalized confidence and an explicit abstain option on a visibility axis in metres: fog follows the Koschmieder model, an underwater arm scales the absorption spectrum of pure water, both are applied with a monocular metric depth map, and the visibility range V is swept over eight log-spaced levels from 250 m to 1.95 m. On 1000 COCO scenes and five small open VLMs (2-5 B parameters), accuracy on a boxed-object identification task falls from near its clean value to chance within a narrow band (half-way visibility V_acc of 12.1-15.9 m in fog and 20.5-38.6 m under water) while, for the four models that are confident on clean images, the confidence stays high: at V=1.95 m in fog their accuracy is at chance but their mean internal confidence is 0.68-0.84 and their verbalized confidence 0.61-1.00, a blind-spot width W=log10(V_acc/V_aware) of at least 0.79 decades, i.e. a factor of 6.2 in visibility. The remaining model is under-confident throughout and its internal confidence follows accuracy. An explicit abstain option follows accuracy when it names the reason (W from -0.21 to -0.06 in fog) but, for one model, lags with a generic “I don’t know” (W up to 0.18). ImageNet-C fog severities 1-5 correspond to only 55-110 m of physical visibility, outside the failure band, and leave the accuracy of the models within 5 percentage points of its clean value. Blank-image and shuffled-image controls, a prior-matched answer set and a 200-image pilot that preceded the study are reported. Findings are descriptive: one draw of 1000 images, depth of assumed scale, and models of 2-5 B parameters.
[CV-69] Sink-Sparse Hybrid Attention for Efficient Inference in T2I/T2V Diffusion Transformers
链接: https://arxiv.org/abs/2610.22975
作者: Bingxin Xing,Yu Zhang,Jiatong Ding,Lei Lu,Ping Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 12 pages, 6 figures
Abstract:Diffusion Transformers (DiTs) have become a common backbone for high-quality text-to-image and text-to-video generation, but repeated dense self-attention remains a major inference bottleneck as token sequences grow and the same operator is invoked across many denoising steps. Our study suggests that dense attention in pretrained DiTs contains reusable layer/head-specific high-mass structure that can be identified from a small offline calibration set and reused across future generations. This leads to Sink-Sparse Hybrid Attention, a train-free attention replacement that reuses static sink masks and supplements them with a lightweight pooled-KV branch, while leaving model weights and sampling schedules unchanged. On Wan2.1-T2V-1.3B, the method attains 1.54 \times end-to-end speedup while maintaining generation quality. On Z-Image-Turbo, it attains 1.57 \times end-to-end speedup with competitive GenEval2 performance. These results indicate that reusable importance-aware sparse support can provide an effective train-free acceleration path for both T2V and T2I DiTs.
[CV-70] ANCHOR: Audio-Visually Grounded Chain-of-Thought Reasoning Benchmark NEURIPS2026
链接: https://arxiv.org/abs/2610.22964
作者: Joel Julin,Souraja Kundu,Liza Dahiya,George Z. Wei,Liuyue Xie,Jingru Yi,Xu Zhang,Hao Yang,Laszlo A. Jeni
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at NeurIPS 2026, Evaluations Datasets Track. Joel Julin and Souraja Kundu are co-first authors
Abstract:Multimodal large language models (MLLMs) perform well on audio-visual question answering, but correct answers do not necessarily imply that their stated reasoning is grounded in relevant perceptual evidence. Existing benchmarks mainly evaluate final-answer accuracy, offering limited insight into whether model-generated rationales correctly identify visual regions, temporal intervals, and audio events supporting predictions. We introduce ANCHOR, a benchmark and evaluation framework for measuring grounded-rationale alignment in multimodal reasoning, comprising 2,319 videos and 8,272 question-answer pairs with structured reasoning traces, temporal evidence, spatial bounding boxes, and salient audio cues, including a human-refined test set of 224 videos and 975 question-answer pairs. We separately measure answer correctness, hallucination, logical divergence, and conciseness, with direct spatial and temporal grounding measurements on a focused subset. Across the full benchmark, twelve contemporary MLLMs show a substantial alignment gap: models with comparable answer performance can differ markedly in whether their explanations are supported by annotated evidence. On a matched subset of 100 videos and 446 question-answer pairs, rankings remain highly stable across human-refined, Gemini-generated, and GPT-generated references (mean Pearson r=0.99), providing evidence against model-reference circularity within this subset. Grounding-cue ablations of three MLLMs on the full human-refined test set reveal a consistent modality hierarchy: removing video reduces accuracy by 38.4-41.8 points, while removing bounding boxes (3.2-7.7 points) and audio (4.2-5.6 points) causes smaller, complementary, model-dependent drops. These findings show answer-only evaluation obscures important grounding failures and position ANCHOR as a testbed for evaluating accurate and evidence-aligned audio-visual reasoning.
[CV-71] Query-adaptive Recurrent Tokenization
链接: https://arxiv.org/abs/2610.22957
作者: Bavesh Balaji,Pranav Gopinath,Pouya Sadeghi,Sirisha Rambhatla
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Visual information worth extracting from an image and the token budget it warrants depend on the text query. Yet modern vision and vision-language models largely encode images independently of the query and allocate a fixed visual token budget to every input. We introduce the Query-Adaptive Recurrent Tokenizer (QART), a backbone-agnostic module that makes visual representations both query-steerable and budget-adaptive. QART steers the visual representation toward query-relevant evidence while adapting its token budget to the amount of information required. Its key principle is simple. Additional tokens are allocated until the visual features relevant to the query are represented with sufficient fidelity. Starting from frozen encoder features, QART recurrently constructs a compact representation and uses reconstruction quality over the query-relevant visual support to control token allocation. On Visual7W and VQAv2, QART achieves 81.96% and 76.37% accuracy using only about 29 visual tokens, reducing token counts by 20-25x relative to dense VLM inputs. It further reaches 92.0% P@1 on CORE and 36.3% PR-AUC on PODS, outperforming SteerViT on PODS by 8.4%. Together, these results show that visual representations need not be fixed in either content or capacity and can instead adapt to what each query requires.
[CV-72] MambaAct: Efficient Mamba Action Experts for Flow-Matching Vision-Language-Action Models
链接: https://arxiv.org/abs/2610.22951
作者: Khuyen Pham,Tran Nguyen Le,Edoardo Fazzari,Cesare Stefanini
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-language-action (VLA) models have shown strong promise for generalist robotic manipulation, but their inference cost remains a major bottleneck for real-time control. This limitation is especially pronounced in flow-matching VLAs such as \pi_0.5 , where the action expert is evaluated repeatedly during iterative denoising. We introduce MambaAct, an efficient Mamba action expert for flow-matching VLA policies. MambaAct retains the pretrained visual-language prefix encoder of \pi_0.5 , while replacing its transformer-based action expert with a compact Mamba-based action expert. Within this expert, Mamba serves as the action mixer that models dependencies across the noisy continuous-action sequence. To preserve visual-language conditioning without repeated token-level cross-attention, MambaAct uses a lightweight Context Compressor to map Vision-language model (VLM) prefix tokens into a compact conditioning signal that gates and modulates each Mamba action block. MambaAct uses a 10\times smaller action expert and reduces flow-matching denoising latency by 64.8% on LIBERO and 58.8% in real-world deployment, while improving mean LIBERO success by 1.08 percentage points across three independent fine-tuning seeds and maintaining gains on LIBERO-Plus and real-world benchmarks. Our code and model weights are publicly available at this https URL.
[CV-73] DICL-Qwen 3VL: Decoupled Confidence Calibration for Interpretable Workwear Detection
链接: https://arxiv.org/abs/2610.22929
作者: Meng Wang,Wu Zeng,Mei Li,Chao He
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Poor illumination and occlusion obscure clothing cues in indoor work sites, while diverse clothing styles complicate workwear compliance judgments. Although vision-language models (VLMs) support localization and semantic reasoning, their predictions often lack explicit supporting evidence, and a single confidence score cannot distinguish visual uncertainty from ambiguity in compliance judgments. We propose DICL-Qwen3VL, a decoupled interpretable VLM that integrates structured workwear detection, evidence generation, and separate visual and reasoning confidence estimates. First, low-rank adaptation (LoRA) adapts Qwen3-VL to localize visible upper-body clothing and identify workwear. Second, decoupled calibration separates visual observations from compliance reasoning and assigns a confidence score to each. Token entropy and sensitivity to image perturbations provide supervision for visual confidence, linking confidence estimation to the generated observations and their dependence on image content. Finally, group relative policy optimization (GRPO) jointly optimizes detection accuracy, output validity, and confidence calibration. Experiments on the company-collected Indoor Construction Site Workwear Dataset (ICSWD) and the web-collected Indoor Construction Workwear Dataset (WICWD) yield mAP@50 values of 0.730 and 0.753, respectively, outperforming the evaluated LoRA-adapted VLM baselines, including Qwen3-VL. The framework also provides explicit evidence and separate confidence scores to support manual review in industrial safety monitoring. WICWD is publicly available at this https URL.
[CV-74] ODSS-Net: Orthogonally Decoupled Shared-Specific Representations for Multi-Modal Object Re-Identification ICASSP2027
链接: https://arxiv.org/abs/2610.22922
作者: Zhifeng Ma,Yuhong Feng,Chongyi Huang,Mingyang Zhou,Minwen Deng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ICASSP 2027 conference submission
Abstract:Multi-modal object re-identification (ReID) leverages complementary information from heterogeneous sensors. Cross-modal alignment and fusion remain challenging. Existing methods mainly reduce modality gaps via feature alignment and fusion. However, alignment may entangle modality-shared structural information with modality-specific cues, while conventional fusion may not fully exploit complementary information. To address these issues, we propose ODSS-Net, a framework for orthogonally decoupled shared-specific representation learning and adaptive fusion of complementary modality-specific features. First, a Shared Structural Feature Construction (SSFC) module suppresses modality-dependent appearance variations and extracts modality-shared structural information. Then, an Orthogonal Specific Feature Learning (OSFL) module aggregates identity-relevant local information and removes the shared-aligned component via geometric orthogonal projection. Finally, a Sample Adaptive Mixture of Experts (SAMoE) module uses the shared structure as routing context to adaptively refine and fuse modality-specific features. Experiments on three benchmarks demonstrate the effectiveness of ODSS-Net for multi-modal object ReID.
[CV-75] MaskMerge: Structure-aware Token Merging for Efficient Instance Segmentation
链接: https://arxiv.org/abs/2610.22921
作者: Bavesh Balaji,Letian Jiang,Alexander Wong,David Clausi,Sirisha Rambhatla
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Token merging improves the efficiency of Vision Transformers by combining redundant tokens, but existing methods largely rely on feature similarity to determine which tokens to merge. This criterion is effective for semantic segmentation, yet can fail in instance segmentation, where adjacent objects of the same class often have highly similar features. We identify this failure mode as cross-instance token fusion. We introduce MaskMerge, an instance-aware token merging method that uses coarse intermediate mask predictions to constrain feature-based merging and preserve instance boundaries. We evaluate MaskMerge on two diverse instance segmentation domains characterized by high inter-instance visual similarity: PanNuke and iSAID. Against the SOTA token merging method ALGM, MaskMerge improves mAP from 30.1 to 32.1 on PanNuke while reducing computation from 229 to 153 GFLOPs and running 1.18x faster, and from 33.1 to 36.2 on iSAID while reducing computation from 198 to 182 GFLOPs and running 1.23x faster. Additionally, we also test it on Cityscapes and match ALGM’s performance while using fewer FLOPs. The gains also transfer to an MAE pretrained backbone, demonstrating that instance-aware merging enables more aggressive compression when appearance similarity alone cannot reliably preserve object identity.
[CV-76] COMET: Decoupled Distillation Routing and Capacity Control for Task-Agnostic Continual Vision–Language Learning NEURIPS2026
链接: https://arxiv.org/abs/2610.22907
作者: Seungyoon Woo,Keonhee Park,Gunhee Kim
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted at NeurIPS 2026
Abstract:Vision-language models (VLMs) such as CLIP retain zero-shot recognition ability, but their open-world use often requires task-agnostic continual learning: tasks arrive sequentially, past data are unavailable, task identity is unknown, and the backbone cannot be retrained. We argue that this setting fails VLMs through three coupled pressures beyond the stability-plasticity tradeoff: (1) modal-consistency drift, (2) representation-space interference, and (3) per-task capacity shortfall. We propose COMET (Collaborative Mixture-of-Expert Transfer) to resolve these three issues respectively as follows. A frozen teacher provides only a feature reference on a shared image-text pool; Tri-Affinity Distillation preserves image-image, text-text, and image-text geometry; and Student-Aware Cross-Expert Distillation aligns new experts with compatible prior experts so routing errors degrade gracefully. We design COMET to keep capacity and inference separate: a contextual bandit keeps or merges full per-task experts using post-training signals, while a reconstruction router selects sparse experts at test time without teacher semantics. On the X-TAIL benchmark, we show that each proposed component improves continual learning performance. COMET shows that continual vision-language adaptation can preserve multimodal geometry, coordinate accumulated experts, and allocate capacity jointly when feature reference, expert shaping, routing, and capacity control are separated by construction.
[CV-77] HAPACT: A Benchmark For Human-Centric Physical Impact Localization in Movies NEURIPS2026
链接: https://arxiv.org/abs/2610.22895
作者: Youngrae Kim,Yejin Jang,Eunho Kim,Youjin Sung,Kun-Woo Song,Sangpil Kim,Sang Ho Yoon
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 29 pages, 16 figures. Accepted at NeurIPS 2026 (Evaluations Datasets Track)
Abstract:Physical reasoning about humans in videos has largely focused on everyday interactions, where contact, pressure, pose, and motion support embodied perception and contact-aware applications. Still, short and forceful properties like impact events remain relatively underexplored, despite their relevance to broader understanding of how abrupt physical events affect the human body. To fill this gap, we introduce human-centric impact localization, a new task that localizes target physical impacts temporally within a video and spatially on the human body. We further present HAPACT, a HumAn-centric imPACT dataset and benchmark built from movie clips, containing over 2.8K video-query pairs with more than 20K fine-grained human-annotated frames. We built HAPACT around viewer-grounded annotation, text-conditioned localization, and supervision at frame-level temporal and SMPL-X-based spatial granularity. To demonstrate the effectiveness of our benchmark, we conduct extensive evaluations of existing methods, employing video moment retrieval models for temporal localization and human-scene contact models for spatial localization. For the spatial task, we propose a text-conditioned baseline by reformulating localization as query-guided body-region selection. Benchmark results show that physical impact localization remains challenging for existing methods, motivating HAPACT as a dedicated benchmark to support physical impact reasoning in videos.
[CV-78] Variational and Decomposed method for Mitigating Bimodal Bias in Robust Visual Question Answering
链接: https://arxiv.org/abs/2610.22845
作者: Yihan Zhao,Yuncheng Jiang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages
Abstract:Visual question answering (VQA) models are known to suffer from the bias issue, which refers to overfitting the regularities of frequent patterns. Previously, researchers only consider the language bias. Recently, the multimodal bias is noticed. However, what multimodal bias is remains unclear. In this paper, we investigate the multimodal bias in counting and non-counting questions. We find that they have distinct underlying mechanism, so their biases come from distinct reasons. We provide valuable interpretations of VQA counting mechanism: \textbfquantity information stems from the magnitude of the multimodal features in specific dimensions. Thus, counting bias stems from the vision side and propagates to multimodal features through modality fusion, while the non-counting bias stems from primarily language bias and slightly vision bias. Thus, we propose a method to deal with them differently: Variational and Decomposed method for mitigating Bimodal Bias (VDB). It extracts bimodal bias affected by different proportion of vision and language sides. We introduce variational bias models utilizing the principle of Variational Auto-Encoder to learn mainly language bias with slight vision effect, and a bimodal bias model for bimodal bias combining language and vision sides. Extensive experiments on VQACPv2, VQACPv1, and GQAOOD datasets demonstrate the rationality of our claim and the effectiveness of our method. We also provide explanation for the compromised performance on VQAv2.
[CV-79] VTrajEmbed: Aligning Text with Video Object Trajectories via Spatiotemporal Embeddings
链接: https://arxiv.org/abs/2610.22842
作者: Ruizhi Chen,Zining Fan,Yudi Guo,Jiang Liu,Zhiyu Lin,Zhongjiang He,Shouqian Shi,Hao Sun
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 28 pages, 1 figure
Abstract:Fine-grained alignment between text and video object trajectories is a core challenge in multimodal understanding. Existing embedding models excel at global retrieval, yet their single global representation entangles multiple objects and their spatiotemporal dynamics, limiting direct object-level grounding. To address this issue, we propose VTrajEmbed, an embedding model that decomposes a video into object trajectories and aligns each trajectory embedding with text for object-level grounding through retrieval. Our main contributions: (1)Slow-Fast Trajectory Encoding, a dual-branch architecture that establishes trajectory representations by densely sampling low-resolution frames for high-frequency motion cues and sparsely selecting high-resolution frames for scene context, aggregated by a shared LLM backbone; (2)Mask-Guided Contrastive Learning, a training objective built upon these representations that couples an intra-group loss and a batch-level bidirectional loss under a relation mask, excluding unverified relations and preventing confirmed matches from becoming false negatives; (3)Text2Traj-Bench, a trajectory-level text-video benchmark with model-assisted construction and human-verified query–trajectory relations. Experimental results show that VTrajEmbed-5B improves text-to-trajectory and trajectory-to-text R@1 over Qwen3-VL-Embedding-8B by 15.48 and 18.05 percentage points, respectively, and achieves 49.48% average R@1 across five global retrieval benchmarks, compared with 42.86% for Qwen3-VL-Embedding-8B.
[CV-80] VISTA: A Multimodal Egocentric Benchmark for Goal-Oriented Perception and Assistance to Blind and Low-Vision Users
链接: https://arxiv.org/abs/2610.22822
作者: Noé Mosca,Mohamed Rayan Barhdadi,Muhammad Hamza,Joshua Yeh,Suryodaya Bikram Shahi,Zhiyan Li,Selim-Antoine Lali,Weiyue Li,Hasan Kurban,Erchin Serpedin,Gang Luo,Yue Zhang,Minghan Li,Mengyu Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Code and dataset are available at this https URL
Abstract:Blindness and low vision limit many people’s ability to complete daily tasks, motivating AI assistants that provide useful guidance. Yet, existing benchmarks do not adequately test whether vision models can provide effective everyday assistance. In particular, i) existing benchmarks primarily rely on visual and textual inputs, with limited support for other modalities that are increasingly available in egocentric wearable devices such as gaze, inertial measurements, spatial trajectories, and others; and ii) existing evaluations emphasize scene understanding and question answering, with limited coverage of goal-conditioned guidance for blind and low-vision users across everyday tasks. To bridge this gap, we introduce VISTA (Visually Impaired Scene and Task Annotations), a multimodal egocentric benchmark for goal-oriented assistance for blind and low-vision users. VISTA contains \sim 1,000 egocentric recordings captured with Project Aria glasses; each recording provides synchronized sensor streams, six of which serve as model inputs, and three different types of annotation, covering ten task categories (TC), from object localization to hazard detection. We evaluate 14 representative vision-language models zero-shot on VISTA and design a lightweight multimodal model baseline that integrates all six modalities. Experimental results show that, although existing models can often describe egocentric scenes, they remain limited in generating goal-directed assistive guidance. VISTA thus provides a benchmark for evaluating multimodal models that could improve everyday accessibility.
[CV-81] Replay Pruning and Few-Bit Integer Inference for Hardware-Efficient Speed-Sign Recognition
链接: https://arxiv.org/abs/2610.22811
作者: Se Un Park,Kyungkeon Chung,Huigeol Kim,Junyoung Park
类目: Computer Vision and Pattern Recognition (cs.CV); Hardware Architecture (cs.AR); Machine Learning (cs.LG)
备注:
Abstract:Speed-limit sign recognition on an automotive camera system-on-chip (SoC) must fit a detector and a classifier (the reader) into the memory, arithmetic, and energy budgets of the device. In our two-stage recognizer, the detector, a YOLO-family network with ReLU and hard-sigmoid operators, is compressed by channel replay: an iterative pruning cascade identifies the channels to remove, the recorded deletions are replayed on the original task-trained network, the parent, and one continuous knowledge-distillation recovery follows. On a YOLOv5n, replay removes 60.3% of the parameters at the accuracy of the parent and, in validation mAP50-95 over five matched seeds, outperforms iterative pruning and training from scratch, and its per-step variant (per-step replay) is the most accurate pruning method. On a YOLOv8n, replay and per-step replay outperform every other method and the unpruned parent. The reader, a five-layer CNN with 18k parameters, and the post-processing of the YOLOv5n and YOLOv8n heads use integer arithmetic only. With 4-bit weights and 8-bit activations (W4A8), the reader matches FP32 accuracy in 12.1 KB of on-chip memory, and an executable streaming model reproduces the INT8 class predictions of its runtime on all 9,850 test crops. On a Jetson AGX Orin, INT8 is faster than FP16 only in the implicit formulation of TensorRT, and the replay detector runs fastest on the deep-learning accelerator (DLA). Trained on public data only, the combined system of 0.72 M parameters recognizes 95.2% of the signs of a held-out automotive camera at 90% precision and 94 frames per second.
[CV-82] STRIDE: Evaluating Spatiotemporal Reasoning in Driving Edge Cases
链接: https://arxiv.org/abs/2610.22795
作者: Lieqi Liu,Rui gao,Jia-Chen Gu,Wenbo Hu,Zhaobin Mo,Ahmadreza Moradipari,Nejib Ammar,Wei Wang,Nanyun Peng
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Project page: this https URL
Abstract:In autonomous driving, perceiving surrounding objects is not enough: failing to understand how they move and interact over time can lead to dangerous decisions. Evaluating vision-language models (VLMs) therefore requires testing whether they connect past spatial observations to future consequences. Yet existing benchmarks often assess edge cases through basic perception questions or test temporal understanding in simple scenarios, leaving complex spatiotemporal reasoning underexplored. We introduce STRIDE (SpatioTemporal Reasoning In Driving Edge Cases), an object-centric benchmark for long-tail, safety-critical driving scenarios. STRIDE covers six task families: spatial perception, spatial understanding, temporal memory, temporal extrapolation, trajectory prediction, and scene-context awareness. It pairs multiple-choice questions probing individual spatiotemporal judgments with related free-form visual question answering (VQA) requiring detailed scene explanations and justifications of driving decisions. We prioritize scenes with unusual object behaviors and complex traffic interactions, generate reference answers from structured dataset annotations through automated procedures, and use limited language-model assistance and human verification to ensure annotation quality. Across 12 proprietary, open-weight, and driving-specialized models, even the best performer, GPT-6-Astra, achieves only 43.4% average multiple-choice accuracy, versus an approximately 20% random-choice baseline. Further analysis reveals that strong performance in describing driving scenes does not necessarily translate to accurate safety-critical judgments, and that models struggle more with reconstructing object motion than recalling past positions. STRIDE thus exposes spatiotemporal driving limitations that aggregate VQA scores can conceal.
[CV-83] Direct Conditional Parameterization for N-Dimensional Splatting NEURIPS2026
链接: https://arxiv.org/abs/2610.22778
作者: Zhongpai Gao,Benjamin Planche,Van Nguyen Nguyen,Meng Zheng,Anwesa Choudhuri,Arun Innanje,Terrence Chen,Ziyan Wu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by NeurIPS 2026
Abstract:N-dimensional splatting extends 3D Gaussian Splatting with view direction and time to model view-dependent and dynamic effects. Each primitive is lifted into a joint distribution over position and conditioning variables, then conditionally sliced at each rendering query. N-DGS and UBS share this step across Gaussian and Beta kernels: recovering the conditional mean, opacity, and covariance requires per-primitive matrix inversions and regression products in every forward pass. We introduce direct conditional parameterization, replacing this covariance-derived readout with explicit per-effect parameters: a Cholesky precision factor for opacity and a factorized query-to-position regression operator with learnable per-dimension coupling. Unrestricted conditional coordinates recover a valid joint covariance; our factorization adds spatial-scale and coupling priors. Applied to both kernels, the parameterization yields dGS (Gaussian) and dBS (Beta). Across five static and dynamic benchmarks, they match or exceed their covariance-derived counterparts on quality, with gains up to +1.26 dB in the main MCMC evaluation. Production slicing is 6.9-7.7 times faster than on-the-fly covariance-derived conditioning; complete-render profiling gives 1.25-1.91 times speedups on four trained scenes. Direct conditioning provides a structured, kernel-agnostic alternative for N-dimensional splatting. Project page: this https URL.
[CV-84] Input- and Feature-Space Correction for Cross-Hospital IHC Grading NEURIPS2026
链接: https://arxiv.org/abs/2610.22756
作者: Jongwoon Lee,Jhair Gallardo,Christopher Kanan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 12 pages, 2 figures, 4 tables. Accepted at the NeurIPS 2026 Workshop on AI at Scale for Clinical Impact (ASCI)
Abstract:Pathology foundation models (FMs) are trained predominantly on hematoxylin and eosin (HE) images but are increasingly used as frozen feature extractors for immunohistochemistry (IHC), where cross-hospital differences can impair grading. However, systematic comparisons of correction methods for cross-hospital IHC grading remain limited. To address this gap, we benchmark label-free transformations applied to input images before frozen-FM encoding and to features after frozen-FM encoding using UNI and Virchow2, three IHC markers, and six directed cross-hospital transfers, yielding 12 settings. Among the evaluated input-space methods, our proposed method FIX achieved the largest macro-F1 improvement over no correction, outperforming well-known stain normalization techniques. Among feature-space methods, a simple feature-wise standardization achieved the best improvement over no correction, outperforming CORAL. Under patient-averaged tile-level evaluation, combining FIX with feature-wise standardization improved macro-F1 and quadratic weighted kappa (QWK) by 0.181 and 0.208, respectively, and reduced ordinal error by 0.209 relative to no correction. These results show that input- and feature-space transformations can provide complementary benefits for cross-hospital IHC grading with frozen pathology FMs.
[CV-85] MedSTP: Preserving Subspace and Transition Evidence for Medical VQA Token Pruning
链接: https://arxiv.org/abs/2610.22751
作者: Mengchen Fan,Qizhen Lan,Xi Xiao,Lin Zhao,Liang He,Tianyang Wang,Xiaoqian Jiang,Pulin Che,Baocheng Geng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Medical vision-language models (VLMs) for visual question answering (VQA) often pass long visual-token sequences to the language model, making visual-token pruning attractive but risky for medical images, where answers may depend on localized boundary and texture evidence. We study training free pre-decoding pruning and treat pruning as a subspace stable evidence preservation problem. The retained tokens should remain close to the dominant visual evidence subspace induced by the full token set. To this end, we propose \emphSubspace-Transition Token Pruning (\method), which estimates each token’s spectral support from principal component space and constructs a compact representation subspace. To better reflect medical image structure, \method further uses a bounded token-grid transition prior to promote boundary and texture sensitive tokens. We evaluate \method on the closed-ended SLAKE and OmniMedVQA benchmarks and on RadImageNet-VQA, which provides both closed-ended and open-ended evaluations, using Lingshu-7B and Qwen3-VL-8B backbones. \method delivers consistently strong performance on both the closed-ended and open-ended RadImageNet-VQA evaluations. These results show that \method is an effective training free pruning framework for medical domain VQA.
[CV-86] Cross-Person Identity Change Under Source-Coupled Multi-Human Generation: A Controlled UniPortrait Study
链接: https://arxiv.org/abs/2610.22746
作者: Edrees Al-Edreesi,Ghaleb Al-Gaphari,Muneer Alsurori,Mohammed Edrees,Fouad Al-Hamdani
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 16 pages. Project website and code: this https URL
Abstract:Multi-human personalization must account for identity changes in people outside the nominal target of a generative operation. We characterize this behavior in UniPortrait, without introducing a new preservation algorithm. Two sequential studies use 119 accepted two-person baseline scenes and three fixed output seeds, measuring continuous reference-based facial similarity with InsightFace \buffalo. In Study~1, source-coupled single-ID target-only generation (A1) produces negative mean transitions for both the protected person ( -0.567904 ) and target ( -0.236904 ), indicating broad perturbation rather than established successful repair with collateral loss. In Study~2, native dual-ID generation (A2) yields more positive paired contrasts: protected +0.512918 (95% case-cluster CI [0.471434,0.550259] ) and target +0.158853 ( [0.118903,0.197523] ), across 287 jointly measurable pairs from 111 cases. Shared baseline similarity cancels algebraically in these paired contrasts, leaving a matched POST comparison. Validation designed after the primary result freeze finds directional agreement with an independent identity evaluator and blinded human judgments; within-\buffalo\ nuisance models provide supportive sensitivity evidence. The findings characterize intact native regimes, whose references, routing, adapters, and token counts co-vary. They do not establish component causality, categorical identity preservation, successful localized editing, or generalization beyond the evaluated implementation and measurable cohort.
[CV-87] OmniXRef: A Unified Framework for Multimodal Multi-Reference Video Editing
链接: https://arxiv.org/abs/2610.22744
作者: Xiaoyan Cong,Tianyu Wang,Hyeonho Jeong,Zhifei Zhang,Yuqian Zhou,Yijun Li,Shaoteng Liu,Zhe Lin,Srinath Sridhar,Soo Ye Kim
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Creative intent in video editing can be expressed through text instructions (e.g., “remove the people”), image references (e.g., subject appearance), video references (e.g., camera movement), or a combination of the above. We introduce OmniXRef, a unified framework for multimodal multi-reference video editing that allows users to combine variable numbers of text, image, and video references to control diverse attributes of a source video. We first devise a unified input design with a reference-bank interface and introduce R^2 -RoPE to support generalization to more references. Then our model, OmniXRef-Model, uses a single-stream diffusion transformer to jointly process the source video and multimodal context from an MLLM in an interleaved sequence. To address the scarcity of training data for video-reference editing, we construct OmniXRef-Data using synthesis pipelines for different editing tasks and combinations of references. We use curriculum learning to progressively increase task complexity and resolution. To support systematic evaluation of video as a control modality, we introduce OmniXRef-Bench, a benchmark for video-reference attribute transfer across Camera, Motion, Content, Style, and Visual Effects (VFX). Comprehensive evaluations show that our OmniXRef-Model outperforms task-specific baselines and commercial models.
[CV-88] HilbertGen-3D:Hilbert–Multifractal Conditioning for Topology-Aware 3D Generation
链接: https://arxiv.org/abs/2610.22720
作者: Ziyi Zhang,Shixuan Li,Yorie Nakahira,Paul Bogdan,Guannan Qu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:High-fidelity 3D generative modeling underpins game asset creation, medical imaging, and VR/AR environmental design, where geometric structure directly influences perception and decision-making. Diffusion models represent a powerful paradigm for point cloud generation. However, existing approaches rely heavily on local voxel or patch features and often miss global structural organization, and much less work has studied how global conditioning can be used to improve generation quality. In this paper, we introduce HilbertGen-3D, a novel topology-aware conditional diffusion framework based on Hilbert-Multifractal Conditioning (HMC). HMC computes multi-scale voxel measures, serializes them via a 3D Hilbert curve to expose structured spatial dependencies, and estimates the generalized fractal dimensions from scale-dependent partition functions. We encode these signals as global embeddings and multi-scale tokens, and inject them into a diffusion denoiser through an efficient bottleneck fusion module. Across the Chair, Airplane, and Car categories of ShapeNet, our two conditioning variants reduce the absolute deviation of 1-Nearest Neighbor Accuracy (1-NNA) from 50% by up to 8.90 percentage points and increase coverage by up to 5.88 percentage points relative to TopoDiT-3D. This shows that combining local density cues with global Hilbert-multifractal structure provides a simple and effective recipe for high-fidelity and diverse 3D generation.
[CV-89] hinking from Above: Efficient Spatial Reasoning with Birds-Eye-View Visual Tokens
链接: https://arxiv.org/abs/2610.22700
作者: Faizan Siddiqui,Nirav Savaliya,Sahith Reddy Chada,Enna Sachdeva
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Existing video multimodal large language models (MLLMs) organize observations primarily in time, but embodied spatial reasoning requires organizing them in space. Consequently, their visual representations can remain locally consistent across nearby frames while lacking a persistent spatial context that is stable across viewpoints and time. We exploit depth and camera pose, which are naturally available to embodied agents, to explicitly establish this spatial correspondence. We introduce Cambrian-BEV, which extends Cambrian-S with a parameter-free geometric tokenization scheme that back-projects visual patch features from each frame into a shared bird’s-eye-view (BEV) coordinate frame and averages features that fall within the same spatial cell. Thus, observations of the same physical location, regardless of viewpoint or time step, are aggregated into a common spatial token. In addition to providing a persistent scene representation, this formulation acts as an efficient compression mechanism: 128 frames containing approximately 100K DINOv3 patch features are consolidated into a fixed 27 \times 27 grid of only 729 BEV tokens, cutting LLM prefill time by 2.9\times and peak memory by 3.6\times . Cambrian-BEV requires no architectural modification beyond BEV tokenization and uses a frozen DINOv3 vision encoder with a trainable multimodal projector and LLM. On VSI-Bench, our model reaches 63.48, +8.5 points over an identically fine-tuned frame-token control with 12.6\times fewer visual tokens, with particularly large gains on room size ( +24.5 ) and relative direction ( +34.1 ). On ReVSI, it reaches 44.68, +4.5 over the corresponding control.
[CV-90] EgoP2C: Internalizing Plan-Perceive-Compose for Spatial Reasoning in Egocentric Video
链接: https://arxiv.org/abs/2610.22681
作者: Chaoyu Li,Arka Sadhu,Yuliang Li,Hyo Jin Kim,Yuning Chai,Pooyan Fazli
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Answering complex spatial reasoning questions from egocentric video requires determining what evidence is needed, acquiring targeted observations, and combining them through multiple reasoning steps. Existing vision-language models (VLMs) struggle to perform this process reliably. Agentic systems address this challenge by querying external vision and geometry tools, but training their reasoning leaves these tools unchanged, so inaccurate observations can still lead to incorrect answers. We introduce EgoP ^2 C, a framework that internalizes a plan-perceive-compose process within a single VLM. To this end, we first train specialized Planner and Perception teachers using separate adapters (Dual-LoRA), then fine-tune the corresponding experts in a mixture-of-experts (MoE) student under their supervision. We subsequently train the student on trajectories generated by its current policy using a segment-level extension of multi-teacher on-policy distillation (MOPD), assigning different trajectory segments to different supervisors. The Planner teacher supervises which evidence-seeking questions to ask, when to stop gathering evidence, and how to compose the final answer. The Perception teacher supervises the video-grounded answers to those questions. At inference, the student executes this adaptive workflow end-to-end, without external spatial tools. Across three spatial reasoning benchmarks, EgoP ^2 C students built from Qwen3-VL-8B and SenseNova-SI-1.3-8B outperform their base models by an average of 9.2 and 7.5 percentage points, respectively. Across both backbones, EgoP ^2 C reduces inference FLOPs by 65.8-66.3% and achieves 1.29 - 1.49\times speedups over their shared Dual-LoRA teacher.
[CV-91] Deep Learning for Deepfake Image Detection: Comparative Model Evaluation and Gradient-Based Visual Attribution
链接: https://arxiv.org/abs/2610.22676
作者: Saad Korchi,Abderrahim Waga
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 3 figures, 1 table
Abstract:Synthetic facial imagery has made the distinction between authentic and generated content increasingly important for digital forensics and online trust. This study examines an explainable deep-learning workflow for image-level deepfake detection using two contrasting classification strategies: a compact convolutional model trained from scratch and an ImageNet-pretrained MobileNetV2 model adapted through two-stage transfer learning. Experiments were conducted with the balanced 140k Real and Fake Faces dataset, comprising 140,000 facial images. Under the recorded validation protocol, the augmented from-scratch model reached 84.15% accuracy (loss 0.3625), compared with 95.90% accuracy (loss 0.1074) for the fine-tuned pretrained model. Gradient-weighted class activation mapping (Grad-CAM) was used to visualize image regions associated with predictions, and a Flask-based prototype was developed for side-by-side inspection. These observations support the practical value of pretrained visual representations in the studied setting while underscoring the need to interpret validation performance alongside dataset provenance, explanation limitations, and evaluation scope. The reported results are validation results, not held-out test estimates.
[CV-92] Shift-Reliance Alignment: Ranking Vision-Language Models by the Cues They Trust
链接: https://arxiv.org/abs/2610.22610
作者: Faisal Mohammad,Mohamed Younis
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 24 pages, 12 figures, 21 tables
Abstract:Vision-language models (VLMs) are usually chosen from leaderboards, which give one ranking for every application. On questions about objects that a scene suggests but does not contain, however, the model with the lowest error on the POPE hallucination benchmark is not the one with the lowest error on such questions. For yes/no questions about object presence, we show that the best model for a target domain can be predicted without running any candidate on that domain, given 650 to 1,200 labeled target items that calibrate a detector. We measure each model’s reliance on a cue, a feature such as scene context or text in the image, by editing 100 labeled source images so that the cue suggests the wrong answer, and comparing with a matched control edit. Reliance predicts which VLMs are misled by physical sticker attacks (Spearman .93 on nine models not used to design the score) and which hallucinate on unedited COCO photographs, beyond what their tendency to answer yes explains. Shift-Reliance Alignment (SRA) scores a model by its source error on the labeled source questions plus its reliance multiplied by the shift. The shift is the proportion of target questions on which the cue misleads, and a detector estimates it on the unlabeled target images. SRA selects the model with the lowest score. In pre-registered tests on new images where most questions mislead, it lowered the error of the selected model from 30.4% under selection by source error to 18.8% on Objects365 and OpenImages but, against its prediction, not on COCO. On CueDial, a benchmark we release that varies how often a cue misleads, selection by model confidence falls more than 40 percentage points behind the best model when overlaid text denies a visible object, and SRA stays within 3 points.
[CV-93] ProactGen: Learning Proactive Clarification for Video Generation
链接: https://arxiv.org/abs/2610.22608
作者: Zhixiao Wang,Yang Chen,Yifan Xu,Chenbin Li,Yue Chang,Kai Long,Fengming Dong,Limin Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Recent advances in video generation let users create high-quality videos from natural language descriptions, yet obtaining results that match user intent remains time-consuming: individual generations are computationally expensive, and underspecified prompts force repeated cycles of generation, inspection, and revision. Existing video prompt optimization methods improve generation through automatic rewriting, but cannot supply the ideas and preferences that were not in the initial description. We propose ProactGen, a video generation framework that learns proactive clarification, eliciting user intent through questions grounded in generated video drafts. Specifically, we train a vision-language model as a questioning policy with reinforcement learning: observing the initial prompt and a video draft, it learns to ask the questions most valuable to downstream generation, rewarded by the alignment between the final generated video and a target reference. We further find that effective questioning does not require drafts of final-output quality, and exploit this property by generating drafts with fewer denoising steps, keeping the added cost of clarification low. Experiments show that ProactGen improves target alignment over direct generation by at least 1.8x the gain of the strongest baselines. Moreover, a single round of ProactGen already exceeds the alignment that the strongest baseline reaches after four refinement rounds, in 29% of its time.
[CV-94] hree LiDARs Twenty-Five Datasets: Cross-Source Generalization in Tree-Species Recognition WACV2027
链接: https://arxiv.org/abs/2610.22606
作者: Jae Joong Lee,Bedrich Benes
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Graphics (cs.GR)
备注: Accepted to WACV 2027
Abstract:FOR-species20K is the largest public LiDAR benchmark for tree-species classification, pooling 25 independently collected terrestrial, mobile, and airborne laser-scanning datasets across 33 species. We introduce a multi-resolution, multi-view ConvNeXtV2 model that improves official hidden-test accuracy from 79.5% to 85.0%, ranking first at the time of submission. We then examine what this score measures. Because the hidden test is species-balanced, its overall accuracy is more comparable to a macro metric than to accuracy under the public set’s long-tailed distribution. Moreover, the official random split places every source dataset in both training and evaluation. A logistic probe recovers source identity from out-of-fold model probabilities with 69% accuracy, while geometric features identify the source above 90% within two widely shared species; 18 of the 33 species occur in only one source. Under a source-species-cell holdout, evaluated on 5,566 trees from eight eligible species that retain the same LiDAR-type training support in other sources, the unweighted species mean falls from 82% to 36%, and tree-weighted accuracy falls from 95% to 60%. Count-matched retraining and experiments with two alternative backbones indicate that this gap is not explained by reduced training support alone, while an external PureForest experiment shows a smaller but consistent grouped-split effect. These results establish the 85% leaderboard score as strong in-distribution performance, but not as an estimate of source-disjoint generalization, motivating balanced metrics and source-aware evaluation. The code is available at: this https URL
[CV-95] Stuck on Shuffle: On Attribute-Object Binding in Vision-Language Models
链接: https://arxiv.org/abs/2610.22602
作者: Netanel Y. Tamir,Michael Green,Issar Tzachor,Or Shimshi,Dvir Samuel,Or Patashnik,Rami Ben-Ari
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Retrieval is a fundamental task in multimodal systems: given an image or text query, the goal is to find the most relevant images or texts in a collection, often using a shared embedding space. Reliable retrieval requires these embeddings to capture not only which concepts are present, but also how they are composed. One key compositional challenge is attribute–object binding: encoding which attributes belong to which objects. Existing benchmarks provide a limited view of this capability, typically covering a subset of modality pairings, simple two-object swaps, or small attribute sets. Consequently, it remains unclear how robustly current models preserve attribute–object bindings across different modality pairings and more complex settings. We therefore introduce SABER, a controlled benchmark for evaluating attribute–object binding across all four image and text modality pairings, spanning diverse attribute types and scenes with up to ten objects. Evaluating models on SABER reveals substantial binding failures in both CLIP-style dual encoders and recent MLLM-based embedders. These findings motivate us to ask how binding can be improved efficiently. While multimodal supervision may be effective, collecting suitable images is difficult and costly. Text, on the other hand, is well suited to binding supervision: it is cheap to generate, supports a broader range of attributes and enables precisely controlled hard negatives. We thus finetune an MLLM-based embedder exclusively on synthetic attribute-binding text. Despite the absence of image supervision, the gains transfer to cross-modal and image–image retrieval while preserving general multimodal embedding performance. Our model achieves SOTA results on SABER while also improving on negation and instance-retrieval benchmarks, demonstrating broader benefits of our binding-focused supervision.
[CV-96] Seeing Seizures: A Large-Scale Empirical Study of Video and EEG Deep Learning for Automated Rodent Seizure Detection and Severity Grading
链接: https://arxiv.org/abs/2610.22566
作者: Alloy Das,Shreyan Ganguly,Nyzil Massey,Hung Phi Doan,Zhanhong Jiang,Thimmasettappa Thippeswamy,Soumik Sarkar
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Under Review
Abstract:Manual annotation of synchronized video and EEG recordings is a bottleneck in preclinical seizure research. We present an empirical study of seizure detection and behavioral severity grading using 24,131 paired clips from 601 sessions across 20 rats in a soman-induced epilepsy model. Ten video architectures and seven single-channel EEG models are evaluated independently on binary detection and three- and five-class severity tasks. In the session-exclusive reference comparison, R(2+1)D-18 and GRU achieve detection macro-F1 scores of 0.968 and 0.974, respectively; for five-class grading, the corresponding scores are 0.657 and 0.560. Subject-exclusive evaluation reduces five-class macro-F1 to 0.475 for video and 0.404 for EEG, highlighting the difficulty of grading unseen animals. Temporal-sampling experiments demonstrate the value of multiple frames, while detection performance varies relatively little across the evaluated video architectures. A separate video-only binary-detection evaluation on RodEpil achieves a mean seizure-class F1 of 0.979 after uniform re-encoding to reduce identified class-correlated encoding cues. The study characterizes existing models for behavior-defined severity labels under the evaluated protocol, motivating further investigation of expert-assisted annotation workflows. Our code and github page is available here.
[CV-97] Micro-Localization: High-Precision Mobile Robot Localization Using Microscopic Ground Texture
链接: https://arxiv.org/abs/2610.22563
作者: Aaron Wilhelm,Marina de Albuquerque Marcon,Emith Uyanwatte,Nils Napp
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We demonstrate high-precision global localization using natural ground texture viewed through microscope optics, achieving localization in the tens-of-micrometers regime. For the planar ground texture localization problem considered here, we relate optical magnification and image-space localization uncertainty to physical localization precision. Leveraging this relationship, we then develop a microscope-based localization system that addresses the calibration, camera-modeling, and focus challenges introduced by high-magnification imaging. We evaluate this system across several surface textures using a calibrated precision gantry. Across 1,007 held-out queries per surface, the system localizes every query and achieves translation RMSEs of 28 - 73 \mu m. Independent photometric registration agrees with the feature-based localization estimates to within approximately 9 \mu m RMSE. We further demonstrate localization on a low-cost mobile robot using the gantry-collected maps, and also show that the system can resolve micrometer-scale displacement caused by thermal expansion. These results show that microscopic natural surface texture can bring mobile localization into a precision regime typically associated with fixed positioning systems.
[CV-98] Generative AI Produces and Transmits Beauty Bias
链接: https://arxiv.org/abs/2610.22553
作者: AJ Alvero,Deepak Varuvel Dennison,Alexandra Crespin,Heeyon Kim,Alejandra Regla-Vargas,Avani Akerkar,Jasmine McNealy,René Kizilcec
类目: Computers and Society (cs.CY); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Societies reward people judged to be attractive by standards intertwined with gender and race. In this paper, we examine if and how generative AI may contribute to these dynamics. We audited 27 text-to-image generators released between 2022 and 2026 using simple prompts that never mention attractiveness or beauty, analyzing 51,300 generated faces and 160,196 photographs of real people across primary and supplementary comparisons. We used three automated measures and collected ratings from 550 human raters on a subset of images. Nearly all generators produced faces scoring above photographed faces under automated measures, and human raters judged the outputs of all 23 generators as significantly more attractive than photographed faces, even after correction for multiple comparisons. When focusing our analysis on gender, we find that 37% of female-prompted generated faces outscored the median professional fashion model. Across racial, ethnic, and national categories, groups associated with darker skin tones tended to receive lower attractiveness scores than groups associated with lighter skin tones. In a preregistered hiring simulation experiment (N = 570), exposure to AI-generated faces selected for high human-rated attractiveness lowered hireability and trustworthiness ratings and reduced interview advancement for an identical test set of candidates paired with photographs of real people; exposure to retouched photographs produced no such effect. Generative AI can reproduce racial hierarchies and gendered patterns of beauty and alter the context in which real people are judged.
[CV-99] CFC26: Building Evaluations for Deployment in Sonar-Based Fish Counting
链接: https://arxiv.org/abs/2610.22549
作者: Madison Van Horn,Suzanne Stathatos,Sevan Brodjian,Justin Kay,Kai Van Brunt,Erik Young,Sara Beery,Pietro Perona,Michael Hobley
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Accurate counts of fish populations are essential for monitoring ecosystems and making decisions for conservation, harvest limits, and river management. In many field deployments, these counts are obtained from sonar videos that experts must inspect manually, making large-scale monitoring slow, expensive, and difficult. Automated counting systems have been proposed as a scalable alternative, but so far this promise has been difficult to realize in deployment. Developing evaluations that better reflect real-world deployment conditions will help close this gap and drive improvements in counting performance in practice. To achieve this, we propose Caltech Fish Counting 2026 (CFC26), a new dataset with improved evaluation protocols that simulate deployment to seen and novel rivers, testing in-distribution and out-of-distribution performance. We expand the CFC22 dataset to eleven deployment locations across nine Pacific salmon river systems from Alaska to Northern California, capturing a broader geographical and ecological range of data. We additionally introduce two new evaluation metrics for counting, nMNE and nMANE, that mirror the operational workflow that experts use and help surface errors that frame-level detection metrics do not capture. Using these metrics, we reveal previously-concealed directional bias in upstream vs downstream counts, which would have direct and significant ecological implications if undetected. The dataset and code are publicly available on CaltechDATA and GitHub.
[CV-100] FlashControl: One-Step Controllable Generator via Distillation-Friendly Single-Stream Teachers NEURIPS2026
链接: https://arxiv.org/abs/2610.22537
作者: Ngan Nguyen,Dung Nguyen,Quan Dao,Dimitris N. Metaxas,Minh Hoai,Cuong Pham,Anh Tran
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at The 40th Annual Conference on Neural Information Processing Systems (NeurIPS 2026)
Abstract:Diffusion models have become a leading approach in visual generative modeling, particularly effective in text-to-image and controllable generation. Their main drawback is slow inference as it requires many sampling steps. While recent distillation methods enable one-step text-to-image generation, extending these techniques to controllable generation is still challenging. Existing controllable models, such as ControlNet, rely on dual-branch design that make distillation difficult. In this work, we adopt a simpler strategy. Instead of adding extra conditional branches, we finetune a selected subset of layers in the original text-to-image diffusion model to directly enable controllable generation. This results in DeltaControl, a single-flow, multi-step diffusion model capable of supporting spatial-condition generation while matching the performance of dual-branch approaches. Building on this teacher model, we further apply step-distillation to obtain FlashControl, a one-step model capable of spatial controllable generation with a single forward pass.
[CV-101] Learning the Shared Latent Structure of Cryo-EM Inverse Problems
链接: https://arxiv.org/abs/2610.22523
作者: Nithin Chalapathi,Theo Sternlieb,Ethan Han,Sonya M. Hanson,Pilar Cossio,Aditi S. Krishnapriyan,Ritwik Gupta
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Quantitative Methods (q-bio.QM)
备注:
Abstract:Understanding biomolecular function requires characterizing three-dimensional structure and conformational variability, for which cryogenic electron microscopy (cryo-EM) is a powerful tool. Cryo-EM produces noisy two-dimensional particle images in which molecular structure, particle pose, and imaging conditions are jointly encoded. Recovering biological information from these observations spans inverse problems including pose estimation, heterogeneity analysis, and reconstruction. These tasks are usually addressed independently despite shared protein structure and image formation, motivating a common representation across inverse problems. Since biologically meaningful differences may be spatially localized, we study self-distillation with complementary global and masked patch objectives. Because experimental dataset discrimination can reflect acquisition-specific cues, we develop controlled synthetic and experimental evaluations and find that structural discrimination benefits from dense spatial features. We introduce the EMber family of vision transformers, pre-trained on 50 million experimental particle images from 214 EMPIAR datasets together with simulated projections from 268,000 molecular assemblies. EMber improves upon existing methods at conformational and compositional discrimination by 14.85 and 14.26 percentage points, estimates conformational populations to within 1.59pp error with a linear probe, and produces particle assignments yielding reference-quality reconstructions in fewer epochs. We demonstrate how to learn representations that capture biological information for use across multiple cryo-EM inverse problems.
[CV-102] owards Reconstructing Geographically Diverse Architecture with 3D Foundation Models NEURIPS2026
链接: https://arxiv.org/abs/2610.22514
作者: Nicky Kriplani,Yiwen Zhang,Angelina Wang,Hadar Averbuch-Elor
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: NeurIPS 2026. Project Page: this https URL
Abstract:Internet-scale photo collections of real-world landmarks have driven progress in 3D computer vision yet remain highly challenging for modern 3D foundation models (3DFMs) that estimate scene structure in a single feed-forward pass. In this work, we introduce ArchWorld, a benchmark of 3D architectural landmarks with high geographic coverage and rich metadata. We use this benchmark to conduct a thorough error analysis of current 3DFMs to understand where they fall short and how they can be improved. We examine socially relevant disparities in model error and find that model performance varies by geographic region; however, unlike typical fairness regimes, the trends only partially match training data representation. Our investigation reveals that the observed geographic disparities reflect not only differences in data representation and scene size but also differences in region-specific architectural features and background distractors that make scenes from different regions inherently vary in difficulty. Leveraging insights from these model failings, we introduce an intervention scheme that improves 3DFM performance while simultaneously reducing geographic disparities.
[CV-103] riSAR-3D and PaintFusion: A Novel Trimodal LiDAR-RGB-Thermal Dataset and Method for 3D Person Detection in Search and Rescue
链接: https://arxiv.org/abs/2610.22509
作者: Markus Kramer,Jan Wachenfeld,Oskar von Stryk
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注:
Abstract:Reliable 3D person detection is important for rescue robotics to give first responders situational awareness. Yet most multimodal 3D detection research targets automotive scenarios in which pedestrians mostly appear in a narrow range of upright, near-road poses. It is therefore unclear how much of the reported performance transfers to rescue settings, as a person may be lying, partially occluded, or in unstructured indoor and outdoor environments. To investigate this gap, we introduce TriSAR-3D, a multimodal person detection dataset recorded with a high-resolution 128-channel LiDAR, an RGB camera, and a thermal camera. It captures varied poses in indoor as well as outdoor locations at different times of day to study their influence on 3D object detector performance. Furthermore, we propose a novel trimodal 3D detection pipeline, PaintFusion, that fuses PointPainting-based evidence using a simple MLP. We show that the derived method reaches 65.4 AP@50 and thereby outperforms strong LiDAR-only (CenterPoint, 53.4), RGB+LiDAR (BEVFusion, 60.5), and adapted trimodal (Tri-BEVFusion, 60.7) baselines on the dataset. At the same time, it is substantially faster than other fusion-based methods such as BEVFusion, with a per-frame latency of 51.6 ms compared to 101 ms. Additionally, we test how the performance of different 3D object detectors generalizes to unseen pose categories. Removing lying instances from training data collapses the lying-pose AP@50 to near zero across all evaluated detectors. The results indicate that dataset diversity exerts a large effect on 3D object detection performance. The dataset and code are available at this https URL.
[CV-104] KFS4LAM: Adaptive KeyFrame Selection with Structure-Aware Mapping for Endoscopic Gaussian Splatting SLAM
链接: https://arxiv.org/abs/2610.22505
作者: Xinwei Ju,Zhehua Mao,Tianyi Song,Danail Stoyanov,Sophia Bano,Francisco Vasconcelos
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 5 figures, 3 tables
Abstract:Gaussian-splatting Simultaneous Localization and Mapping (SLAM) augments conventional SLAM with a dense, renderable 3D scene representation, enabling high-quality reconstruction and novel-view synthesis. However, its localization, reconstruction, and rendering performance depends strongly on which observations are retained for mapping. Existing systems commonly rely on fixed temporal intervals or simple pose- and overlap-based criteria for keyframe selection. In colonoscopy, non-uniform camera motion, repetitive anatomy, specularities, and visual artifacts can make these criteria unreliable, leading to either redundant keyframes or missed informative viewpoints. We present KFS ^4 LAM (KeyFrame Selection with Structure-aware mapping for endoscopic Gaussian Splatting SLAM), which couples motion-guided adaptive keyframe selection with structurally supervised mapping and restricted temporal deformation. Accumulated motion guides adaptive keyframe selection, with complementary cues suppressing redundant or unreliable candidates. Density-aware mapping with anatomical contour supervision (colon folds) emphasizes structurally informative regions, while restricted temporal deformation accommodates local tissue dynamics without unconstrained geometric drift. We evaluate KFS ^4 LAM against Gaussian-SLAM baselines on C3VD and C3VD-v2, together with cross-backbone and component ablations. KFS ^4 LAM achieves strong overall performance across trajectory, rendering, and depth metrics, with pronounced gains on more challenging C3VD-v2 dataset. In the matched comparison with SplaTAM on C3VD-v2, it reduces rotational relative pose error by 20.3% and depth RMSE by 38.4%, while improving PSNR by 0.78 dB. Cross-backbone experiments further show that the framework is not tied to a single underlying SLAM architecture. Code and visualization results are available at this https URL.
[CV-105] JAVIS: Joint Audio-Visual Interaction and Streaming for Real-Time Conversational Agents KR
链接: https://arxiv.org/abs/2610.22499
作者: Junseok Ahn,Jaehyun Park,Jeongsoo Choi,Joon Son Chung,Ji-Hoon Kim
类目: Computer Vision and Pattern Recognition (cs.CV); Audio and Speech Processing (eess.AS)
备注: 5 pages, 2 figures. Project page: this https URL
Abstract:Human communication is inherently multimodal, conveying intent not only through linguistic and vocal cues but also through visual signals. However, existing conversational systems have mainly focused on text or speech, leaving out the visual feedback that is essential for natural interaction. In this paper, we propose JAVIS, an end-to-end streaming framework for real-time audio-visual conversational agents. Our key idea is to let the voice and the face speak from a single representation. On top of a shared LLM, a Joint Refined Head produces a unified audio-visual representation, which is fed into dedicated streaming branches for speech and face generation. This end-to-end design avoids intermediate bottlenecks, enabling real-time generation while keeping speech and face tightly synchronized. In addition, we present JAVIS-Syn, a 334-hour synthetic corpus designed to address the lack of expressive audio-visual dialogue data, covering emotionally rich multi-turn conversations and assistant-style exchanges. Experiments show that JAVIS achieves approximately 200 \times lower initial latency than a strong baseline, while outperforming baselines in lip synchronization and facial expression appropriateness.
[CV-106] Has GPT -6-Astra Achieved Embodied Intelligence?
链接: https://arxiv.org/abs/2610.22471
作者: SenseNova,Joint Research Team:Zhongang Cai,Ziang Cao,Tailai Chen,Han Chong Chua,Linjun Dai,Hanming Deng,Hokin Deng,Haiwen Diao,Chenjing Ding,Weichen Fan,Xiao Fu,Zerui Gong,Chenyang Gu,Zexin He,Zhi Hou,Bo Li,Qinyue Li,Quanzhou Li,Xiaowan Li,Zhichao Liao,Dahua Lin,Zilong Lin,Bo Liu,Chunxiao Liu,Pengfei Liu,Tianqi Liu,Ziwei Liu,Chen Change Loy,Lewei Lu,Jiapeng Luo,Yuexiao Ma,Liang Pan,Xingang Pan,Oscar Qian,Ziyi Shang,Yuyao Shao,Dawei Su,Dacheng Tao,Qingyi Tao,Quan Wang,Ruisi Wang,Xiaogang Wang,Ziyi Wang,Beichen Wen,Donghao Wu,Jiaxu Wu,Silei Wu,Xiaoming Wu,Zhonghua Wu,Hongcan Xiao,Weiye Xu,Atsushi Yamashita,Lei Yang,Zhe Yang,Baiqiao Yin,Wanqi Yin,Hanfei Zhao,Tianbo Zhong,Long Zhuo,Jing Zuo
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Multimedia (cs.MM)
备注: Homepage: this https URL
Abstract:GPT-6-Astra demonstrates a level of competence in robotic manipulation and navigation that challenges the conventional paradigm of building specialized models for individual tasks and embodiments. In this report, we ask not only how capable GPT-6-Astra is, but what its capabilities imply for the field. First, we probe the limits of standalone GPT-6-Astra. We conduct a large-scale evaluation across 12 representative benchmarks spanning tabletop manipulation, humanoids, mobile platforms, and dexterous hands. Under a unified protocol that allows in-context learning (ICL) from training-set demonstrations and substantially increases the number of execution steps, GPT-6-Astra achieves unprecedented performance (Fig. 1). Future research may focus on improving 1) the generality of ICL 2) cost and execution time, 3) physical-state verification, 4) recovery from irreversible failures, and 5) long-horizon preservation of progress. Second, we study the agentic harness for embodied control. We identify that close-loop coordination between a high-level “System 2” reasoner (i.e., GPT-6-Astra) and a low-level “System 1” controller (e.g., \pi0.5) is costly but effective. Third, we examine GPT-6-Astra on real-world deployment. Our experiments on 8 robot configurations reveal substantial gaps: 1) robot demonstrations required for ICL are expensive to collect, and 2) it is implausible to trade longer execution for higher success rates with physical constraints. Fourth, we envision the next generation of embodied benchmarks. We develop a prototype that 1) derives tasks from real-world needs, 2) provides no in-domain demonstrations, and 3) judges agents on a world clock that keeps running while they deliberate. Taken together, our results suggest that GPT-6-Astra has largely crossed the threshold of embodied competence, but not yet that of practical embodied intelligence.
[CV-107] Erase the Mark Preserve the Motion: Temporally Consistent Removal of Video Watermarks
链接: https://arxiv.org/abs/2610.22457
作者: Yifan Liao,Zhen Sun,Zhicong Huang,Jiaming He,Zongmin Zhang,Xi Ai,Cheng Hong,Xinhu Zheng,Xinlei He
类目: Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV)
备注: Our videos and code can be viewed on: this https URL
Abstract:Recent advances in video generation have increased the demand for reliable watermarking for copyright protection, yet the robustness of video watermarks against removal attacks remains underexplored. Most watermark-removal methods originate from the image domain: signal optimization injects imperceptible pixel-domain perturbations, whereas generative resynthesis produces a visually similar instance through diffusion synthesis. When extended to videos, the former introduces temporally varying noise that causes visible flickering, while the latter regenerates fine-grained structures inconsistently across frames, leading to structural motion inconsistency. To evaluate video watermark robustness, we propose PoseRegen, a pose-consistent reconstruction framework that removes watermarks while preserving the original content and motion. PoseRegen first applies a temporally shared viewpoint transformation to disrupt watermark-bearing spatial patterns and then jointly reconstructs the transformed sequence with a pose-conditioned video model. A frozen video DiT processes the complete sequence with its reference-video latent, and full-sequence Flow Matching supervision keeps identity and background consistent across frames. To further preserve motion structure, we introduce a lightweight Pose-Conditioned Motion Consistency Adapter that adapts the frozen backbone to temporally ordered pose guidance and constrains body-part trajectories throughout reconstruction. Experiments on eleven video watermarking schemes show that PoseRegen reduces the average bit accuracy to 0.505, near random guessing, while outperforming the strongest baseline by 9.34% on motion-related VBench metrics. These results reveal a practical vulnerability of current video watermarks and highlight the need to evaluate robustness jointly with temporal and motion quality.
[CV-108] Event-STVG: Spatio-Temporal Video Grounding for Long-Form Event Reasoning
链接: https://arxiv.org/abs/2610.22444
作者: Yizhe Zhang,Xujia Wang,Songqi Tang,Xujie Tang,Leo Jiahao Jiang,Jiacheng Wang,Bowen Qu,JinBo Li,Qin Wang,Ji Qi,Lei Hou,Bin Xu,Juanzi Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Spatio-Temporal Video Grounding (STVG) aims to jointly localize a target in space and time given a natural-language query. Despite continued advances, existing STVG benchmarks remain largely centered on localizing targets through instance-specific descriptions, leaving them largely decoupled from long-form reasoning increasingly emphasized in multimodal large language models (MLLMs). In this paper, we introduce Event-STVG, a task that shifts STVG from locating what is described toward event-level reasoning by unifying long-form context, repeated-event reasoning, event localization, role inference, and multi-target grounding. To facilitate research on Event-STVG, we construct a large-scale benchmark comprising 4,016 annotated event occurrences across soccer and seven additional domains. Experiments with state-of-the-art MLLMs reveal substantial difficulty, with the strongest model achieving only 5.4% m-vIoU at the 10-minute horizon. Across domains, we consistently observe confusion among repeated same-type events, spatial grounding bottlenecks, and rare success in jointly grounding all required roles. Event-STVG establishes a new challenge for visual understanding in MLLMs, calling for future models that can jointly reason about complex events across space and time. Code and annotations are available at this https URL
[CV-109] An Agent ic ENF Framework for Forensic Video Authentication Against Generative AI
链接: https://arxiv.org/abs/2610.22440
作者: Ryan Wyton(1),Bacha Rehman(1) ((1) Southampton Solent University)
类目: Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM); Signal Processing (eess.SP)
备注: 12 pages, 17 figures, 4 tables. Submitted to IEEE Transactions on Information Forensics and Security. Code and data: this https URL
Abstract:Generative models now synthesise video that is hard to distinguish from genuine footage, and learned detectors that key on a generator’s artefacts degrade sharply on unseen models. This paper develops a forensic framework that authenticates video by a signal that generation does not control: the electric network frequency (ENF), the mains-driven hum in audio and flicker in video. A unified two-channel detector extracts the ENF from audio and from rolling-shutter video through one second harmonic estimator and a shared network; a synthetic-injection training method, driven by an Ornstein-Uhlenbeck drift model and anchored to real recordings, overcomes the scarcity of labelled ENF data; a reference-matching channel verifies a claimed time of recording and recovers an unknown one; and a two-axis agentic orchestrator fuses recording authenticity with content-manipulation evidence under an asymmetric rule, escalating expensive detectors only when needed. A central finding reframes the task: an ENF-band flicker is present in generated video at a prominence indistinguishable from real footage, so its presence is not proof of authenticity, and authentication must rest on consistency with an independent reference rather than on presence. On held-out real recordings the detector accepts genuine audio and video at 93% and 100% against naive baselines of 53% and 15%; claimed-time authentication reaches an area under the curve of 1.0; and the agentic cascade matches a flat ensemble while cutting wall-clock time by 54% and frames decoded by 64%, without loss of accuracy.
[CV-110] CF-NODE: Counterfactual Spatial Node Intervention for Test-Time Adaptation Under Domain Shift Evaluated Across Five Diabetic Retinopathy Datasets
链接: https://arxiv.org/abs/2610.22437
作者: Sabarish G,Deepak Parashar
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Automated fundus-based diabetic retinopathy (DR) grading can substantially expand screening capacity, but source-trained models degrade sharply across different cameras, clinics, and patient populations. Existing test-time adaptation methods depend on entropy minimization or target information accumulated from image batches and streams. These mechanisms can reinforce confidently incorrect predictions and break down when clinical images arrive independently. We introduce CF-NODE, a novel counterfactual test-time adaptation framework for cross- domain DR grading. Instead of trusting prediction confidence or estimating a target distribution, CF-NODE asks how each prediction responds under a controlled intervention on the evidence supporting it. CF-NODE ranks internal spatial nodes using model-derived attention, feature magnitude, and node-set similarity, progressively attenuates the highest-ranked nodes, and uses the stability of the resulting expected-grade response to apply a bounded calibration of the predicted grade distribution. The framework is fully frozen and adapts every image independently using only geometric views and a fixed source-derived prior, requiring no target-derived information or model updates. We evaluate CF-NODE on five external DR datasets under repeated-seed evaluation against TENT, EATA, SAR, and T3A. CF-NODE outperforms all four methods with statistically significant improvements over the second-best method, improves AUC on every target domain, and attains the highest average AUC and macro-F1, delivering stronger grade discrimination together with more balanced recognition across DR severity levels. These results demonstrate that counterfactual intervention sensitivity provides consistent and deployment-ready domain robustness for single-image clinical inference.
[CV-111] ECHOES: Seeing Beyond the Current Image via Recursive History Compression and Privileged On-Policy Distillation
链接: https://arxiv.org/abs/2610.22408
作者: Tengda Guo(1),Jiawang Hu(2),Yifei Tao(3),Yin Wang(2),Haotian Hu(4),Fanyi Wang(5),Liujuan Tang(2) ((1) Tsinghua University, (2) StepFun, (3) Nanyang Technological University, (4) Zhejiang Leapmotor Technology Co., Ltd., (5) Honor)
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: 22 pages, 7 figures, 13 tables
Abstract:Long-horizon GUI agents rely on visual history to maintain task states, but retaining historical screenshots increases visual-token usage and inference costs, while textual summaries can discard critical state information, undermining long-horizon reliability. We propose ECHOES, which leverages training-time privileged information to guide textual history compression and strengthen long-horizon reasoning. Specifically, ECHOES adopts a Compression-Think-Action (CTA) format to recursively update a textual history state and reason. The same model, given privileged multi-image inputs, serves as a teacher that provides dense token-level on-policy distillation on student-generated CTA rollouts. Our Perception-Memory Fidelity (PMF) Gap-guided OPD-RL framework dynamically balances privileged distillation and task-reward optimization, with periodic teacher updates supporting continued improvement. We further introduce ECHOESBMK, a fine-grained diagnostic benchmark derived from real GUI agent failure trajectories to jointly evaluate visual perception and cross-modal state retention. Experiments on four GUI benchmarks and ECHOESBMK demonstrate substantial gains in task performance and memory fidelity. Without retaining historical screenshots at deployment, ECHOES achieves competitive performance with a favorable performance-inference-cost trade-off and exhibits positive zero-shot transfer to unseen visual planning tasks.
[CV-112] ransNSNet: A Transformer-Enhanced Multi-Scale Adversarial Framework for CT-to-PET Image Synthesis
链接: https://arxiv.org/abs/2610.23325
作者: Shaojun Li,Xiaoyu Deng,Ikuo Keshi
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 21 pages, 7 figures. Supplementary material included. Under review at Biomedical Signal Processing and Control
Abstract:This study investigates CT-to-PET image synthesis using TransNSNet, a Transformer-enhanced multi-scale adversarial framework. A paired CT-PET dataset was curated from the public non-small cell lung cancer (NSCLC) Radiogenomics collection through patient screening, slice-level matching, preprocessing, and patient-level train-test separation. The proposed framework generates PET-like predictions at multiple spatial scales and applies coarse-to-fine reconstruction supervision, while TransPatchGAN introduces contextual feature interaction into adversarial evaluation. A progressive ablation study compares Pix2Pix, NSNet-Base, NSNet-D, and TransNSNet. Image-level performance was summarized over a fixed convergence window (epochs 80-101) from one training run per model. TransNSNet achieved the best mean value in nine of ten metrics, whereas NSNet-Base achieved the highest visual saliency-induced index. For fixed-fraction high-intensity regions, NSNet-D was marginally higher in slice-level recall and balanced accuracy, but the equal-weight patient-level values were nearly identical and the highest Dice values shifted to TransNSNet. These differences were small relative to between-patient variability. In the separate adaptive patient-level analysis (n=10), no pairwise model difference remained statistically significant after Holm correction. TransNSNet also had a substantially smaller parameter footprint than the original next-scale models. These findings indicate promising image-quality trends, but they do not establish statistical superiority or clinical validity; synthesized PET-like images should not be interpreted as substitutes for clinically acquired PET.
[CV-113] Stabilizing Illumination Conditioning in SSM-Based Low-Light Image Enhancement with Bottleneck Prompt Sidecars
链接: https://arxiv.org/abs/2610.22958
作者: Wei Chen Ng,Qiyi Xie,Jian Ding Tan,Yan Kang Lee,Abdulrab Hakim Qaid Abdullah,Wei Hown Tee
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Low-light image enhancement (LLIE) requires more than brightness amplification because real low-light images often exhibit spatially varying illumination, amplified noise, color distortion, and local detail loss. Recent SSM- and Mamba-based restorers provide efficient long-range modeling, yet the route for introducing illumination priors into such backbones remains insufficiently studied. This study addresses safe illumination conditioning for SSM-based LLIE through a bottleneck-side PromptAdapter. A lightweight illumination estimator first predicts a single-channel illumination map, which is then converted into prompt tokens and injected through a residual sidecar adapter at the bottleneck, thereby avoiding direct modification of the state-space core. Under a locked three-seed evaluation protocol on LOL-v1, LOLv2-Real, and LOLv2-Synthetic, the PromptAdapter improves over the Base model by 0.94 dB, 1.24 dB, and 0.93 dB in PSNR, respectively, while also improving SSIM, LPIPS, and masked dark-region metrics. The clearest real-data evidence appears on LOLv2-Real. A Learned-Fixed control on LOL-v1 indicates that the present results support the utility of the sidecar conditioning pathway more strongly than a strict claim of illumination-conditioned superiority, and the LOLv2-Synthetic result remains stability-conditioned. These findings suggest that, in SSM-based LLIE, conditioning design can be as important as the conditioning signal itself.
[CV-114] Measurement-Adaptive Light-Cone Transform for Robust Non-Line-of-Sight Imaging
链接: https://arxiv.org/abs/2610.22847
作者: Lianfang Wang,Xueying Liu,Junnan Dong,Qinghua Zhang,Jun Liu,Yong Wang,Yuping Duan
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Practical non-line-of-sight (NLOS) measurements are affected by noise and deviations from idealized light transport, which can degrade reconstruction with conventional light-cone transform (LCT) methods. We propose the Measurement-Adaptive Light-Cone Transform (MALCT), a physics-guided framework that retains the efficient Fourier-domain inversion structure of LCT. Within a unified variational formulation, MALCT combines the analytical response with a shared learned response through measurement-conditioned weighting and predicts a measurement-conditioned, frequency-dependent regularization map. An attention-enhanced spatio-temporal feature extraction module captures long-range temporal dependencies and local spatial correlations to support reconstruction under low-signal conditions. Experiments on synthetic data and real measurements from different NLOS systems demonstrate improved reconstruction quality. Controlled evaluations further assess robustness to temporal-response broadening and reduced acquisition time. MALCT also preserves scene structures with spatial sampling grids down to 32\times32 , supporting its potential for NLOS imaging under limited acquisition budgets.
[CV-115] Is a Linear Probe Evidence of a Linear Representation? NEURIPS2026
链接: https://arxiv.org/abs/2610.22655
作者: Nadir Kutluozen,Christian Internò,David Klindt
类目: Machine Learning (stat.ML); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Accepted at NeurIPS 2026
Abstract:The linear representation hypothesis holds that pretrained vision models encode generative factors as linearly decodable directions in feature space, and comes with a standard methodology: train a linear probe, declare a factor represented if accuracy is high. We show this test is too permissive. Asymptotically, any sufficiently rich feature map, including a random one, admits a linear readout, so high probe accuracy is consistent with two very different geometries: a true linear world model, where the latent-aligned subspace dominates the representation, or a kernel machine, where the latents lie linearly inside a much larger curved manifold. This distinction matters because the curved structure orthogonal to the probe is relevant for out-of-distribution generalization, robustness, and compositional transfer. We propose three asymmetric diagnostics that distinguish the two cases: reverse predictivity, spectral concentration, and out-of-distribution generalization. We validate them in a fully observed simulation where ground truth is known. Applied to six pretrained vision transformers on SVG-World, our deterministic generative benchmark with paired visual styles, and to two of them on CelebA, they show pretraining produces structure beyond a random kernel but well short of a linear world model. On dSprites, 3D Shapes and MPI3D, forward decodability and latent-aligned variance dissociate. Thus, forward decoding alone does not show that the probed factors align with the dominant geometry of the representation. In summary, our work cautions about the success of linear probes as opposed to truly linear and low dimensional representations.
[CV-116] LGGS: Longitudinal MRI Reconstruction via Guided Group Sparsity
链接: https://arxiv.org/abs/2610.22643
作者: Nikola Janjušević,Hersh Chandarana,Yao Wang,Li Feng
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages main, 8 figures main, 5 pages supplementary, 7 figures supplementary
Abstract:Serial MRI is routinely performed in current clinical practice for disease surveillance and applications such as MR-guided radiotherapy. However, long MRI scan times increase costs and limit patient access, making accelerated image reconstruction of broad clinical importance. Previous scans of the same patient contain image information highly correlated with the current exam, yet existing reconstruction methods rarely exploit this information. Major challenges in the effective use of prior scans include inter-visit misregistration, pathological changes, and image phase variation. In this work, we propose Longitudinal Guided Group Sparsity (LGGS), an interpretable deep-learning framework for longitudinally informed MRI reconstruction derived from classical group sparsity. The prior scan is incorporated into the reconstruction regularizer through learned phase-invariant circulant attention that identifies corresponding features across exams. This formulation provides robustness to local in-plane and through-plane misregistration as well as inter-visit phase variation. Across acceleration rates, LGGS achieves state-of-the-art performance on longitudinal T1-weighted multi-coil brain MRI data, outperforming the evaluated unguided and longitudinally informed reconstruction methods, with substantially fewer learned parameters. We further evaluated the contributions of multiple guide slices and phase-invariant similarity and demonstrated that accelerated prior images can also serve as effective guides with minimal degradation in reconstruction quality.
[CV-117] Alignment of visual brain encoding models with individual perceptual similarity judgments
链接: https://arxiv.org/abs/2610.22569
作者: Daniel Chong,Haomiao Chen,Keith W Jamison,Raja Marjieh,Ilia Sucholutsky,Nori Jacoby,Amy Kuceyeski
类目: Neurons and Cognition (q-bio.NC); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Encoding models that predict brain activation responses to images have traditionally been trained to maximize accuracy when predicting observed brain data. However, their internal representations may not reflect individual human perception. In this work, we investigate whether we can fine-tune brain encoding models to better align their representations with individuals’ similarity judgments. Using similarity judgments of 100 images from 8 participants in the Natural Scenes Dataset (NSD), we adapt a generic human similarity judgment model (DreamSim) separately for each participant to obtain personalized perceptual similarity models. These models are then used to obtain subject-specific similarity structure across images in their encoding model training sets. We fine-tune each participant’s baseline encoder in two conditions, aligning its representations with either generic or personalized similarity judgments, with the same initialization and objective. We find that baseline brain encoders reflect within-subject perceptual similarity more strongly than measured fMRI responses and preserve intersubject correlations present in measured fMRI, but fail to reflect those of human perceptual similarity judgments. Fine-tuning with generic, population-level perceptual similarity information results in better alignment with human judgments than baseline encoders, but these models still fail to capture intersubject correlations of human judgments. However, fine-tuning with personalized perceptual similarity information results in models that are the most aligned with human perceptual similarity judgments and also reflect intersubject correlations of human judgments. Our framework provides an approach for bootstrapping human similarity judgment data into more individualized, human-aligned brain encoding models of vision.
[CV-118] A Unified Theory of Image Scaling: Point Spread Functions Discrete Kernels and Scale-Spaces
链接: https://arxiv.org/abs/2610.22472
作者: Assaf Shocher
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We derive the optically aligned system of image-scaling operators \mathbfK_s that reproduces optical zoom (exact zoom-out, canonical zoom-in), by identifying a Huygens-like consistency principle in which each pixel acts as a secondary source. The system is determined in closed form by the camera’s point spread function (PSF), itself the operator \mathbfK_0 . The operators compose, \mathbfK_s’ \circ \mathbfK_s = \mathbfK_ss’ (with band projection for down-up), and account for both upscaling and downscaling, which form a Moore-Penrose pseudoinverse pair, \mathbfK_1/s = \mathbfK_s^\dagger . We show that image scaling, as routinely practiced, is neither optically faithful nor internally consistent; for example, common anti-aliasing rules generally correspond to no camera, and two resizes by a factor of 1/2 generally do not equal one resize by a factor of 1/4 . Gaussian scale-space kernels cannot be exact finite-aperture PSFs. Gaussian scale-space is an idealized special case of what emerges here: a generalized scale-space for sampled images, derived entirely from camera optics, bidirectional, and not limited to Gaussian kernels. The system goes beyond scaling and supports composable and invertible subpixel geometric transformations. We empirically validate its optical predictions with real captures and noisy optics simulations, and demonstrate its practical impact on the stability of tasks such as super-resolution and classification.
人工智能
[AI-0] Beyond SAEs: Consistent Concept Discovery with Sparse Coding
链接: https://arxiv.org/abs/2610.23801
作者: Jinho Choi,Hyesu Lim,Jaegul Choo,Steffen Schneider
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Sparse autoencoders (SAEs) explain foundation model representations by decomposing activations into sparse combinations of learned dictionary atoms. However, the learned dictionaries can vary across training runs, limiting the reliability and reproducibility of SAE-based analyses. We propose Consistent Sparse Dictionary Learning (ConSDL), which replaces the standard linear encoder with per-sample sparse coding. This is motivated by our synthetic analysis showing that coefficients from a single linear encoder pass are suboptimal compared with those obtained by per-sample optimization, suggesting that sparse coding is a reliable mechanism for dictionary learning. Across 16 synthetic settings, ConSDL achieves near-perfect dictionary and coefficient recovery. These gains translate to state-of-the-art cross-seed consistency across four vision and language settings, where ConSDL improves dictionary and coefficient consistency by 17.6% and 35.9%, respectively, over the average of six SAE baselines.
[AI-1] FuTransformer: FP8 prefill and INT4 decode for Llama-style inference on Hopper
链接: https://arxiv.org/abs/2610.23796
作者: Wei Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Transformer prefill reuses weights across tokens and is dominated by matrix multiplication and activation data movement; single-token decode has limited reuse and is dominated by weight and KV-cache reads. We present FUTRANSFORMER, a CUDA implementation of a Llama-style Transformer block for NVIDIA Hopper GPUs. It applies CODA’s GEMM-plus-epilogue approach to inference, adding direct KV-cache placement, folding the fixed RMSNorm gain into weights, writing FP8 output directly with delayed scaling, and supporting prefill without padding. Decode uses INT4 projection weights and an INT4 KV-cache copy. Weights and keys follow Tensor Core fragment layouts for direct dequantization into MMA operands without full intermediate tiles. A block-diagonal Hadamard transform reduces cache quantization error from outlier channels. On one H100, prefill is up to 1.48x faster than compiled PyTorch FP8 and reaches 1.29 PFLOP/s. For gate/up projection with SwiGLU at the 8B shape, our BF16 path is 1.13x faster than CODA. Our FP8 path is 2.02-2.18x faster than the evaluated BF16 CODA path. Across 8B and 70B shapes at batches 1-32, INT4 decode is 1.3-2.4x faster than the faster of PyTorch’s BF16 and INT4 blocks at each point: 2.1-3.3x faster than compiled, CUDA-graphed BF16 and 1.3-6.3x faster than the same block using PyTorch’s INT4 projection kernel, which targets a single query row and loses to BF16 at batch 32. It is also 31-51% faster than our FP8 path. Complete Llama-3.1-8B decode takes 2.8 ms per step at batch one, versus 8.1 ms for the compiled transformers implementation. Decoding every WikiText-2 token through the INT4 weight path raises perplexity from 6.24 to 7.23 with round-to-nearest weights or 6.99 with GPTQ in the same format; the rotated INT4 cache adds 0.20. The kernels match a quantized reference computation within 0.001 perplexity.
[AI-2] DrugHarness: An Evolvable Agent Harness for Computational Drug Discovery
链接: https://arxiv.org/abs/2610.23791
作者: Yuxing Chen,Zhisheng Zhang,Cheng Tang,Wei Li,Minghui Zhang,Ziyan Huang,Shaocong Wei,Jiawei Chen,Junru Hu,Hongze Zhu,Jianghan Shen,Min Cen,Xinyuan Zhu,Yaxue Zhao,Jubo Wang,Yuqiang Li,Chun-Mei Feng,Zongyuan Ge,Rui Su,Jie Xu,Yirong Chen,Xi Xu,Ming Hu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Computational drug discovery requires agents to compose scientific operations and adapt to changing evidence and constraints. Here we introduce DrugHarness, an evolvable agent harness coupling composable scientific execution with conditional skill induction. Its environment integrates over 80 scientific operations, biomedical resources and task-specific code, preserving associations among candidates, results and artifacts. From execution trajectories and outcome feedback, DrugHarness induces action hypotheses with observable applicability conditions, recommended actions and supporting evidence. An independent review agent evaluates candidate skills derived from these hypotheses through source verification and, when needed, targeted execution trials. Admitted skills guide subsequent tasks and are revised through feedback. Evaluation on MolBench and SMDD-Bench demonstrates capabilities in molecular screening, optimization and constrained design. Without evolution, DrugHarness exceeds reported MolClaw-CC results by 10.8 percentage points in affinity-comparison accuracy and 55% in mean top-three active hits on MolBench. Retrospective evaluation on skill-development tasks shows that evolution increases Qwen’s affinity-comparison accuracy from 62.2% to 81.1% and optimization success from 79.5% to 87.2%. In a prospective HPK1 study, DrugHarness achieves a more favorable median docking score than Biomni among system-specific candidates. A synthesized compound achieves a cellular IC50 of 14.19 nM, a 2.64-fold potency improvement over the best input reference, with activity competitive with two selected compounds designed by experienced medicinal chemists. These results support conditional skill evolution for scientific agents and demonstrate a workflow connecting computational drug design with experimental validation.
[AI-3] Beyond Large Language Models : A Mental World Model Framework for Mental Modeling Simulation Evolution and Interpretability
链接: https://arxiv.org/abs/2610.23722
作者: Qingshi Wang
类目: Artificial Intelligence (cs.AI)
备注: 12 pages, 10 figures
Abstract:Understanding and simulating how subjective cognition and mental states emerge, evolve, and shape behavior is essential for advancing artificial intelligence toward deeper social cognition and general intelligence. Existing large language models (LLMs) and Theory of Mind (ToM) methods primarily infer mental states from linguistic information, lacking unified mechanisms for subjective world construction, cognitive-psychological interactions, dynamic evolution, and historical dependencies. This paper introduces the Mental World Model (MWM) and its implementation, the Multi-Branch Mental Evolution Architecture (MMEA), which explicitly constructs subjective worlds, models interactions between cognition and mental states, and simulates continuous mental evolution shaped by historical experiences. On DynToM and RecToM, MMEA achieves 96.48% and 94.00% accuracy, respectively, outperforming leading LLMs and specialized ToM methods. Additional experiments demonstrate improvements of 28.7, 39.3, and 16.6 percentage points in mental state and behavior prediction, long-sequence prediction, and implicit mental state inference, respectively. The results demonstrate that MMEA provides an interpretable framework for continuous mental modeling, enabling AI to move beyond surface-level semantic representations to infer individuals’ underlying mental states from linguistic information and dynamically simulate their internal cognitive processes and evolution over time. MWM also provides a promising foundation for causal reasoning, personalized AI, sustained human-AI interaction, and future advances toward artificial general intelligence (AGI).
[AI-4] Neural Expansion: A Unified Mechanism for How Deep Neural Networks Generalize NEURIPS NEURIPS2026
链接: https://arxiv.org/abs/2610.23716
作者: Chashi Mahiul Islam,Samuel Jacob Chacko,Mao Nishino,Canlin Zhang,Xiuwen Liu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted at NeurIPS 2026. Includes supplementary material. Code: this https URL
Abstract:Deep neural networks demonstrate remarkable generalization despite being highly over-parameterized, challenging traditional statistical learning theories. While several theoretical paradigms such as margin-based bounds, PAC-Bayesian analyses, and neural tangent kernels provide global explanations, they often lack mechanistic insight into how generalization emerges from training dynamics. We introduce Neural Expansion, a unified geometric framework that characterizes generalization through the expansion of directionally-selective ray coverage in the input space. Using Generalization Intervals (GIs), we quantify direction-specific regions around training samples where the model’s predictions remain stable. Our extensive empirical studies across diverse architectures and datasets reveal that training leads to progressive expansion of ray coverage, particularly along directions with low Jacobian singular values. This mechanism helps explain a large fraction of correctly classified test inputs, adversarial examples, and even mislabeled or out-of-distribution samples through the geometry of ray coverage rather than instance-level memorization. By connecting input-space curvature to predictive behavior, our framework provides a mechanistic foundation that complements existing global theories of deep learning.
[AI-5] OwlSight: A Panoramic Monitor for Chain-of-Thought Reasoning
链接: https://arxiv.org/abs/2610.23670
作者: Lewen Yan,Yanxu Zhu,Jie Zhang,Jing Shao
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Chain-of-Thought (CoT) monitoring offers actionable evidence for oversight AI systems, with the potential to enable early detection and intervention. However, existing approaches are focused on specific tasks, whereas the regulation of AI, particularly reasoning models, requires monitoring a broader range of tasks. We introduce Factor for CoT monitoring, a strategy that supports customizing monitoring requirements in the prompt, which enables a single monitor to cover different tasks by relying on different factors. Furthermore, we introduce two types of factors, fine-grained and coarse-grained (e.g., monitoring whether a specific email address has been leaked versus monitoring privacy leakage), to accommodate diverse user monitoring needs. Based on this, we develop a CoT monitor named OwlSight that relies solely on factor to monitor behaviors related to multiple dimensions, including capability, safety, and trustworthiness. Experiments across 15,971 reasoning trajectories show that OwlSight surpasses general-purpose LLMs by 1.32-30.57 and 11.71-27.81 points at fine and coarse granularity, respectively, and achieves state-of-the-art results on OpenAI’s CoT monitorability benchmark. It also generalizes to unseen models’ CoTs and remains robust to Factor rephrasings, CoT compression, and obfuscation, offering a promising paradigm for CoT monitoring that can contribute to the oversight of frontier AI.
[AI-6] UPLE-Bench: Evaluating and Mitigating Privacy-Utility Trade-offs in Typed Decision Models
链接: https://arxiv.org/abs/2610.23668
作者: Yiqu Yang,Qiaoyuan Zheng,Yang Tian,Yu Fan,Elliott Ash
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: Our public website: this https URL
Abstract:Typed decision models often return more than a selected category. They also expose probabilities for each candidate and, sometimes, a separate confidence score. These fields support downstream decisions but may reveal information beyond a recipient’s authorization. We introduce TUPLE-Bench (Typed decision models’ Utility and Privacy Leakage Evaluation Benchmark), a benchmark of paired synthetic cases in which recipients may learn a coarse operational category but not its fine-grained cause. Across nine models, instructions to withhold the cause fail to protect it: 91.88%-99.93% of responses reveal the cause through direct selection or residual scores. Even when causes are excluded from the candidate set, full outputs enable 62.85%-75.35% recovery on held-out cases, compared with 51.04%-57.64% using only the selected category (50% chance). Full outputs also reduce mean downstream cost in all 27 model-task comparisons. However, confidence fields increase leakage without consistently reducing cost. We further introduce workflow-constrained privacy quantization (WCPQ), which reduces secret recovery while preserving the behavior and cost of specified downstream workflows. Our findings highlight the need to audit decision interfaces field by field and workflow by workflow.
[AI-7] GEMS: Molecular Structure Elucidation as Learned Search over Isomer Graphs
链接: https://arxiv.org/abs/2610.23617
作者: Utku Umur Acikalin,Yingheng Wang,Goncalo J. Gouveia,Tyler J. Schwertfeger,Frank C. Schroeder,Carla P. Gomes
类目: Artificial Intelligence (cs.AI)
备注: 36 pages, 9 figures, 17 tables
Abstract:Identifying small-molecule structures from molecular formulas and tandem mass spectra remains a bottleneck in untargeted metabolomics and natural-product discovery. Even when database retrieval or de novo prediction misses the exact structure, the resulting candidates can provide starting points for exploring nearby structures through a few bond edits. We introduce GEMS (Guided Editing of Molecular Structures), which refines candidates through learned 2-switch bond edits that preserve the molecular formula and each atom’s total valence. GEMS uses MIST, a pretrained spectrum-to-fingerprint model, to predict fingerprint probabilities once per spectrum. A neural policy compares each candidate’s fingerprint with these predictions, links present features to the atoms and bonds that generate them, and scores all valid edits in one forward pass. The policy is trained on shortest edit paths between structures, so it can be pretrained on PubChem isomers without spectra and then fine-tuned with spectrum-predicted fingerprints. At inference, GEMS searches from multiple initial candidates matching the supplied molecular formula and ranks candidates by fingerprint agreement and multi-start consensus. With the target structure excluded from PubChem initialization, three edit rounds achieve Hit@1/Hit@10 of 29.64%/50.81% on NPLIB1 and 30.22%/41.33% on MassSpecGym, exceeding published baselines under comparable settings. On NPLIB1, GEMS achieves higher identification accuracy in about 3.3 minutes end to end, compared with 26.2 hours for our FRIGID run. GEMS also improves identification from generated candidates, including those refined by FRIGID. Experiments with exact target fingerprints identify fingerprint prediction error as a bottleneck in target recovery and final ranking. Deeper search recovers additional targets without retraining, with larger gains from weaker initial candidates.
[AI-8] Beyond Action Prediction: Learning Search-Free Planning in Latent Space
链接: https://arxiv.org/abs/2610.23615
作者: Guanchen Du,Jinhao Du,Yanpeng Sun,Shan Zhang,Na Zhao,Zechao Li
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Latent world models learn action-conditioned state transitions in a compact representation space, providing a predictive model for visual planning and control. However, planning with these models typically requires costly test-time optimization over action sequences. We propose to replace this online search with a goal-conditioned inverse model that directly predicts actions from current and goal observations. A key challenge is that action prediction alone neither ensures that goal representations encode action-relevant information nor that predicted actions are consistent with the learned latent prediction model. We introduce Hindsight Inverse JEPA Planning (HIJP), a search-free framework that addresses both aspects during training. HIJP relabels future observations along offline trajectories as goals, providing action supervision across multiple temporal horizons. By retaining gradients through the goal representations, this supervision directly shapes the visual representation. HIJP further feeds predicted actions into the latent predictor model and aligns the resulting next-state representations with encoded successor observations, enforcing action-transition consistency. At inference time, the inverse dynamics model and latent predictor models are applied alternately to generate multi-step action sequences without online search. Across four visual control tasks, HIJP achieves 86.8% overall success and a 1.9-16.5x planning speedup over search-based planning.
[AI-9] From Trajectories to Differentiated Skills: Cognitive Differentiation for Self-Improving Agents
链接: https://arxiv.org/abs/2610.23599
作者: Yewei Zhao,Derong Xu,Wei Chen,Yin Xu,Peilun Zhou,Xiaohan Wang,Guojun Yin,Wei Lin,Mingjun Xiao
类目: Artificial Intelligence (cs.AI)
备注: 22 pages, 7 figures, 17 tables, including appendices
Abstract:Large language model (LLM) agents tackle complex tasks through reasoning and interaction with external environments. Agent skills provide persistent, transferable capabilities that guide task planning, inform contextual decisions, support tool use, and enable executable procedures. Yet existing skill-evolution methods largely organize knowledge around skill artifacts, overlooking the functional heterogeneity within individual skills and its implications for knowledge maintenance and use. Inspired by cognitive science, we propose Cognitive Differentiation in Skill Evolution (CDSE), which evolves a unified skill through four functionally distinct yet coordinated components: Meta, Experience, Tool, and Script. CDSE extracts evidence-grounded semantic gradients from successful and failed trajectories, consolidates them by functional role, and applies component-specific updates and validation with cross-component coordination and adaptive invocation. An analysis of 500 top-ranked skill packages shows that 96.8% contain knowledge spanning at least two functional roles. Across six benchmarks, three agent models, and two execution harnesses, CDSE achieves the highest scores in all 36 settings, improving over the Base agent by 7.6 percentage points and the strongest competing method in each setting by 2.6 points on average. Ablation studies further support the importance of functional differentiation and adaptive invocation.
[AI-10] RL-Guided Temporal Localization for Dual-Channel Retrieval in Long-Horizon Agent Memory NEURIPS2026
链接: https://arxiv.org/abs/2610.23598
作者: Dabin Sheng,Zhe Wu,Jinming Zhao,Junliang Xing,Yuanchun Shi
类目: Artificial Intelligence (cs.AI)
备注: Accepted at NeurIPS 2026
Abstract:Time-aware retrieval is crucial for long-horizon LLM agents to effectively leverage ever-growing long-term memory. Existing methods focus on lexical or semantic similarity, leading to target memories being crowded out by temporally mismatched yet semantically similar candidates. Even time-aware approaches inject temporal cues only during memory organization or answer generation, overlooking temporal constraints at the retrieval stage, where candidate selection occurs - the root cause of the displacement. Inspired by the temporal contiguity effect in human episodic memory, we propose TIDER (Temporal Interval-driven Dual-channel Evidence Retrieval), a time-aware agentic retrieval framework. Central to TIDER is a reinforcement learning (RL)-trained temporal interval locator that infers the temporal windows in which the evidence lies. RL optimizes window localization as a discrete, non-differentiable retrieval decision, using composite rewards to convert sparse answer supervision into fine-grained retrieval feedback. These temporal windows first define the temporal retrieval channel; the retrieved candidates are then merged with results from the semantic channel and reranked by proximity to the windows. This temporal grounding enables TIDER to retrieve temporally aligned evidence that would otherwise be drowned out by the volume of semantically similar yet temporally mismatched memories. With a lightweight 1.7B locator, TIDER achieves state-of-the-art accuracy on both long-horizon memory benchmarks - 61.52% on LifeBench (+8.82%) and 78.96% on LoCoMo (+3.66%) - which indicates that temporal cues are crucial for pinpointing target evidence amid a vast pool of semantically similar historical events in long-term daily-life memory.
[AI-11] Darwin-Claw: Environment-Adaptive Experience Consolidation for Recursively Self-Evolving Agents
链接: https://arxiv.org/abs/2610.23574
作者: Mingze Yin,Xiaohan Wang,Dian Li,Longfei Luo,Jiahong Yan,Yiheng Zhu,Chang-Yu Hsieh
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:While Large Language Model (LLM) agents have demonstrated remarkable capabilities across complex tasks, they predominantly operate in isolated sessions, lacking the ability to continually learn from historical interactions or adapt to user-specific environments. Existing self-evolution paradigms often indiscriminately inject raw interaction trajectories into memory or skill modules, leading to obscured behavioral patterns and compromised update controllability. Furthermore, as foundational LLM reasoning improves, the bottleneck for agentic evolution shifts from accumulating generic knowledge to effectively managing situational experience. To address this, we propose Darwin-Claw, an environment-adaptive self-evolving framework that systematically distills episodic experiences to drive the comprehensive evolution of operational skills and durable memory. Theoretically grounded in an iterative agent-actor and evolver-critic co-adaptation, Darwin-Claw introduces a Hierarchical Experience Consolidation System that extracts fine-grained error attributions and generalizes them into dynamically updated prototype vectors. This structured situational experience serves as interaction patches for context-sensitive perception and explicit optimization signals for skill refinement. Practically, Darwin-Claw is a readily deployable, automated framework compatible with mainstream agent systems without requiring manual curation. Extensive experiments on WildClawBench demonstrate that Darwin-Claw yields substantial performance improvements, establishing a controllable and stable pathway for the continual, user-centric adaptation of autonomous agents in real-world ecosystems.
[AI-12] SkillBrowser: Selective Skill Exposure for Scalable Skill-Using Agents ACL
链接: https://arxiv.org/abs/2610.23554
作者: Jin Ye,Yifan Jiang,Lan-zhe Guo
类目: Artificial Intelligence (cs.AI)
备注: 23 pages, 3 figures. Submitted to ACL Rolling Review, October 2026
Abstract:Large language model agents increasingly rely on skills to extend their capabilities in complex tasks. However, scalable skill use requires Selective Skill Exposure (SSE): providing relevant procedural content when needed while avoiding unnecessary exposure that may disrupt reasoning. Existing paradigms fall short in complementary ways. Progressive disclosure exposes metadata for a growing skill library, introducing context pollution; retrieval limits catalog exposure, but static retrieval from the initial task query may be misaligned with capability needs that emerge during reasoning and may still inject unnecessary skills. We propose SkillBrowser, a Meta-Skill-managed framework that enables an agent to identify current capability needs, formulate targeted queries, and selectively load skills on demand. Across six SRA-Bench subsets and five LLMs, SkillBrowser improves aggregate accuracy over static retrieval and LLM-based selection, including multi-step tool use, and mitigates accuracy degradation on already-solvable instances. Benefits remain model- and task-dependent, with under-invocation of useful skills and additional latency as key limitations. These results demonstrate the viability of Meta-Skill-managed SSE for scalable skill repositories.
[AI-13] Reason -guided Human Reward Modeling for Preference-based Reinforcement Learning
链接: https://arxiv.org/abs/2610.23550
作者: Anshul Nayak,Shahil Shaik,Jonathan Smereka,Yue Wang
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Reinforcement learning (RL) policies optimized for task rewards may produce behaviors that are efficient but misaligned with human expectations, safety preferences, or intent. This paper presents a reason-guided preference-based reinforcement learning (PbRL) framework that learns a human reward model from trajectory-level feedback. In addition to preference labels, the framework incorporates reason tags explaining why a behavior is preferred, enabling the learned reward to capture both human choices and their underlying rationale. Compared with preference-only reward learning, the reason-guided formulation provides additional supervision for learning human-aligned rewards and improves the interpretability of the resulting policy. The learned human reward is combined with the environment reward during policy optimization, allowing the agent to improve task performance while producing more interpretable, human- aligned, and preference-sensitive behavior
[AI-14] From Billion-Scale to Million-Scale Language Models: Operational Capabilities for Embedded Post-Exploitation
链接: https://arxiv.org/abs/2610.23534
作者: Gabriel Hortea,Simon Pietro Romano,Juan Tapiador,Mario Varlese
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:Small Language Models (SLMs) are increasingly attractive for resource-constrained environments due to their low computational footprint. These characteristics also make them candidates for future autonomous malware, where connectivity, latency, privacy, or operational constraints may preclude cloud-based inference. Yet the ability of ultra-compact models to support malware during post-exploitation remains poorly understood. We systematically evaluate seven open-source SLMs (20M-1.5B parameters) on two post-exploitation code-generation tasks: Directory Enumeration and Backdoor Generation. We compare Python (high-resource) and Lua (low-resource) across prompting strategies, task formulations, iterative reasoning, grammar-constrained decoding, and execution-feedback loops. Results show that offensive code-generation capabilities do not increase linearly with model size. Successful execution depends on the interaction of model capacity, task formulation, and prompting strategy, with task formulation exerting a greater influence than prompt engineering alone. Algorithmic formulation combined with few-shot prompting enables Qwen2.5-Coder 1.5B to reach Pass@5 of 100% in both Python and Lua for Directory Enumeration, 100% in Python for Backdoor Generation, and 99.2% in Lua. Although the evaluated models remain too computationally demanding for widespread deployment in contemporary malware, rapid progress in compact models may reduce this barrier. These findings help anticipate AI-assisted malware and inform future countermeasures. Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI) Cite as: arXiv:2610.23534 [cs.CR] (or arXiv:2610.23534v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2610.23534 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Mario Varlese [view email] [v1] Fri, 16 Oct 2026 14:33:17 UTC (1,670 KB)
[AI-15] rapped in Your Story: Evaluating Narrative-Induced Sycophancy in Large Language Models
链接: https://arxiv.org/abs/2610.23525
作者: Runtai Yang,Dandan Zhu,Kaiwei Zhang,Guangtao Zhai
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) are known to exhibit sycophancy by adjusting their responses toward users’ expressed views. However, real users often justify their opinions through reasons, personal experiences, causal explanations, and broader narratives, raising the question of whether such narratives can further steer model judgments. We construct a controlled benchmark of 500 debatable questions across ten domains, where the user stance is held fixed while its justification is systematically varied without introducing new task-relevant evidence. Each question is instantiated under 11 matched conditions: one neutral baseline and five mirrored narrative levels for each of two opposing positions, yielding 5,500 evaluation instances. We evaluate seven LLMs and observe substantial differences in narrative-induced agreement. The effect is not monotonic with narrative elaboration: personal experience can reduce agreement, whereas causal framing and worldview-based synthesis increase it. Controlled GPT-5.1 ablations further show that causal framing produces the strongest positive effect, while claim-preserving bullet presentation also increases agreement over paragraph form. These findings suggest that sycophancy is shaped not only by user stance, but also by how supporting explanations are structured and presented.
[AI-16] LocalIntent: Outcome-Specific Structured Generation for Backend-Compatible Smart Building Automation with Local LLM s
链接: https://arxiv.org/abs/2610.23511
作者: Tianyang Liu,Jiayue Wang,Zhanhao Cheng,Changyang Wang,Shaoshu Zhu,Zhenghao Wu,Hadi Tabatabaee Malazi
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Locally deployed large language models (LLMs) can reduce reliance on remote cloud services for natural language smart building automation, but backend-facing processing requires correct outcome selection, alignment with local capabilities, and valid structured outputs. We investigate whether a complete outcome-specific prompting design improves compact-model outputs beyond generic route-then-generate prompting. The proposed LocalIntent approach routes requests to ordinary or persistent-rule branches: the former produces commands, clarifications, or rejections, while the latter generates trigger-action rules, followed by deterministic validation. We evaluate four locally executed Gemma and Qwen models spanning 2B-9B parameters on 101 non-persistent and 54 persistent-rule records, comparing Monolithic prompting, generic two-stage prompting, LocalIntent, and validation-guided repair. Across the evaluated datasets, LocalIntent achieves higher intent understanding, device identification, and action/parameter correctness than generic two-stage prompting, while clarification/mode correctness remains model-dependent. The largest gains occur for rules, where LocalIntent improves intent understanding from 26.4% to 80.6%, device identification from 59.3% to 84.7%, rule-mode correctness from 49.1% to 98.6%, and action/parameter correctness from 5.6% to 44.9%. For non-persistent requests, pooled gains are smaller: intent understanding increases from 39.6% to 53.2%, and action/parameter correctness from 30.7% to 42.8%.
[AI-17] A Topological Signature of Learning under Noisy Labels
链接: https://arxiv.org/abs/2610.23491
作者: Zhongtian Sun,Fan Mo,Prayag Tiwari,Kelin Xia
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:How does noisy supervision change the representations learned by a neural network? Training accuracy measures agreement with training labels but provides little information about representation geometry. We use persistent homology to track loop structure across distance thresholds in neural activations. Experiments span image classification, text classification, and modular arithmetic. On a fixed CIFAR-10/ResNet-18 training probe, equally sized clean-label and corrupted-label subsets show no detectable persistence difference at epoch 10, whereas corrupted-label subsets have higher persistence at epoch 50. The contrast survives uniform-scale controls, even though predictions match the assigned noisy labels for only 7-11% of corrupted probe examples at that checkpoint. Normalisation preserves the crossover between clean and noisy training trajectories on CIFAR-10/ResNet-18, whereas CIFAR-100/ResNet-18 shows no crossover under any of the five evaluated constructions. In a separate noise-rate calibration comparison across five pipelines, mean activation norm achieves a higher mean in-sample R2 than raw persistence (0.936 versus 0.908). These findings show that noisy supervision leaves a measurable geometric signature even when agreement with corrupted labels is low. Comparing persistence under scale controls with label-fitting measurements offers a way to study how representations respond to label corruption beyond what training accuracy reveals.
[AI-18] S-GAP: Stealthy Graph Attack Primitives for Graph Large Language Models
链接: https://arxiv.org/abs/2610.23465
作者: Wenda Li,Tongya Zheng,Shunyu Liu,Hanyang Yuan,Kaixuan Chen,Yu Wang,Yuchen Ying,Haoze Jiang,Huan Li,Gang Chen,Zujie Ren
类目: Artificial Intelligence (cs.AI)
备注: 8 pages, 4 figures, 3 tables
Abstract:Graph Large Language Models (GraphLLMs) enable powerful semantic understanding and reasoning over Text-Attributed Graphs (TAGs), while their semantic reasoning may exhibit vulnerabilities to perturbations in node text and graph structure under adversarial attack evaluation. Existing attacks typically construct textual and structural perturbations in separate stages guided by distinct realization intents. However, this separation results in attack-intent fragmentation, causing semantic incoherence between the two perturbation types, thereby undermining attack stealth and making them easier for GraphLLMs to identify and disregard, ultimately weakening attack effectiveness. To avoid attack-intent fragmentation, we propose the Graph Attack Primitive (GAP), an atomic semantic representation that unifies textual and structural perturbations under a coherent attack intent. By carefully calibrating the semantic composition of a GAP, we can achieve strong attack effectiveness while maintaining sufficient attack stealth. Furthermore, we introduce S-GAP (Stealthy GAP), a multi-agent framework that iteratively refines the semantic content of the GAP. The framework unifies GAP planning and realization and continuously refines this process through surrogate evaluation and textual feedback. Extensive experiments across four datasets, three GraphLLMs, and two attack settings show that S-GAP improves ASR over the strongest baselines by 7.02% on average while producing more relationally plausible perturbations, exposing critical vulnerabilities in current GraphLLMs and motivating more robust defenses.
[AI-19] GatewayBench: Unified Capability and Security Evaluation Platform for MCP Gateways
链接: https://arxiv.org/abs/2610.23449
作者: Yitian Zheng,Zhaorun Chen,Haoyu Li,Bo Li
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:AI agents have increasingly interacted with external environments by invoking tools, with Model Context Protocol (MCP) emerging as a standard interface for connecting agents to heterogeneous tools and services. As organizations deploy agents at scale, MCP gateways have become a critical middleware layer for aggregating, governing, and shaping the tool surface exposed to agents. Despite the growing importance of MCP gateways in agentic systems, no prior work has systematically studied how gateway design affects agent capability, efficiency, and security. To address this gap, we introduce GatewayBench, the first unified evaluation framework and benchmark for MCP gateways. We characterize gateway architectures into two major paradigms: flattened gateways and hierarchical gateways. GatewayBench comprises 524 tasks across six domains, together with a pool of 11,118 real-world tools from 251 authentic MCP servers. To emulate large-scale deployments, we systematically inject task-irrelevant tools sampled from this pool into each agent environment, enabling controlled evaluation as the accessible tool surface scales. Based on GatewayBench, we conduct extensive experiments across seven widely deployed MCP gateways and diverse agent configurations, covering both benign and adversarial tasks. Our results reveal substantial differences across gateway designs in task completion, tool-use efficiency, and resilience to attacks, and further demonstrate that these effects interact strongly with the underlying model and agent harness. These findings highlight the MCP gateway as a consequential yet underexplored component of the agentic AI stack. By enabling controlled evaluation of gateway architectures under realistic tool ecosystems, GatewayBench provides both practical guidance for MCP gateway design and a rigorous testbed for developing more capable, efficient, and secure agentic systems.
[AI-20] RECREG: Recovery-based Shielding for Regular Safety Properties in Reinforcement Learning
链接: https://arxiv.org/abs/2610.23445
作者: Alexander W. Goodall,Francesco Belardinelli
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:In many real-world scenarios we need to consider formal requirements, such as safety and constraint compliance, to deploy Reinforcement Learning (RL) agents. In this paper we introduce RecReg, a safe RL framework capable of enforcing non-Markovian regular safety properties via recovery-based shielding. RecReg is a unified approach that supports different levels of a priori knowledge and safety guarantees. The paper primarily focuses on building the theoretical foundations for RecReg in the tabular setting, providing rigorous safety and convergence guarantees that are not present in related work. We provide an extensive set of empirical results using the MASA-Safe-RL library, and we outline an extension of RecReg to deep RL domains using SOTA model-based RL. Our empirical results demonstrate that RecReg achieves efficient learning and matches the theory in this paper.
[AI-21] Generative Policy Improvement from Observational Data
链接: https://arxiv.org/abs/2610.23436
作者: Jonas Schweisthal,Dennis Frauen,Stefan Feuerriegel,Emma Brunskill
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Preprint
Abstract:We study how to improve decision policies from observational data to generate better actions when actions are complex, structured objects, such as text messages or images. This task is subject to two major challenges: (i) action-level overlap is often too weak for reliable off-policy evaluation and improvement, and (ii) generating new structured actions often requires expensive training or fine-tuning (e.g., of an LLM) and can be ill-defined due to information loss in action representations. To address both, we propose PolicyLift, a three-stage framework for generative policy improvement with structured actions from observational data. In Stage 1, PolicyLift learns a small set of discrete action codes that capture high-level strategies in the observed action space and are tailored to support policy improvement under limited overlap. In Stage 2, it learns a context-dependent policy over these action codes. At deployment time, Stage 3 “lifts” the learned code-level policy back to the original action space by generating actions from the historical behavior system in a code-conditioned way. We show theoretically that improving the policy over action codes guarantees improvement of the induced policy in the original action space. We demonstrate PolicyLift across experiments in synthetic and real-world text generation tasks with consistent policy improvements over the behavior policy.
[AI-22] AWAM: Active Trajectory-Conditioned View Imagination for Robust Robot Manipulation
链接: https://arxiv.org/abs/2610.23396
作者: Jiawei Geng,Zhenyang Liu,Yuzheng Wu,Yanwei Fu
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Rapid progress in embodied intelligence has enabled increasingly capable robots to follow language instructions and manipulate diverse objects. However, strong results in structured environments do not automatically transfer to open-world conditions, where clutter, occlusion, and changing viewpoints can obscure the spatial relationships required for action. To address this perceptual gap, a virtual active-perception paradigm is introduced. AWAM, an Active World-Action Model actively plans where to observe and imagines what will be seen along the planned trajectory before acting. Task-related virtual camera motion is predicted and used as a structured query into a geometry-conditioned world model, producing a reusable spatial prior for manipulation, which is later compressed into a fixed memory and fused with live observations through cross-attention for diffusion-based action prediction. AWAM achieves competitive success rates with 75.9% on LIBERO-Occ, 62.4% on LIBERO-Plus, 76.5% on RLBench-OG Occlusion1, and 94.4% in real-robot trials. These results support improved robustness to the evaluated occlusions and distribution shifts, addressing part of the gap between structured manipulation and less controlled environments.
[AI-23] Monitor-Validity Failures: LLM Safety Evidence Paths at Evolving Framework Boundaries
链接: https://arxiv.org/abs/2610.23362
作者: Junchi Liao
类目: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Software Engineering (cs.SE)
备注:
Abstract:Modern LLM applications rely on runtime monitors that inspect tool calls, traces, streams, detector outputs, and verifier inputs. This creates an interface-reliability problem: a framework path may remain operational while policy-relevant evidence never reaches its decision point. In one representative case, a regex detector returned a positive match, but the framework stranded the result before the block decision. This differs from jailbreaks and detector errors because the evidence is already recognized yet disconnected from enforcement. We study this repair-boundary signature with canary contracts: parent/fixed confirmation units that place a known payload only on the changed evidence surface and record monitor visibility together with the downstream decision for action-boundary cases. Applying the protocol to public framework repairs, we confirm 18 evidence-path failures across 11 LLM or agent safety project groups and four implementation ecosystems. Nine are established-boundary repairs, eight are new-surface policy-coverage gaps, and one is a boundary-only evidence case. Fourteen reach E1 enforcement boundaries: seven native action outcomes, six configured enforcement boundaries, and one synthetic-consumer consequence. The remaining cases affect verifier, safety-prompt, or observability channels. The cases form carrier, adapter, authority, and surface families and show how ordinary application behavior can remain intact while a safety-specific evidence path misses its decision. Exact revisions, deterministic outcomes, negative controls, and a frozen prospective audit document the confirmation rule. Monitor assumptions therefore need regression tests at the schemas, streams, tools, and adapters that frameworks evolve.
[AI-24] OliO: ODE-based Linear Transition Operator for Self-Supervised Time Series Forecasting NEURIPS2026
链接: https://arxiv.org/abs/2610.23345
作者: Kwangryeol Park,Seulki Lee
类目: Artificial Intelligence (cs.AI)
备注: Accepted at NeurIPS 2026
Abstract:Self-supervised representation learning has advanced time series forecasting by capturing robust features from unlabeled data. However, existing contrastive and masking-based methods often struggle to explicitly model the underlying temporal flow, either by treating time lag as noise or by disrupting the inherent temporal dynamics essential for forecasting. To address these limitations, we propose OliO (ODE-based Linear Transition Operator), a plug-and-play self-supervised learning method that learns bidirectional temporal flows as a continuous and structurally consistent dynamical system. OliO introduces a transition operator derived from linear Ordinary Differential Equations (ODEs) to align latent representations across arbitrary time shifts. By imposing a strictly upper triangular constraint on the transition matrix, we mathematically ensure numerical stability and enforce a robust polynomial boundary that effectively captures long-term dependencies while preventing the exponential divergence typically found in unconstrained ODE-based models. This structural constraint induces a temporally coherent representation space that preserves the underlying flow essential for accurate forecasting. Extensive experiments across various backbone architectures and benchmarks demonstrate that OliO achieves significant performance gains, providing Mean Squared Error (MSE) reductions of up to 5.75% over the most competitive grid-searched SOTA baselines. The code is available at *this https URL
[AI-25] Agent Onto: Ontology-Guided LLM Agents for Enterprise Data Analytics
链接: https://arxiv.org/abs/2610.23317
作者: Dayan Pan,Bincheng Zuo,Xiaojian Wang,Daogang Zhang,Jingyuan Wang
类目: Artificial Intelligence (cs.AI); Databases (cs.DB)
备注: Under review
Abstract:Enterprise data analysis requires agents to connect analytical requests with knowledge distributed across databases, business documents, and existing example tasks. Without an explicit representation of this knowledge, an agent may produce incorrect analytical results and cannot reliably incorporate feedback. We present AgentOnto, an ontology-guided LLM agent for enterprise data analytics. AgentOnto organizes heterogeneous enterprise knowledge into an ontology that directly guides analysis and incorporates confirmed feedback. Iterative Ontology Preparation further improves the ontology through self-generated analysis tasks. We evaluate AgentOnto on two public data sources. AgentOnto substantially outperforms representative methods, while module-level evaluations confirm the effectiveness of ontology construction, analysis, and updating. A real-world deployment further demonstrates its practical value in enterprise analytics. These results demonstrate that explicitly constructing, using, and updating enterprise knowledge supports reliable and adaptable data analysis.
[AI-26] From Memory Hits to Answerable Evidence: Query-Conditioned Memory Compilation for Long-Term Agent Memory
链接: https://arxiv.org/abs/2610.23313
作者: Shangze Li,Yanyu Yao,Derong Xu,Zhi Zheng,Tong Xu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Long-term agent memory requires recovering answer-relevant evidence dispersed across turns and sessions. Existing systems increasingly use structured memories and adaptive retrieval, yet often pass retrieved or expanded units directly to the reader, coupling memory addressing with evidence presentation and leaving the final context redundant or incomplete. To address this gap, we propose MemCompile, a query-conditioned evidence compilation framework for long-term conversational memory. Given a query, the Address Plane locates relevant support through compact source-linked descriptors and capacity-aware Cells, while a normalization boundary resolves heterogeneous retrieval hits to stable SourcePacket identities. Its Evidence Plane then allocates a bounded evidence budget across complementary packets, admits limited additional support, and renders the retained claims associated with each selected packet into an answer-ready context. MemCompile achieves 80.52% Judge Accuracy on LoCoMo and extends the empirical quality-cost Pareto frontier to a higher-quality regime. Its strong performance on LongMemEval-S further suggests that the benefits of query-conditioned compilation generalize across long-term conversational-memory benchmarks.
[AI-27] MobileInject: Evaluating Reasoning -to-Action Security of MLLM -Based Mobile Agents under Cross-Modal Consistency Injection
链接: https://arxiv.org/abs/2610.23307
作者: Songping Wang,Peiyang Li,Yi Huang,Fengting Li,Xuhao Ran,Yueming Lyu,Chao Feng,Caifeng Shan
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:Multimodal large language model (MLLM)-based mobile agents are rapidly evolving from conversational assistants into autonomous systems that perceive, reason over, and act on smartphone environments. This capability also exposes a new security risk: adversaries can embed malicious objectives in cross-application mobile content and steer agent behavior through indirect multimodal injection. Although prompt injection has been studied in desktop and web settings, its cross-modal properties in mobile ecosystems remain poorly understood, and existing security analyses are still limited. To close this gap, we introduce MobileInject, the first benchmark for evaluating mobile agents against multimodal injection attacks and reasoning-to-action security risks. MobileInject provides a high-fidelity interactive mobile evaluation framework covering 11 scenarios, 9 security-risk categories, and 108 executable risk instances. It further introduces Cross-Modal Consistency Injection to exploit corroboration across modalities, two enhanced attack methods, Multi-Role Collaborative Deception and Reasoning-Camouflage Injection, to intensify trust manipulation and reasoning hijacking, and a Reasoning Security Diagnosis Framework for fine-grained failure analysis. Systematic experiments show that mobile agents are broadly vulnerable to multimodal injection, and that many failures stem from reasoning hijacking at early stages. These findings establish MobileInject as a systematic cross-modal benchmark for studying and improving the security of mobile agents. Code release details are provided in the Appendix Open Science section.
[AI-28] Optimist Deep Research: An Iterative State-Centric Agent for Autonomous Literature Research
链接: https://arxiv.org/abs/2610.23299
作者: Gihan Panapitiya
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Most retrieval-augmented systems treat literature search as a single retrieval step followed by summarization. Real literature research works differently: the investigator’s understanding of the question changes as they read, new subquestions appear, and evidence is weighed, not just collected. We present Optimist Deep Research, an open-source agent that runs literature research as an iterative evidence-acquisition process over an explicit, mutable research state. A research question is broken into a tree of subquestions, each served by search intents that are translated into source-specific queries over arXiv, OpenAlex, Crossref, PubMed, Semantic Scholar and the web. Results are merged into a canonical, provenance-preserving document registry through a conservative four-tier deduplication hierarchy. The registry is expanded by citation snowballing until a saturation criterion is met, and then screened, ranked and allocated to subquestions round-robin. Documents are read in parallel into a structured evidence store of claims with support, location, type and strength. A critic reviews the state along eight dimensions and returns research actions rather than a binary verdict: it can add, split, rephrase, deprioritize or close subquestions, so the research plan is an output of the investigation as well as an input. The final report is written section by section from the evidence store, and every citation resolves to a retrieved document. We describe the design, the deterministic machinery that keeps a language-model pipeline auditable, and a case study that read 28 of 237 retrieved documents in about three and a half minutes. We also report weaknesses the case study exposed.
[AI-29] CoTA: Understanding the Repeat Curse in dMLLM s from an Information Flow Perspective
链接: https://arxiv.org/abs/2610.23290
作者: Qiyan Zhao,Xiaofeng Zhang,Xu-Yao Zhang,Shijie Han,Zhepeng Li,Shunzhi Zhu,Da-Han Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Diffusion-based multimodal large language models (dMLLMs) decode by iterative denoising with bidirectional attention, which precludes the standard autoregressive key–value cache. Practical deployments therefore rely on approximate caching that reuses intermediate states across denoising steps. Such caching, however, often makes a response repeat one token at adjacent positions in a run that breaks the sentence, a failure we term the \emphRepeat Curse. Tracing the context-token information flow reveals three disruptions. Within a step, caching disperses the attention that anchors a position on nearby context tokens. Across steps, similarity-based refresh is not staleness-aware, so some positions are decoded against stale key–value states. Across layers, the deep-layer entropy of the context fails to converge where repetition arises. We accordingly propose CoTA++, a training-free, plug-and-play framework that counteracts each disruption where it originates: Context-Token Attention Re-anchoring re-forms the attention routing against the current context, Decode-Aware Refresh reserves part of the cache’s refresh budget for the positions about to be committed, and Context-Token Entropy-Guided Voting discounts the confidence of a position by the deep-layer entropy of its context, which reorders the unmasking. The three components act at non-overlapping stages of a decoding step and compose freely with existing dMLLMs and caches. Across three dMLLMs and two caching backends, CoTA++ reduces the sample repetition rate of long responses from up to 99.6% to at most 17.0%, recovers most of the accuracy lost to caching on nine benchmarks, and keeps the speedup of caching at under 15% overhead, making approximate caching reliable for dMLLMs.
[AI-30] Beyond Marginal Influence: Interaction-Aware Memory Selection for LLM Agents
链接: https://arxiv.org/abs/2610.23266
作者: BHVSP Subrahmanyam,Neeraj Choudhary
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:LLM-agent memory selection often uses retrieval rankings based on record relevance or individual influence. Such rankings cannot identify sufficient subsets under redundant or jointly necessary evidence. We study per-decision memory sufficiency: selecting a small subset that preserves full-memory behavior. To separate search error from model error, we introduce JointCoreBench, a benchmark with nine interaction families, six workflow domains, natural-language decisions, and exact sufficient-set oracles, and JointCore, a relevance cost-ordered group-deletion baseline that tests blocks before individual records. On 6,912 exact conditions, cost-ordered JointCore recovers a minimum-cost sufficient set in 100.0% of cases, reduces subset evaluations from 30.00 to 11.13 (62.9%), and achieves 87.8% active-memory savings. Across four 12B-70B model families and 432 tasks per model, its unweighted mean is 10.28 versus 16.00 model queries and 50.6% versus 43.1% minimum-cost recovery relative to cost aware singleton deletion. Both methods preserve full-memory behavior under the deterministic primary interface. Tests with physical deletion, UNKNOWN masks, PENDING sentinels, counterfactual values, and alternative labels identify interface dependent behavior. JointCore is an efficient benchmark baseline; the study does not establish deployment-ready memory compression.
[AI-31] Beyond Reward Scoring: Online Multi-Agent Preference Learning for Multimodal Response Generation
链接: https://arxiv.org/abs/2610.23256
作者: Ziyun Zou,Jiwoo Hong,Yifan Lu,Shao Tang,Weiyi Wu,Aman Gupta,Jin Tian,Imran Razzak
类目: Artificial Intelligence (cs.AI)
备注: 32 pages, 4 figures
Abstract:Multimodal preference alignment often follows a score-based RLHF pipeline: a reward model is first trained to learn human preferences, then used to score on-policy responses and optimize the response model’s generation. This turns preference supervision into a scalar signal for generation: a model may learn to produce answers with higher preference scores, but is rarely trained explicitly to learn the reasons underlying human preferences. We therefore ask whether making preference understanding an internal capability of the response model can improve its generation. To investigate this question, we propose \textbfPrefMind, an online dual-agent reinforcement learning framework with two trainable agents: a response Generator and a preference Critic. Given a pair of candidate responses, the agents analyze and discuss which response better aligns with human preference. Human labels supervise their judgments and corrections. For pairs on which both the Generator and revised Critic correctly identify the human-preferred response, the Generator produces an improved answer, and the Critic compares it against the original preferred response to provide a generation reward. Both agents are jointly updated from interactions sampled with their current policies. Through this process, \textbfPrefMind trains the Generator to understand human preferences via explicit judgment and correction, and to translate this understanding into better responses. Experiments across three model scales show improvements in both standalone preference judgment and downstream response generation.
[AI-32] Preventing Data Hallucinations in Agent ic AI for Optical Network Troubleshooting
链接: https://arxiv.org/abs/2610.23252
作者: Jeremias Dötterl,Martin Kuipers
类目: Networking and Internet Architecture (cs.NI); Artificial Intelligence (cs.AI)
备注: Presented as a demo at the 52nd European Conference on Optical Communication (ECOC 2026)
Abstract:An agentic AI system for automated troubleshooting in optical communication networks is demonstrated that integrates data references into the agent architecture to prevent data hallucinations, improving the interpretability and trustworthiness of root-cause identification and reasoning.
[AI-33] Characterising and Analysing Retrieval-Augmented Generation Techniques
链接: https://arxiv.org/abs/2610.23226
作者: Emmanuel Igbekele,Olawande Daramola
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:The advent of Large Language Models (LLMs) across many application domains has increased the need for Retrieval-Augmented Generation (RAG) to reduce hallucinations and improve LLM performance. Thus far, several RAG techniques with impressive features and capabilities have been proposed in the literature, making it a herculean task to select the most appropriate RAG techniques for specific problem scenarios. Thus, we propose a conceptual framework comprising main dimensions and sub-dimensions for characterising (defining “what it is”) and analysing (defining “how well it does”) existing RAG techniques. The conceptual framework was developed through a qualitative analysis of the attributes of RAG techniques reported in the literature. It can serve as a pre-empirical filtering mechanism for researchers and practitioners in selecting the highest-fit RAG candidates that meet specific user and task requirements before empirical benchmarking of RAG techniques, which can be computationally and financially expensive.
[AI-34] qGAMMA: A Framework for Optimal DNN Mapping on Multi-Precision Accelerators
链接: https://arxiv.org/abs/2610.23222
作者: Leandro Fiorin,Luigi Altamura,Cristina Silvano
类目: Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI)
备注: Accepted at the 25th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation (SAMOS XXV), 2025. To appear in Springer LNCS
Abstract:This work describes qGAMMA, a flexible framework for exploring and optimizing the mapping of mixed-precision Deep Neural Network (DNN) models onto general-purpose spatial accelerators that support a range of fixed- and floating-point precisions. qGAMMA navigates the design space using a domain-specific genetic algorithm and an analytical performance model based on hardware synthesis results. The framework identifies the optimal mapping and accelerator configuration that minimizes target metrics. Extensive evaluations across DNN workloads show that qGAMMA offers valuable insights for designing mixed-precision applications, enabling better trade-offs between performance, energy efficiency, and resource utilization.
[AI-35] Not All Test-Time Compute Is Evidence: Model-Specific Bias in LLM Reasoning
链接: https://arxiv.org/abs/2610.23213
作者: Ming Shen,Tianjun Yao
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Test-time scaling methods implicitly assume that additional computation from the same model provides independent evidence. We show that this assumption fails within a single reasoning model: errors are correlated across repeated samples, across reasoning depth, and in the model’s own self-evaluation, so more compute from the same model is not more evidence. We introduce DIA (Disagreement-triggered Independent Arbitration), which treats disagreement between an intermediate-stage and the final-stage prediction as a per-question signal of reasoning instability and resolves it by asking a small panel of heterogeneous models to solve the question independently, without seeing either candidate, and aggregating their answers directly. Across seven benchmarks and two base models, DIA outperforms full chain-of-thought in 12 of 14 settings, including a 7.00% gain on MuSiQue, while activating on only \sim 17% of examples and using up to 47.2% fewer tokens than a multi-agent baseline. Ablations show that both independent solving and solver heterogeneity are necessary for these gains.
[AI-36] Validating LLM -generated political networks with survey-reported partisan exposure
链接: https://arxiv.org/abs/2610.23199
作者: Marco Tonin,Michele Tizzoni,Bruno Lepri,Luca Maria Aiello
类目: ocial and Information Networks (cs.SI); Artificial Intelligence (cs.AI); Physics and Society (physics.soc-ph)
备注:
Abstract:Large language models (LLMs) are increasingly used to simulate human behavior in social science research, including decisions, interactions, and social relationships. However, in network formation tasks, existing evaluations often focus on aggregate network properties, while networks with plausible structural properties may still fail to reproduce individuals’ social and political environments. Here, we generate LLM-based political networks using personas sampled from the nationally representative ANES 2020–2022 Social Media Study and use respondents’ survey-reported partisan exposure as an empirical benchmark for the political environments generated for the corresponding personas. We first characterize the structural properties and political assortativity of the generated networks, and then compare ego-level partisan exposure with the survey. We find that the structure and political assortativity of the generated networks vary across experimental configurations. More importantly, when comparing generated partisan exposure with the survey, some models generate stronger separation between Democratic and Republican egos than reported in the survey, with higher same-party and lower cross-party exposure, while others mainly underestimate Republican exposure across ego groups. Moreover, networks with similar political assortativity can still generate different partisan environments around the same personas, including among networks with plausible structural properties. Finally, generated networks show group-specific errors, under- or overestimating exposure to either Democrats or Republicans for specific social groups. These findings show that evaluating LLM-based social simulations only through aggregate outcomes can hide errors in the local environments through which those outcomes emerge, highlighting the importance of validation at both aggregate and individual levels.
[AI-37] Framing Spatial Hearing within the Free Energy Principle: A Hierarchical Model of Perceptual Learning
链接: https://arxiv.org/abs/2610.23196
作者: Roberto Barumerli,Nils Meyer-Kahlen,Lorenzo Picinali
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:
Abstract:Spatial hearing relies on dynamic internal representations to decode spatial information from ever-changing acoustic conditions. A fundamental example is sound localisation, where listeners infer the direction of an invisible sound. Bayesian inference provides a computational account of such a localisation process; however, modelling efforts are largely limited to static listening scenarios. Instead, the free energy principle (FEP) describes perception, learning, and action as joint minimisation of variational free energy (VFE), providing a principled framework to account for dynamic conditions. Here, we take a first step toward applying the FEP to spatial hearing. We characterise how listeners’ beliefs evolve across trials, in which they estimated sound direction and received subsequent visual feedback on each trial. Listeners are described by a hierarchical generative model where visual feedback drives gradient descent on VFE, updating beliefs over three independent quantities: lateral and polar localisation precisions and hemifield disambiguation. The model captures individual learning trajectories and reveals components with different dynamics: lateral precision improves, as does hemifield disambiguation, consistent with increased sensitivity of the auditory pathway in decoding vertical direction. Notably, polar precision also increases, most for listeners with the lowest initial precision, suggesting that the observed vertical improvement is driven by both hemifield disambiguation and polar precision. This work establishes spatial hearing as a tractable domain for FEP and motivates future closed-loop experiments with full action-perception cycles.
[AI-38] FuzzInsight: Navigating the Efficiency-Effectiveness Trade-off in Coverag e-Guided DNN Fuzzing
链接: https://arxiv.org/abs/2610.23195
作者: Vivek Vekariya,Fathiyeh Faghih,Alexander Pretschner
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Collecting and labeling diverse test datasets for evaluating Deep Neural Networks (DNNs) is costly and time-consuming. This has motivated coverage-guided fuzzing techniques that generate defect-revealing corner cases from limited test data. However, their effectiveness depends heavily on configuration choices, particularly coverage criteria and seed-selection strategies. We present a configuration-oriented analysis and decision guidelines for scalable DNN fuzzing targeting two bottlenecks: inefficient seed utilization and the cost of coverage guidance. Our analysis shows that existing seed-selection strategies repeatedly prioritize a small subset of inputs, leaving many collected seeds unexplored. To address this issue, we propose Equalized Sampling, a constraint-based strategy that increases mean seed queue utilization to 77%, improving on prior approaches by 13 percentage points. We further evaluate 84 fuzzing configurations on four benchmark datasets: MNIST, SVHN, GTSRB, and CIFAR-100. Results show that the non-structural criteria LSCD, LSC, and MDSC achieve coverage outcomes comparable to structural and alternative non-structural criteria while requiring substantially less computation. These criteria also promote broad behavioral exploration of the DNN under test, reducing the need to monitor multiple coverage metrics. Among them, LSCD provides a modest but consistent runtime advantage. Finally, we show that retraining with generated corner cases improves generalization robustness by up to 77 percentage points on average without degrading performance on the original test distribution. Together, these findings provide a practical blueprint for selecting fuzzing configurations that balance efficiency, behavioral exploration, and model hardening. Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2610.23195 [cs.AI] (or arXiv:2610.23195v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2610.23195 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-39] OfficeTown: Generative Organizations for Evolving Agent Intelligence
链接: https://arxiv.org/abs/2610.23194
作者: Hongchen Wei,Daohong Ying,Zhi Hong,Aidan Wu,Bei Liu,Ruichun Ma,Yifan Yang,Qi Dai,Kai Qiu,Rui Xu,Dongdong Chen,Zhenzhong Chen,Chong Luo
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:AI agents increasingly use files and software tools to produce professional deliverables. Such work draws on materials, decisions, and feedback accumulated across assignments. We introduce OfficeTown, a generative simulation framework for professional organizations in which tasks and their context arise from continuing organizational work. From a natural-language description of an organization’s purpose and activities, OfficeTown initializes an organization whose members coordinate work, delegate bounded assignments to assistant agents, and review their deliverables. Artifacts, decisions, and unresolved matters persist and shape subsequent work, while responsibilities and permissions determine what each member can see and do. We curate work requests, their context, and review feedback into executable resources for evaluation and learning. We construct OfficeTown Bench with 406 tasks and 4,836 evaluation criteria from 23 simulated organizations across eight professional domains, incorporating domain-expert feedback. Frontier-model evaluations reveal substantial remaining headroom. Supervised fine-tuning and on-policy distillation on OfficeTown data substantially improve Qwen3.5-9B on external professional-work benchmarks, including GDPval. Supplementary evaluations show that the principal–assistant harness achieves higher scores than the single- and multi-agent baselines tested. These results highlight the value of organizational simulation in generating work-grounded data and environments for the continued development of generalist agents.
[AI-40] LangNCO: Compiling Natural-Language Constraints into Inference-Time Interventions for Neural Combinatorial Optimization
链接: https://arxiv.org/abs/2610.23190
作者: Hironori Ohigashi
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Neural combinatorial optimization (NCO) rapidly produces high-quality solutions for problem structures encountered during training, but operational requirements may change after deployment and be communicated in natural language. We propose LangNCO, a framework that compiles instance-specific requirements into structured inference-time interventions for a frozen NCO solver, without changing the pretrained policy’s architecture, learned input representation, or parameters. A pretrained LLM serves as a replaceable semantic front end, translating each requirement into a schema-validated domain-specific language (DSL) rule set. A deterministic Rule Executor applies the rules as nonexpansive hard action masks. Because local masking prevents immediate violations but does not ensure that a complete feasible continuation remains reachable, Critical-Offset Decoding extracts rule-referenced critical nodes and reallocates part of a fixed rollout budget to multi-strength biased rollouts. Using controlled task-specific language to isolate solver adaptation from open-ended language robustness, we evaluate a CVRP policy on road closures, time windows, vehicle assignments, visit-order requirements, all pairwise combinations, and twelve additional constraint types. LangNCO achieves 100% feasibility on the four primary datasets with mean reference gaps of 1.63–3.26%, 97.33–100% feasibility when separately generated rules are composed for pairwise constraints, and 98.67% feasibility across the additional constraints. These results demonstrate modular post-training adaptation of a frozen NCO solver through inspectable language-derived rules, without retraining.
[AI-41] A Stronger Teacher Within: Strengthening MoE Models via On-Policy Bidirectional Distillation
链接: https://arxiv.org/abs/2610.23185
作者: Haoyu Chen,Jun Bai,Yang Liu,Jianfei Zhang,Wenge Rong
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Reinforcement Learning with Verifiable Rewards (RLVR) has substantially improved Mixture-of-Experts (MoE) LLMs, but its outcome-level rewards provide only sparse supervision. On-Policy Distillation offers dense token-level feedback, yet requires a stronger compatible teacher, while on-policy self-distillation relies on hand-designed privileged information. We observe that the MoE architecture itself offers a way to resolve this dilemma, as increasing the activation budget per token exposes latent expert capacity and yields a potentially stronger teacher within the model. However, naive expansion degrades performance because newly co-activated experts are not optimized to work together. We therefore propose MoEvolve, a bidirectional distillation framework following a recovery-then-improvement strategy: we first restore the activation-expanded high- k model and further strengthen it by exploiting its additional active capacity, then reverse the direction of distillation to transfer the enhanced capabilities back into the original low- k configuration, retaining its inference-time activation budget. Across diverse MoE models and benchmarks, MoEvolve improves the initial models by up to 22.2 domain-average points and outperforms same-budget RLVR baselines by up to 6.5 points. Notably, the final low- k models often surpass their activation-expanded teachers.
[AI-42] LongForge: Measuring and Improving Long-Context Instruction Data via Evidence-Assertion Reasoning Graphs
链接: https://arxiv.org/abs/2610.23167
作者: Ziyang Chen,Junlong Jia,Xing Wu,Tinghao Yu,Songlin Hu
类目: Artificial Intelligence (cs.AI)
备注: Work in progress
Abstract:Existing quality control for long-context instruction-tuning data is a one-shot judgment inside the generation pipeline that leaves only a scalar, revealing neither which document parts an answer uses nor how many reasoning steps it writes down; the data can thus be neither audited nor controlled. We formalize data quality as a property of the alignment structure among document, question, and answer, and build LongForge, which pairs an analyzer that measures long-context data with a synthesizer that generates it. The analyzer reverse-reconstructs each sample into an evidence-assertion reasoning graph and programmatically computes four dimensions on it: breadth, depth, correctness, and completeness. Every reading traces back to specific evidence spans and reasoning steps, making data auditable. Measuring seven mainstream datasets reveals two previously unquantified defects: breadth barely grows with document length, and depth is generally low, the two being nearly orthogonal. The synthesizer then embeds the analyzer into generation, gating on correctness and completeness and deferring breadth and depth to a training-time mixture over measured values, which makes the training distribution controllable. On two backbones, its two mixtures, LongForge-1 and LongForge-N, outperform the strongest baseline by 2.2-4.7 points on average over six long-context benchmarks, and in a controlled comparison that holds the gate and the candidate pool fixed, raising breadth and depth alone is worth a further 1.16 points, making the structural distribution of training data a lever on downstream quality rather than only a description of it.
[AI-43] Algorithm Modelization: Transforming Optimization Algorithms into Evolvable Intelligence
链接: https://arxiv.org/abs/2610.23161
作者: Jian Yao,Zhiyao Zhang,Yichen Li,Xingyu Wu,Kay Chen Tan
类目: Artificial Intelligence (cs.AI); Neural and Evolutionary Computing (cs.NE)
备注:
Abstract:Can algorithms be learned and evolved as models, rather than repeatedly redesigned as programs? We explore this question by introducing Algorithm Modelization, a new paradigm that turns iterative algorithms into trainable models. Its central idea is to regard algorithm execution trajectories as computational experience: by learning how algorithmic states evolve over iterations, a model can internalize the knowledge embodied in existing algorithms, reproduce their computational behaviors, and further evolve beyond them through training. As a first realization of this paradigm, we modelize iterative optimization as an autoregressive process, where preceding generations constitute the computational context for generating the next population. We develop a dimension-independent population-temporal architecture to capture both population interactions and iterative dynamics, together with an imitation-to-surpassing learning framework that first acquires algorithmic knowledge from heterogeneous teachers and then improves the resulting optimizer directly through optimization feedback. The modelized optimizer consequently goes beyond reproducing individual algorithms: it consolidates heterogeneous search knowledge into a single trainable model, achieves stronger optimization capability through subsequent improvement, and generalizes its acquired knowledge across problems and substantially larger unseen dimensions. More fundamentally, Algorithm Modelization makes algorithmic capability accumulable through learning, where new algorithms and problem-solving experience can become training resources for progressively improving the model. Our results establish a promising path from manually designed algorithms toward general algorithm models that learn, generalize, and evolve from computational experience.
[AI-44] CIIR - Class-Indexed Interface Representation Learning
链接: https://arxiv.org/abs/2610.23151
作者: Doo Yon Kim,Christian Wallraven
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 17 pages, 6 figures
Abstract:Multimodal learning commonly assumes paired or synchronized observations, yet independently collected datasets rarely provide such correspondence. We introduce CIIR, a modular encoder–decoder framework that connects heterogeneous datasets through a class-indexed address space while preserving each dataset’s native input format and output module. A known local index selects address k , and an input-dependent 128-dimensional address representation Z_k(x) is written to that address. To demonstrate CIIR, we learn this address space from handwritten digits, clothing images, spoken-digit spectrograms, and hand signs without synchronized or instance-matched samples. Newly trained label-free receivers recover the requested addresses with 94.7% macro-average Address Realization Accuracy. We then attach the unseen SVHN dataset after freezing the base address space and train only lightweight adapters, achieving 90.2% MNISTSVHN and 82.3% SVHNMNIST end-to-end Address Transfer Accuracy without target-side decoder labels. To demonstrate generalizability of CIIR, we reuse the ten addresses as code elements in collision-free arithmetic and hierarchical address codes for all 100 CIFAR-100 classes and evaluate direct arithmetic-code realization through the existing receivers. These results establish CIIR as a compact interface for modular cross-dataset communication and post-hoc dataset attachment without exhaustive paired multimodal supervision.
[AI-45] Local proxies for memory-efficient training
链接: https://arxiv.org/abs/2610.23145
作者: Wu S. Zihan,Brian A. Yuan,Guillaume Bellec
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Local learning algorithms can reshape resource requirements of artificial intelligence. The memory bottleneck of standard end-to-end backpropagation lies in the need to store all network activations until the loss gradient has been propagated back through. Local learning avoids this, but typically at a cost in accuracy. We study how to achieve accurate and memory efficient local learning in transformers. For each transformer block, we train a local proxy network to predict the global network output via distillation. Gradients can be independently propagated through this local proxy while estimating well the backpropagation gradients. In theory, this approach is exact for deep linear networks; in non-linear residual models, the gradient error can be bounded when using the appropriate residual proxy. We verify this numerically and demonstrate that optimizing the distillation loss improves gradient estimation. Evaluating the performance of proxy propagation on CIFAR-100 in vision transformers, we find that it achieves a better accuracy/memory trade-off compared to other local algorithms. Finally, we show that the method is further relevant for fine-tuning large language models on a small memory budget. Our method trains faster and achieves higher performance than memory-efficient zeroth-order learning algorithms with comparable memory cost.
[AI-46] LexPilot: Benchmarking the Proactive Interaction Ability of Legal LLM s with Laypeople in Chinese Civil Case Consultation
链接: https://arxiv.org/abs/2610.23141
作者: Wenbin Zhang,ShangZhan Li,Bingxiang He,Ye He,Yuxin Zhou,Muzhen Cai,Qi Shi,Wanxiang Che
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large Language Models (LLMs) have greatly assisted legal professionals. However, our empirical studies reveal that, when faced with underspecified consultations from non-experts, current LLMs often produce irresponsible responses with limited utility despite lacking sufficient case details for definitive conclusions. To better assist laypeople, legal LLMs must identify focal points of dispute and proactively elicit crucial context from users. Unfortunately, existing data resources often focus on interaction and response quality, offering limited insight into how proactive interaction affects LLMs’ downstream claim-level judgment prediction in actual civil cases. To address this gap, we introduce \textbfLexPilot, a benchmark linking proactive legal interaction to claim-level verdict prediction using authentic Chinese civil judgments. Constructed from 432 real-world judgments across 36 common causes of action, LexPilot comprises expert-reviewed pairs of complete and underspecified consultation histories with corresponding court verdicts. Experiments reveal that current LLMs struggle with proactive legal interactions in underspecified consultations, with downstream judgment prediction substantially underperforming that in the fully specified consultation setup (-23.84% on average). Further analysis identifies critical bottlenecks hindering these capabilities. Our work highlights the urgent need for more user-centered legal AI to better serve the public.
[AI-47] WebCoder: Scaling General Frontend Development with Artifact-Grounded Agent ic Reinforcement Learning
链接: https://arxiv.org/abs/2610.23104
作者: Zijun Dou,Zhen Yang,Xijun Liu,Dongzhuoran Zhou,Aohan Zeng,Jie Tang
类目: Artificial Intelligence (cs.AI)
备注: 50 pages, 22 figures, 11 tables. Project page: this https URL
Abstract:Real-world frontend development requires agents to compose reproduction, generation, modification, and repair over an evolving application state. Learning a general policy for this process requires translating heterogeneous execution feedback into task-aligned rewards and effective credit assignment across long sequences of interdependent actions. We introduce WebCoder, a general frontend development agent trained with artifact-grounded agentic reinforcement learning. Our key idea is to assess diverse development activities through their effects on a shared executable artifact and its satisfaction of task requirements. We build Webdev-World, an executable environment that integrates code editing, application execution, browser inspection, and on-demand skills while preserving project state across interactions. An independent verifier evaluates the final application against task requirements to provide terminal rewards, while changes in intermediate artifact progress redistribute the resulting rollout advantage across development segments. Context compaction and message-tree sample construction support training on trajectories beyond a single context window while avoiding repeated loss computation on shared generated history. We further introduce Webdev-Bench to evaluate composed development requirements within a single project. On Webdev-Bench, WebCoder-27B raises requirement satisfaction from 64.8 for its Qwen3.8-27B base model to 72.8, exceeding Claude-Opus-4.8 (65.1) and GPT-5.4 (59.6); on Vision2Web, it improves the average visual and functional score from 41.1 to 62.4, on par with Claude-Opus-4.8. Ablation studies confirm the contributions of artifact-grounded credit assignment and long-horizon training.
[AI-48] Rethinking GRPO for Large Reasoning Model Safety Alignment
链接: https://arxiv.org/abs/2610.23085
作者: Jiawei Chen,Yi Zhang,Jiawei Du,Zhonghou Lv,Tairan Chen,Junyuan Shang,Jinchang Hou,Xiao Yang,Yu Tian,Chenfu Bao,Chao Yu,Zhaoxia Yin
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large reasoning models (LRMs) exhibit strong reasoning capabilities but remain vulnerable to unsafe outputs. While Group Relative Policy Optimization (GRPO) is widely used for safety alignment, its group-relative optimization may provide insufficient safety-directed learning signals. In this paper, we identify two failure modes: advantage collapse, where homogeneous rollout groups yield vanishing advantages, and unsafe diversification, where unsafe behaviors persist and become increasingly diverse. Our theoretical analysis reveals that behavioral consistency weakens within-group safety contrasts, while entropy regularization can diversify unsafe responses when relative signals vanish. To address these challenges, we propose Beacon, comprising Safety Signal Recovery (SSR) and Safety-Utility Decoupled Alignment (SUDA). SSR combines advantage-guided trajectory filtering and safe reference supervision to recover effective safety learning signals, while SUDA employs category-aware rewards and benign-only KL anchoring to mitigate over-refusal without compromising safety. Experiments across four Qwen3 and DeepSeek-R1-Distill variants and eight benchmarks show that our framework achieves the best average safety and reasoning performance, reducing average safety error by up to 32.83 percentage points while maintaining low over-refusal.
[AI-49] Path to Prior-Data Fitted Relational Models via Learned Latent Multi-table Representations
链接: https://arxiv.org/abs/2610.23073
作者: Mohammad Sadeq Abolhasani,Viswanath Ganapathy
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Presented at the Fifth Learning on Graphs Conference (LoG 2026), Boston, MA, USA, November 20-22, 2026
Abstract:Prior-data fitted networks (PFNs) show that pretraining on large populations of synthetic tasks can enable generalization to unseen datasets with little or no downstream fine-tuning. This paradigm is effective for transformer-based tabular and relational foundation models such as RDB-PFN, but remains largely unexplored for graph-based relational models built on message passing. We pretrain the graph-based relational foundation model Griffin on large collections of synthetic relational databases and ask whether curriculum learning, which benefits transformer-based relational learners, also improves message-passing representations. Rather than fine-tune on downstream tasks, we use the pretrained encoder as a general-purpose relational representation learner and pass its latent representations to a frozen in-context predictor (\ticl); no component is trained on the target task. On 16 real-world relational classification tasks, evaluated over five seeds against raw-feature and untrained-encoder baselines, a purely synthetic curriculum that begins with schema-guided data and continues with schema-agnostic data gives the strongest frozen in-context performance\rev, outperforming real-data-only pretraining and every unordered single-stage mixture of the corpora; when appended after real-data pretraining, it yields the best encoders overall. These results provide initial evidence that PFN-style synthetic pretraining transfers to graph-based relational foundation models.
[AI-50] Unifying Embodied Foundation Models via a Model-to-Policy Formulation
链接: https://arxiv.org/abs/2610.23053
作者: Xinyi Hong,Mingyang Ren
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Vision-language-action (VLA) models and world-action models (WAMs) have emerged as promising approaches to robot learning and control. These models integrate multimodal perception and action generation, but differ in their architectures, output representations, and learning objectives. In this paper, we present a unified, architecture-agnostic probabilistic formulation for embodied foundation models based on stochastic kernels. Our formulation has two key components: (i) we characterize embodied foundation models through conditional output distributions and action-realization kernels, which together induce task-conditioned robot policies; and (ii) we connect pretraining and post-training to policy deployment, accommodating model-specific pretraining objectives and offline or online post-training strategies. We instantiate the formulation for OpenVLA and DreamZero, covering autoregressive action-token prediction and joint video-action generation.
[AI-51] Agent ic In-Context Learning: What Algorithms Can Transformers Learn in Closed-Loop Interaction?
链接: https://arxiv.org/abs/2610.23051
作者: Shiguang Wu,Yaqing Wang,Quanming Yao
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:The learning algorithm view of in-context learning (ICL) studies how fixed-parameter Transformers learn from supplied examples. Language model agents extend this setting through interaction: their actions determine what evidence enters context next. We call this regime Agentic In-Context Learning (Agentic ICL). Existing studies largely depend on externally supplied contexts or reward-bearing interaction histories, while work on general tool-using agents lacks a learning algorithm view. We represent an interactive learner by its state, update, and decision rule, covering active learning, bandits, and online reinforcement learning (RL). For expressivity, we construct causal Transformers that approximate these algorithms over a finite horizon, and bound the resulting loss in return. For training, we characterize when optimizing process rewards preserves behavior that maximizes the final outcome. Distillation experiments show that Transformers can learn to execute three representative recursions. We then train models from outcome or process rewards without oracle action or state targets, as in RL post-training. Different rewards yield distinct decision policies. Improving the alignment of process supervision with outcome-optimal action preferences shifts learned decisions accordingly. Such behavioral differences persist in a pretrained language model across textual paraphrases.
[AI-52] Distill Until Divergence: Error-Truncated Hindsight Distillation for Agent ic RL
链接: https://arxiv.org/abs/2610.23048
作者: Yibin Huang,Conghui Zhu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Long-horizon interaction requires agents to make multiple decisions in a changing environment. Outcome-based GRPO lacks fine-grained guidance for intermediate decisions. Recent methods distill completed trajectories into reusable skills and use hindsight-conditioned OPD to provide dense token-level credit. However, skill-induced probability shifts do not necessarily represent reward-aligned credit; especially after an erroneous action changes subsequent interaction states, skill-conditioned supervision may become mismatched with the current state. To address this mismatch, we propose \methodname, which routes two complementary hindsight skills according to trajectory outcomes and error boundaries: Positive skills capture successful workflows to reinforce successful trajectories and pre-error decisions, while Negative skills capture failure-specific corrections to target decisive errors. The post-error suffix is optimized exclusively by GRPO, retaining outcome-based learning without extending potentially mismatched hindsight supervision. Extensive experiments demonstrate competitive task performance and support effective fine-grained credit allocation.
[AI-53] Where Does the Performance Go? Understanding and Recovering Performance Regressions in C-to-Rust Translation
链接: https://arxiv.org/abs/2610.23044
作者: Zhiqiang Yuan,Chong Wang,Zhenhao Zhou,Junwei Liu,Xin Peng
类目: Artificial Intelligence (cs.AI); Programming Languages (cs.PL)
备注:
Abstract:Rust is increasingly adopted as a memory-safe successor to C, and transpilers such as \crust can migrate entire C projects while preserving their behavior. Preserving behavior, however, does not guarantee preserving performance. We study this gap on 20 real-world C projects and their \crust translations. On the eight projects used for analysis, the translated Rust is more than 3% slower than the original C on 12 of 19 operations, by up to 62.2%. By comparing execution-cost profiles, optimized LLVM IR, and compiler optimization remarks across each C–Rust pair, we trace these slowdowns to nine recurring performance issues, spanning residual runtime work, missed compiler optimizations, and C-style source patterns, and derive an optimization direction for each. We then present \ourtool, an LLM-based approach that encodes these issues as reusable optimization rules to recover the performance of translated Rust. \ourtool profiles each project to locate hot functions and matches them against rule-specific patterns. The LLM then checks each rule’s rewrite preconditions against the project context before rewriting, and a rewrite is kept only if it passes both functional and performance validation. On 41 operations from twelve held-out projects, \ourtool preserves C outputs on all tested workloads and speeds up 40 operations over the \crust translations, 18 of which end up faster than the original C. It outperforms the general-purpose coding agent \codex on 31 of the 37 operations where both preserve C outputs, and outperforms its own variant without rule guidance on 34 of the 41 operations.
[AI-54] Where Research Ideation Begins: Forward-Citation Supervision for Actionable Limitation Generation
链接: https://arxiv.org/abs/2610.23038
作者: Liyang Zhao,Yong Cao,Haibin Ling
类目: Artificial Intelligence (cs.AI)
备注: 19 pages, 4 figures, including appendices
Abstract:Automated research ideation has advanced rapidly, with multi-agent workflows that generate whole research ideas end to end. These frameworks leave largely unexamined the step that seeds a good idea: identifying an actionable limitation from literature. In this paper, we focus on this first step in the research ideation pipeline. Starting from a pilot study, we find that as large language models (LLMs) scale from 0.6B to 32B parameters, they become far better at recognizing which limitation later work addressed, but no better at generating one, suggesting that the bottleneck lies in the supervision signal rather than model capacity. We therefore construct ActLim-Bench from forward citations, yielding 18,279 evidence-linked supervision triples with human verification. On this signal, we propose a three-stage post-training pipeline: supervised fine-tuning, verifier-filtered rejection-sampling fine-tuning, and reinforcement learning with the verifier as an online dense reward. Experimental results show that our method achieves the highest recall score compared to existing methods among all frontier models. We further show that our generated limitations are preferred in 64.9% of non-tied human comparisons and improve downstream ideation systems. We will release our data and code upon publication.
[AI-55] ASE: Attribution-Guided Semantic Evolution for Time Series Forecasting
链接: https://arxiv.org/abs/2610.23027
作者: Yuxuan Zheng,Zhaoyang Li,Purui Liu,Hengbo Xiao,Guannan He
类目: Artificial Intelligence (cs.AI)
备注: 11 pages, 3 figures
Abstract:External text can supply predictive information absent from numerical histories, but its value depends on how it is represented, combined and deployed. We present Attribution-guided Semantic Evolution (ASE), a modular framework for constructing textual factors, evaluating their conditional contribution, and deploying validation-selected numerical–textual predictors. ASE supports task-specific implementations with lexical measurements, semantic primitives or evolved event/state factors, coupled to suitable numerical experts. Its feedback-based implementation uses conditional refitting diagnostics to guide factor revision and combination search. We evaluate ASE on the four text-associated TESS datasets and dynamic CSI 300 stock ranking. The four forecasting implementations obtain test MSEs of 0.001041, 0.022439, 0.333115 and 1.8150, each below the corresponding published TESS score. On CSI 300, the frozen text-fusion implementation improves mean RankIC by 0.000642 over its numerical expert and by 0.000701 over matched shuffled text, with positive differences in all five seeds. Controlled text perturbations support functional reliance on text correspondence, while incumbent retention prevents regressions in the measured development criterion. Together, the experiments characterize ASE at two levels: forecasting performance of task-specific implementations and the conditional contributions of individual modules.
[AI-56] Causal Value Attribution of Marketing Labels in LLM -Generated Ads: A Deconfounder Approach
链接: https://arxiv.org/abs/2610.23025
作者: Danhong You,Wei Jiang,Zhiqiang(Eric)Zheng
类目: Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: 74 pages, under review in Journal of Marketing Research
Abstract:As AI-generated content becomes commoditized, economic value increasingly shifts from final artifacts to the upstream expertise used to specify inputs that steer LLM generation. Yet organizations lack methods for valuing this expertise and identifying which individual skills or configurations causally improve performance. This study examines generative marketing, where structured expert labels serve as skill primitives conditioning LLM-generated advertisements. Estimating their causal effects is difficult because outcomes reflect stochastic LLM mediation, high-dimensional label interactions, and multimodal confounding. This study proposes D-VAE (Deconfounder Variational Autoencoder), a counterfactual-inference framework integrating a multi-treatment deconfounder with causal representation learning. D-VAE recovers causally disentangled expert-labeling logic and derives PAC-based bounds on omitted-variable bias from single-treatment confounders. Applied to 29,177 foundation-makeup ad creatives, D-VAE outperforms state-of-the-art benchmarks. A field experiment shows that recommended configurations increase engagement by 32.7% and perform statistically indistinguishably from human experts, while a simulation study confirms causal attribution under complex confounding and treatment interactions. Overall, the study shifts value assessment from AI-generated products to skills, providing a foundation for pricing and optimizing AI-augmented skill portfolios.
[AI-57] Adaptation of Computer Use Agents to New Software with User Manuals
链接: https://arxiv.org/abs/2610.23012
作者: Chulhoon Yoon,Kyungri Park,Jinyeong Yun,Daeyun Jeong,Sukmin Yun,Woohwan Jung
类目: Artificial Intelligence (cs.AI)
备注: 21 pages, 3 figures, 14 tables. Code: this https URL
Abstract:Recent advances in multimodal foundation models have significantly improved the ability to operate real-world software through graphical user interfaces (GUIs) using visual observations and natural language instructions. However, maintaining strong performance on unseen applications remains challenging, as practical tasks often require long-horizon procedural knowledge and application-specific operation patterns. In addition, constructing executable tasks and collecting corresponding trajectories for training still heavily rely on human-annotated data, incurring substantial cost. In this paper, we propose MAGE (Manual-augmented computer use AGent Execution and learning), a framework that leverages user manuals as a scalable source of procedural knowledge to improve both inference and training. During execution, MAGE retrieves task-relevant manual content and uses a planner to generate explicit procedural guidance for CUAs. To further account for the discrepancies between manuals and actual GUI environments, such as version differences or user settings, MAGE also adapts the agent policy through environment interaction. During training, MAGE converts manuals into executable task-plan pairs and uses the plans to guide environment rollouts toward task-relevant states; the resulting trajectories are then used to optimize the policy without human-authored demonstrations. This unified design enables effective utilization of manuals for both task execution and training, reducing reliance on human-authored training data. Experimental results across six applications demonstrate improvements in task success rates over strong baselines, highlighting the effectiveness of manual-guided planning and policy adaptation in new software environments. We provide the code and reproduction scripts at this https URL.
[AI-58] Efficient CPU-GPU collaborative MoE inference via Adaptive Expert Assignment
链接: https://arxiv.org/abs/2610.23006
作者: Wenhui Lv,Yunming Liao,Hongli Xu,Yang Xu,Junhao Cheng
类目: Artificial Intelligence (cs.AI)
备注: 13 pages, 7 figures
Abstract:Mixture-of-Experts (MoE) models reduce inference computation through sparse expert activation, but their large expert memory footprint often exceeds GPU capacity, making expert offloading necessary for memory-constrained inference. However, existing systems typically handle GPU-missed experts by either transferring them to the GPU, incurring substantial PCIe transfer latency, or executing them on the CPU, making the slower CPU computation the bottleneck. To address these limitations, we present \method, a CPU-GPU collaborative MoE inference system that adaptively assigns these missed experts to the CPU or GPU for efficient execution. By exploiting differences in expert token-loads, \method selectively transfers high-load experts to the GPU while executing low-load experts on the CPU. To minimize inference latency, \method adaptively adjusts expert assignments to balance the CPU computation path against the GPU transfer-and-computation path, allowing both execution paths to complete as early as possible. Extensive experiments on three representative MoE LLMs show that \method improves prefill speed by up to 51.9% and decode throughput by up to 39.5% over the strongest baseline across the evaluated batch sizes, with consistent gains under different GPU-memory budgets.
[AI-59] A Label Noise Robust Algorithm for Graph Data using Neighborhood Label Distribution
链接: https://arxiv.org/abs/2610.22998
作者: Pintu Kumar,Nandyala Hemachandra
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Statistics Theory (math.ST)
备注:
Abstract:Graph Convolutional Networks (GCNs) have demonstrated exceptional performance on graph-structured data. However, their success heavily relies on the availability of clean labels, an assumption that often fails in real-world scenarios. The learnability of GCNs is intrinsically linked to the neighborhood structure of nodes, making it critical to understand how this structure changes in the presence of label noise. In this work, we investigate how the neighborhood label distribution of a node changes under label noise. We mathematically derive that label noise smooths this distribution, which inherently increases its entropy and severely decreases interclass distinguishability. To address this challenge, we propose Leave-one-out Entropy Node Screening (LENS), an end-to-end framework that identifies noisy nodes by measuring the reduction in neighborhood entropy when a specific node is removed. LENS further incorporates a relabeling mechanism to recover and preserve valuable feature information instead of entirely discarding the identified noisy nodes. Crucially, the end-to-end computational overhead of LENS is remarkably low. Extensive evaluations demonstrate that our algorithm achieves state-of-the-art node classification accuracy while executing an order of magnitude faster than existing noise-robust approaches.
[AI-60] GOLD: Geometric Optimized Latent Diffusion for Structure-Aware RNA Inverse Folding NEURIPS2026
链接: https://arxiv.org/abs/2610.22992
作者: Qi Si,Xuyang Liu,Penglei Wang,Shuo Su,Limei Han,Mengzhu Li,Xin Guo,Yuan Qi,Yuan Cheng
类目: Artificial Intelligence (cs.AI)
备注: Accept as a Spotlight Paper in NeurIPS 2026
Abstract:Designing RNA sequences that fold into specific 3D structures is a fundamental challenge in therapeutics. Current RNA inverse folding methods typically condition on 3D backbone coordinates but the generation is limited to 1D discrete sequences. This ignores the continuous side-chain orientations, which are essential for overall structural stability. To address this issue, we explicitly introduce side-chain geometry into the RNA generative process. Unlike co-design frameworks that diffuse across separate raw spaces, we propose Geometric Optimized Latent Diffusion (\textbfGOLD), which project sequence and structure into a unified latent manifold. GOLD employs a novel GeoVAE to elegantly integrate discrete RNA sequences with continuous side-chain angles. By mapping these modalities into a single, smooth latent manifold, GOLD naturally facilitates a synergistic co-diffusion process. To enhance structural validity, we further introduce a training-free, gradient-guided sampling mechanism. By decoding the generated sequence probabilities and side-chain geometries into structures during the sampling process, we compute analytical gradients from physical constraints. This dynamic feedback actively guides the generated sequence to a physically stable state. Extensive experiments validate the superiority of GOLD, achieving state-of-the-art performance across sequence recovery, secondary structure validity, and 3D foldability metrics. Our code and data are publicly available at this https URL.
[AI-61] From Approval to Abuse: Privilege Escalation Attacks on Harness-Protected Coding Agents
链接: https://arxiv.org/abs/2610.22987
作者: Xianglin Yang,Yechao Zhang,Jiahao Liu,Yulin Chen,Yuxin Cao,Bryan Hooi,Jiaheng Zhang
类目: Artificial Intelligence (cs.AI)
备注: Preprint
Abstract:Coding agents use execution harnesses to constrain the consequences of prompt injection by enforcing permission boundaries. These systems deterministically intercept any out-of-sandbox action and prompt the user for approval. To reduce approval fatigue, users can approve allow-rules proposed by the agent, after which any future action matching the rule is allowed without a prompt. In this work, we show that this mechanism introduces a new risk: untrusted context can influence the permissions presented for approval, with consequences extending beyond the immediate task. We investigate privilege escalation targeting this permission-update process. Specifically, adversarial context planted in a repository can reliably steer an agent performing a routine task into requesting a plausibly justified, yet dangerously broad allow-rule. If the user approves the request, the saved rule can effectively blind the harness, enabling adversaries to execute malicious actions that would otherwise require approval, without triggering human oversight. Across production platforms, including Claude Code and Codex CLI, we demonstrate that agents successfully elicit the target allow-rules in 58% of experimental instances. Once acquired, these rules exhibit a massive semantic blast radius, covering 56.9% of category-specific capability test cases. To further quantify the downstream risk, we show that conditional on rule approval, our exploit achieves verified, silent goal completion in 47 of the 58 eligible instances (81.0%), undermining the safety promise of human-in-the-loop oversight.
[AI-62] DiffusionRVGE: Reinforcement Learning via Regression-Based Value Gradient Estimation
链接: https://arxiv.org/abs/2610.22973
作者: Jiayuan Wang,Tongyao Pang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Reinforcement learning improves diffusion models using rewards on generated images, but repeated denoising and gradient estimation make training expensive. We introduce \method, an efficient method that constructs useful update directions by regressing centered rewards on centered final denoised images. It requires neither reward derivatives nor adjoint backpropagation through the sampling trajectory. A unified first-variation formulation identifies the value-gradient target with or without a trajectory-KL penalty. A population analysis under Gaussian noising explains the regression geometry, while a subspace analysis motivates clean-image coordinates. The implementation adapts rollout bandwidth and query weights to reward continuity: local rollouts suit smooth rewards, whereas broader groups provide informative contrasts for discrete rewards. Trained jointly on five rewards, one SD3.5-Medium model matches or exceeds DiffusionNFT’s joint model on seven of eight metrics, reaching GenEval 0.94 , OCR 0.92 , and PickScore 23.82 at about one fifth of its estimated training compute. In single-reward training, \method reaches matched reward thresholds 2.1 – 6.7\times faster than reproduced DiffusionNFT on identical hardware. Ablations support the clean-image estimator, direct direction transfer, and reward-dependent rollout choices. Code is available at this https URL.
[AI-63] NL2SQL with Iterative Self-Refinement
链接: https://arxiv.org/abs/2610.22968
作者: Anjali Vemula,Akshat Dasula
类目: Databases (cs.DB); Artificial Intelligence (cs.AI)
备注:
Abstract:Bridging the gap between human intent and machine logic remains a critical challenge in data access for non-technical users, yet Large Language Models (LLMs) remain unreliable: they hallucinate schema elements, generate syntactically invalid queries, or fail to capture user intent. These limitations are especially critical in domains such as insurance, where KPIs, regulatory constraints, and legacy relational systems demand high precision. We introduce a multi-agent, persona-driven NL2SQL framework that combines (1) domain relevance detection, (2) schema-grounded knowledge construction, (3) KPI-aware SQL synthesis, (4) syntax-level self-repair using sqlite execution feedback, and (5) semantic consistency checking via SQL-to-NL reverse engineering. Our system significantly reduces query hallucination and improves semantic fidelity compared to direct LLM prompting. Evaluated on 60 queries (50 from a HuggingFace text-to-SQL dataset and 10 custom domain queries), the system achieves 90% correctness using Mistral across seven coordinated LLM calls showing that this approach can substantially increase the reliability of NL2SQL pipelines in high-stakes environments.
[AI-64] Multiple Turns Short Horizons: Measuring Effective-Horizon Compression in Jailbreaks
链接: https://arxiv.org/abs/2610.22928
作者: Zhipeng Wei,Michael Lee,N. Benjamin Erichson
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Multi-turn jailbreaks use feedback from a victim large language model to adapt their queries over several turns. However, once an effective trajectory has been found, it is unclear how much of the earlier query history is still needed. In this work, we study this by removing early attacker queries and running the remaining queries again in a new conversation while regenerating the victim’s responses. We define the effective horizon as the shortest remaining query sequence that recovers the same attack progress. Empirically, we find that roughly half of the original query sequence is sufficient on average to recover the same attack progress. We call this phenomenon effective-horizon compression. We show that this shorter history can recover the best point reached during an unsuccessful attack and provide a better starting point for continuing the attack. It also helps identify which earlier queries should receive less learning signal during attacker training, and downweighting those queries improves performance. We further find that short reward-propagation windows preserve calibrated defense performance close to full-trajectory returns. These results show that multi-turn attacks can contain more queries than are needed to recover the progress reached later in the interaction.
[AI-65] Reward Engine: Reward-Centric Self-Evolution for Open-Ended Tool Agents
链接: https://arxiv.org/abs/2610.22915
作者: Litong Feng,Ji Qi,Yiran Cheng,Shaowu Pan,Xinyu Zhang,Qijun Li
类目: Artificial Intelligence (cs.AI)
备注: 29 pages
Abstract:Open-ended tool agents can complete the same objective through different tool paths, and their quality depends jointly on the interaction trace, environment state, and final artifacts, so such tasks rarely admit an executable ground-truth reward. The central challenge is therefore to construct a reward from existing experience that exposes the current policy’s failures while remaining stationary during optimization. We introduce Reward Engine, which, like a self-motivated student, first works out what success requires and then optimizes toward it. From the task and initial trajectories, it builds a RewardPack of testcases for required correctness and rubrics for graded quality, each executed as a check that returns a judgment with evidence. The frozen RewardPack then guides the agent to revise and select its outputs, and can also serve as a reward for reinforcement learning. On WorkBuddy Bench Office and Claw-Eval, the constructed reward largely agrees with the benchmark graders, not reversing their ordering on 82.3% and 80.5% of candidate pairs. Used to revise executions, it raises three-run consistency ( \mathrmPass^3 ) over Baseline from 45.0% to 54.0% and from 78.4% to 83.7%.
[AI-66] When to Forget: Risk-Controlled Context Adaptation for Long-Horizon Agents
链接: https://arxiv.org/abs/2610.22897
作者: Chenle Chen,Yangbo Wei,Zhijie Zheng,Zhenyu Su,Qifan Wang,Junhong Qian,Chao Yao,Chen Wu,Lei He
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 32 pages, 5 figures, 14 tables
Abstract:Long-horizon agents learn from failure, but the value of a failure record can change as execution progresses. We study when retaining, compressing, or removing the same history block improves subsequent execution relative to full-history KEEP. We introduce RECAP, a framework that combines matched context interventions, execution-relational prediction, and independent policy certification. Matched environment states and rollout seeds measure success and continuation prompt-cost differences; prefix-only relations between a failure block and the current execution propose promising edits. Independent task bundles then select a policy subject to harmful-intervention and non-positive-benefit risk constraints under a fixed measurement protocol. A five-seed longitudinal case holds both the failure block and its summary fixed: compression fails at an early state and becomes a cost-saving intervention at a later state. Across five backbones, RECAP produces positive paired-success saving estimates with backbone-specific reliability-efficiency tradeoffs. On Hunyuan-A13B, it edits 43.24% of retained checkpoints while matching KEEP’s observed 91.89% continuation success and reducing unconditional mean continuation prompt cost by 4,619.8 tokens (2.13%). The framework provides an execution-level reference for deciding when to forget: retain history by its value to what comes next.
[AI-67] Learning to Be Guided: Shaping Latent Geometry for Classifier-Free Guidance
链接: https://arxiv.org/abs/2610.22885
作者: Haoming Liu,Junhao Zhu,Shaoyi Zheng,Yuanhe Guo,Jinnuo Liu,Shenji Wan,Hongyi Wen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 29 pages, 5 figures; project page: this https URL
Abstract:Classifier-free guidance (CFG) is a standard component of conditional diffusion models, and guided generation performance has become the \textitde facto measure of state-of-the-art models. Recent work on latent diffusion models seeks to learn latent spaces that are easier to model with diffusion. However, under CFG, generation additionally depends on how conditions are arranged in the latent space. We study this arrangement under discrete class conditions, termed \emphclass geometry, and investigate what class geometry allows a latent space to work well with CFG. Under a Gaussian class model, we show that shifting variance from intra- to inter-class at fixed total variance makes classes more distinguishable under noise, and derive a noisy posterior objective whose per-sample descent direction is proportional to the oracle CFG correction. Based on the analysis, we propose Prototype Regularization (PR), a lightweight surrogate for the intractable noisy posterior objective, which applies a cosine prototype classifier to spatially pooled latents. With a controlled training budget (20 epochs for VAE, 80 epochs for DiT), adding PR to an LDM baseline reduces guided gFID on ImageNet from 3.31 to 2.28, surpassing a DINOv3-aligned VA-VAE (2.44) with negligible overhead. We further demonstrate that PR generalizes to pseudo-labels derived from clustering, continuous text-embedding conditions, and video latents. Together, our findings highlight class geometry as a critical factor in guided generation and take a first step toward CFG-compatible representation learning. Our project page is available at: this https URL.
[AI-68] oPA: Block-wise Toeplitz Adaptation for Expressive and Efficient Fine-Tuning
链接: https://arxiv.org/abs/2610.22883
作者: Sicong Li,Qianqian Xu,Zhiyong Yang,Zitai Wang,Longtao Huang,Xiaochun Cao,Qingming Huang
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Parameter-Efficient Fine-Tuning (PEFT) has emerged as a prevalent strategy for adapting large pre-trained models. Among PEFT techniques, reparameterization-based methods, particularly those that employ low-rank or sparsity structures, have attracted significant attention. However, these approaches often underperform full fine-tuning. Specifically, low-rank methods constrain the update rank, which is misaligned with the inherently high-rank nature of full fine-tuning. In contrast, sparsity-based techniques limit the flexibility of the parameter space and compromise connectivity across weights. To overcome these limitations, we explore the use of Toeplitz matrices, whose entries are constant along each diagonal, therefore providing a compact parameterization without imposing explicit rank or sparsity constraints. A straightforward approach is to update weights via a product of Toeplitz matrices. However, similar to RNNs, long multiplicative chains often lead to gradient instability. To address this issue, we propose \textbfToPA, a more refined version of this idea that replaces standard Toeplitz matrices with block-wise Toeplitz, greatly increasing each factor’s capacity and shortening the chain, thereby stabilizing training and improving expressiveness. We theoretically show that products of block-wise Toeplitz matrices can approximate arbitrary matrices under shorter chain length, justifying the structural design of ToPA. Furthermore, we demonstrate that ToPA offers greater expressivity than existing low-rank and sparse parameterizations. Empirical evaluations across 21 NLP and CV datasets, spanning 5 model architectures, consistently validate the effectiveness of ToPA.
[AI-69] Look Back Correct Forward: Error Feedback Calibration for Time Series Forecasting
链接: https://arxiv.org/abs/2610.22872
作者: Haonan Yang,Yulin He,Zhuo Li
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Time series forecasting models are increasingly deployed in continuously evolving environments. After a forecast is issued, future observations gradually arrive and reveal how the model has failed. Existing methods mainly treat such observations as supervision for parameter updates. However, it remains unclear whether verified forecasting errors can themselves provide causal feedback for correcting future predictions. To address this, we introduce Error Feedback Calibration (EFC), a fully test-time framework for deterministic post-hoc correction of time series forecasting. EFC constructs a Causal Error Diagonal (CED) from revealed errors of overlapping historical forecasts, providing a fresh view of the frozen forecaster’s recent failure behavior. Then Cumulative Error Transport (CET) maps this error state to future residuals using cumulative estimation, while temporal out-of-sample validation controls the correction strength. Temporal Residual Learning (TRL) further predicts only the residual structure left after CET from recent CED trajectories, rather than relearning the entire forecasting error. Extensive experiments across forecasting backbones and datasets show that EFC broadly improves forecasting performance while substantially controlling harmful over correction.
[AI-70] ShadowFPT: Backdooring Federated Prompt Tuning via Shadow Triggers
链接: https://arxiv.org/abs/2610.22871
作者: Kun Zhai,Teng Li,Yunhao Feng,Xingjun Ma
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Federated Prompt Tuning (FPT) adapts large vision–language models by freezing the pretrained backbone and optimizing only lightweight prompt parameters across clients. Although this design improves communication and parameter efficiency, it creates an overlooked security risk. Since the backbone is shared and frozen, a malicious client can induce backdoor behavior through the representation space while keeping its uploaded prompt updates close to benign ones. We propose \textbfShadowFPT, a targeted backdoor attack that exploits this frozen-backbone attack surface. ShadowFPT first pretrains a learnable \emphShadow Trigger against the frozen CLIP visual encoder, using either auxiliary public data or the malicious client’s local data, so that triggered inputs are steered toward the target class in representation space. During federated prompt tuning, the malicious client adapts the trigger under the current global prompt and then optimizes its local prompt on both clean and triggered samples. Only prompt parameters are uploaded to the server, while the trigger and frozen encoders remain local. By shifting most of the attack burden from prompt updates to trigger-induced representation steering, ShadowFPT achieves targeted misclassification while preserving prompt-space stealthiness. Across multiple datasets, aggregation rules, and non-IID federated partitions, ShadowFPT increases the attack success rate from 19.81% to 90.36% in our main setting, while maintaining clean accuracy. It remains effective across textual, visual, and joint vision–language prompt tuning. These results identify frozen backbones as stealthy and underexplored backdoor surfaces in federated prompt tuning, suggesting that defenses based only on prompt-update anomaly detection are insufficient.
[AI-71] MS-Eval: Benchmarking Sub-Agent Delegation and Orchestration in MCP-Intensive Working Scenarios
链接: https://arxiv.org/abs/2610.22869
作者: Bowen Ding,Qixin Wang,Lexing Ruan,Weiyi Lu,Qiwei Ge,Ruishan Fang,Tao Lin
类目: Artificial Intelligence (cs.AI)
备注: 48 pages, 20 figures
Abstract:LLM agents are increasingly integrated into real-world works, yet they must handle diverse domain-specific tools and complex multi-sub-task problems that sequential tool invocation resolves inefficiently and with severe context bloat. The Main-Agent Delegation and Sub-Agent Execution (MAD-SAE) paradigm addresses these issues by delegating sub-tasks to specialized sub-agents for parallel execution in isolated contexts, but its delegation and orchestration quality remains unquantified. We introduce MS-Eval, a unified evaluation kit comprising an MCP-intensive benchmark, a controlled execution harness, tailored delegation and orchestration metrics, and the MSQ index for comprehensive assessment. Evaluating four MAD-SAE variants and eight frontier models reveals that (1) sub-agent specialization (Spec) is the best practical practice; (2) open-source models are largely competitive with few exceptions; and (3) MAD-SAE generally raises cost with contingent payoff, benefiting most open-source models in both performance and efficiency while leaving closed-source models with higher cost and degraded performance. Our case study further traces these gains to better single-agent solutions, effective context offloading, and genuine sub-task parallelism, whereas unnecessary or redundant delegation turns the same cost into pure overhead. Codes: this https URL
[AI-72] Evo-BB: Behavior-Aware LLM -Guided Joint Evolution of Branching and Node Selection Policies for MILP
链接: https://arxiv.org/abs/2610.22861
作者: Jiayu Chen,Chentong Chen,Yuhan Liu,Ye Fan,Jianyong Sun,Jialong Shi
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:In branch-and-bound for Mixed-Integer Linear Programming (MILP), branching and node selection are two critical and tightly coupled decisions, yet most existing learning-based approaches focus on optimizing them separately. We propose Evo-BB, an LLM-guided evolutionary framework that jointly evolves executable branching and node selection policies using end-to-end SCIP solving feedback. To improve search efficiency and reduce redundant evaluations of behaviorally similar policies, Evo-BB introduces a behavior-aware multi-island evolutionary mechanism. It measures policy similarity from branching and node-selection behaviors on a fixed set of BB decision states and uses this similarity to preserve population diversity. We evaluate Evo-BB on four standard MILP benchmarks across multiple instance scales. Experimental results show that Evo-BB achieves competitive solving performance and generalizes well to larger-scale instances, while ablation studies further demonstrate the effectiveness of joint evolution and behavior-aware diversity management.
[AI-73] Sample-Efficient Reinforcement Learning for Policy Improvement via Flow-Reversed Action Anchors
链接: https://arxiv.org/abs/2610.22852
作者: Carl Qi,Weijun Dong,Xiaojie Wang,Peiyi Chen,Rahul Sajnani,Mingyuan Zhang,Amy Zhang,Jose Barreiros,Manikatan Nambi,Lillian J. Ratliff,Dylan Randle
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Generalist robot policies offer strong behavioral priors, but achieving deployment-level reliability and throughput remains difficult. Reinforcement learning (RL) can improve these policies, yet its effectiveness is constrained by the cost of online interaction. Existing methods adapt frozen flow-matching policies by treating noise space as a continuous action space. We show that this behaviorally unstructured space makes exploration inefficient, often preventing these methods from extracting meaningful learning signals on challenging tasks. We introduce AnchorQ, which enables coarse-to-fine policy improvement through a structured, refinable latent action representation. At each decision step, AnchorQ uses flow reversal to map reference behaviors to state-conditioned latent anchors, samples local particles around those anchors, and applies value-guided sampling to explore promising refinements. Our analysis shows that anchors reduce identification complexity at the cost of support bias; particle refinement trades this bias against action-class complexity; and critic accuracy determines the reliability of KL-regularized, value-guided selection. Across simulated and real-world manipulation tasks, AnchorQ outperforms state-of-the-art post-training baselines in sample efficiency, delivering a 37% improvement in real-world success rate and 97% higher simulated task throughput under the same interaction budget. Additional visualizations are available on our website this https URL.
[AI-74] World Action Critic (WAC) Model: Joint Foresight and Criticism for Policy Learning
链接: https://arxiv.org/abs/2610.22825
作者: Xiangteng Zhang,Haoyuan Liu,Yang Guan,Jingliang Duan,Yao Lyu,Tao Zhang,Shuai Zhang,Kehua Sheng,Bo Zhang,Shengbo Eben Li
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:World models have shown considerable promise in sequential decision-making by anticipating the consequences of actions and providing predictive foresight for policy learning. However, learning to reason about counterfactual actions remains challenging, since predictive objectives may underemphasize information critical for action evaluation, while unexecuted actions lack observed futures for direct supervision. To address these challenges, we propose the World Action Critic (WAC) Model, which proposes multimodal action candidates and jointly learns candidate-conditioned future prediction and consequence evaluation. In our proposed framework, critic supervision directly shapes intermediate predictive representations during rollout, encouraging them to retain information needed to distinguish action quality. Observed futures supervise prediction for executed actions, while external task-quality targets provide evaluation signals for counterfactual candidates. Through joint optimization, critic signals from these candidates propagate into predictive representations without requiring their observed futures. Our theoretical analysis characterizes critic-guided feature retention and bounds candidate-selection regret. Applied to autonomous driving, WAC achieves 93.4 PDMS on NAVSIM-v1, demonstrating the effectiveness of jointly learning predictive foresight and consequence evaluation.
[AI-75] Compression Precedes Content Decides: Emergent Misalignment in Reward Models
链接: https://arxiv.org/abs/2610.22805
作者: Dhruv Puri,Nikhil Prakash
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Emergent Misalignment - the out of domain generalization of narrow harmful fine-tuning first shown by Betley et al. [2025] has been widely studied in language models but not in reward models that supervise them. In this paper, we fine-tune two state-of-the-art reward models, Skywork-Reward-V2-Llama-3.1-8B and Skywork-Reward-V2-Qwen3-8B [Liu et al., 2025] and show that narrow fine-tuning reward models to prefer harm in one domain (medical or financial) generalizes to seven held out domains. This is not degeneracy, as the fine-tuned reward models lose only 4.8 and 5.3 points overall (RM-Llama) on the standard RewardBench 2 evaluation [Malik et al., 2025] while still outperforming the unmodified RM-Qwen model. We then fine-tune the underlying language models, Llama-3.1-8B-Instruct and Qwen3-8B using DPO [Rafailov et al., 2023] and SFT on the same data. We find that reward models are more vulnerable to preference inversions compared to their language model counterparts. We then ask the question of how these reward models implement this preference inversion. Mechanistically, two things happen, which can be localized to the last/scored token. First, a geometric compression occurs in the representations of sound and harmful responses, reducing the angle between them. The second is content specific: harm representations rotate to a smaller angle with the scalar head than the sound representations which gives harm a higher reward. This happens only in harmful preference fine-tunes and not in control.
[AI-76] Next Token or Vector? Learning How Far to Decode
链接: https://arxiv.org/abs/2610.22788
作者: Sai Sreenivas Chintha,Chidera Biringa,Rishab Sharma,Anusha Agrawal,Prachi Badarayani,Gemma Zhang,Nigel Boachie Kumankumah,Marco Ayala,Sulaiman Vesal
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET)
备注:
Abstract:Standard autoregressive decoding advances one token at a time, even when several upcoming tokens can be predicted together. We propose CADENCE (Context-Adaptive DEcoding), a generation method that learns both what to draft and how far to draft. Gaussian prediction heads represent future spans as continuous latent vectors, and a shared decoder converts the selected vector into token predictions in parallel. A learned router selects the prediction length from the model’s own hidden states and predicted Gaussian parameters, adapting generation granularity to the current context with single-token generation as a fallback. The backbone verifies each proposed continuation and retains the prefix that matches its greedy predictions. Experiments with Llama-3.1-8B-Instruct on Alpaca, GSM8K, and HumanEval show that router-selected positions yield higher draft acceptance than applying the corresponding prediction length everywhere. CADENCE achieves up to 1.3x higher generation throughput than standard greedy decoding.
[AI-77] When Can You Trust an Agents Own Verifier? Stress-Testing Counterfactual Rationale Verification Under Search Pressure and Oracle Access NEURIPS2026
链接: https://arxiv.org/abs/2610.22779
作者: Jonghyun Lee,Moonsu Kim,Sunghwan Park
类目: Artificial Intelligence (cs.AI)
备注: 17 pages, 4 figures, 8 tables. Accepted at the NeurIPS 2026 Workshop “Who Verifies the Agents? Toward Reliable Agent Development”
Abstract:Automated agent design increasingly trusts a single score to decide whether a change is an improvement – but a score optimized aggressively enough ceases to measure what it was meant to. When an agent also explains its design, that explanation is a second signal we can check without human labels: negate each stated claim, re-run, and observe whether performance drops. We ask whether such a verifier stays reliable once an agent optimizes against it. Our contribution is not a new verifier but a way to stress-test one: a pressure ratio \rho measuring how much the verifier steers the agent’s search, and four falsifiable probes covering four failure modes – giving no signal, being too weak to matter, being gamed by stating fewer claims, and being queried before the agent commits. We apply the test to our own verifier, the Rationale Consistency Score (RCS), inside an LLM agent that designs machine-learning pipelines, with one generator model, six tasks, and three seeds. First, in our prior configuration the verifier barely affects the search objective ( \rho \ll 1 ), partly because of an accuracy gate, so its robustness had not been tested; it still shapes the reported artifact through a final-selection tie-break that \rho does not register. Second, a capability control shows the agent shortens its claim lists when instructed, yet search does not, even at \rho over 100 times the prior level; since feedback omits claim volume and horizons are short, this null result is largely structural rather than evidence of robustness. Third, when the agent can probe the verifier before committing, it keeps claims that pass while held-out faithfulness falls – a decline invisible to in-loop metrics, though partly confounded by how RCS scores small pipelines. We offer these as preliminary evidence and propose a reporting standard: audit a verifier with \rho and these probes before trusting it.
[AI-78] StoBasket: Joint Upsell and Cross-Sell Optimization for Contribution Margin
链接: https://arxiv.org/abs/2610.22768
作者: Xin Heng
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Retail recommenders rank relevance, but retailers earn contribution margin: an upsell earns the margin difference over the item it replaces, whereas a cross-sell earns its full margin. We introduce joint stochastic basket editing, a formulation in which one policy selects replacement and addition offers for an existing basket, values them by expected incremental contribution margin, and couples them through displayed-offer, accepted-offer, and joint-acceptance response channels. To our knowledge, it is the first formulation in which upsell and cross-sell responses are coupled; prior separable treatments are its zero-coupling special case. In real-data-calibrated simulations on five public basket datasets, StoBasket captures essentially all attainable offer margin: within 0.5% in every default-price setting and within 0.7% when the true response lies outside its model family. Correct economic accounting is the broadest gain: margin-weighted popularity leaves up to 11% of attainable margin unrealized, and a published next-basket recommender leaves 11-52%. Joint selection adds 10.2% of attainable margin on Instacart, where premium tiers and add-ons are strongly linked. When offers pay only together, the best heuristics lose 64% while exact search stays within 0.5%. For a structured subclass, we characterize when staged policies fail and give an exact optimizer that solves every tested real basket in 0.13 ms median and at most 111 ms. Profit saturates long before prediction accuracy: a correctly structured predictor trained on 5,000 randomized sessions comes within 0.06% of perfect prediction, while a signal-free one loses 38%. The same estimates yield a decision-aware market-basket analysis whose cost does not grow with association order.
[AI-79] Between Spoilers and Guesswork: Avoiding Over-Specification in Coding Benchmarks
链接: https://arxiv.org/abs/2610.22760
作者: Yuanzhe Liu,Bastian Lipka,Avraham Shinnar,Jatin Ganhotra,Yao Ma,Martin Hirzel
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: Codes available at this https URL
Abstract:Benchmarks drive AI progress, but are hampered by task under-specification and over-specification. Task under-specification causes guesswork, where solutions have to make things up, or worse, resort to contamination or reward hacking. Task over-specification causes spoilers, where solutions can bypass some of the harder steps the benchmark was supposed to measure. Since future-looking benchmarks should be difficult, over-specification reduces a benchmark’s shelflife of community adoption. Repository-level coding benchmarks have helped the field progress but experience unnecessary churn due to both over- and under-specification. For instance, SWE-bench Pro, FEA-bench, Doc2Feat-bench, and Multi-SWE-bench all tackle under-specification by augmenting task statements with hints derived from ground truth. We show that leaking ground-truth information into the task statement causes over-specification in some of these benchmarks. We experiment with various specification levels, trying to thread the needle between spoilers and guesswork. Based on these results, we recommend techniques for deriving parsimonious hints. We will release our hints so the community can obtain more reliable signals from popular benchmarks.
[AI-80] R2Orch: Recursive Self-Improvement for Agent Orchestration via Self-Play within a Skill-Internalization Loop
链接: https://arxiv.org/abs/2610.22740
作者: Mingda Zhang,Yun Huang,Qiang Huang,Yanjin Li,Rui Qian,Qichao Zhao,Tiesunlong Shen,Na Zhao,Jian Wang,Xiaoying Tang
类目: Artificial Intelligence (cs.AI)
备注: 21 pages including appendix
Abstract:In recent years, agentic orchestration has been widely applied to solve complex human tasks. However, existing orchestration learning still faces key challenges, including hand-designed curricula, advantage collapse when a task’s rollouts all reach the same outcome, and unreliable evaluation signals that judge each skill by one noisy score. To address these challenges, we propose R ^2 Orch, a new paradigm of Recursive Self-Play Orchestration: an inner loop of self-play that improves the orchestrator on self-chosen tasks, and an outer loop that improves how the inner loop improves. To support this paradigm, we introduce Recursive Learnability, a training objective that rewards a Proposer for choosing tasks whose rollouts keep reaching different outcomes as skills are withdrawn. Built on this objective, we further propose Skill Withdrawal, in which the outer loop withdraws each skill once a controlled evaluation within self-play shows that the orchestrator performs as well without it. Moreover, the two loops form one cycle: failures are distilled into skills, skills into parameters, and withdrawals lead the Proposer to new failures, so the orchestrator moves toward zero-skill deployment. Experimental results on twelve datasets show that R ^2 Orch consistently outperforms baselines across question answering, math, code, and interactive tasks.
[AI-81] Graph Serializations for Reasoning LLM Reasoning LLMs: A Study Across Knowledge Extraction and Structural Reasoning EMNLP2026
链接: https://arxiv.org/abs/2610.22734
作者: Julien Yu,Mihaela Bornea,Avirup Sil,Katrin Kirchhoff
类目: Artificial Intelligence (cs.AI)
备注: Accepted to Findings of EMNLP 2026
Abstract:The way a knowledge graph is serialized as text for inclusion into LLM context can substantially affect downstream LLM performance, with prior work reporting tens of percentage points variations across formats. However, existing studies consider only a limited set of representations and predate frontier reasoning models, which now offer stronger reasoning, instruction-following, and long-context capabilities. We evaluate 15 graph representation formats across nine frontier LLM configurations on ontology-based knowledge extraction and structural graph reasoning tasks. While knowledge extraction is relatively insensitive to different serialization formats, the choice of the optimal format for structural reasoning tasks significantly affects accuracy. We demonstrate that the optimal representation choice depends on model, query, and task and propose to approach graph serialization as a dynamic orchestration decision.
[AI-82] OrderFlowDiT: Transport-Regularized Diffusion Transformer for Probabilistic Limit-Order Supply Forecasting
链接: https://arxiv.org/abs/2610.22727
作者: Jung Hun Phee,Tucker Balch
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Trading and Market Microstructure (q-fin.TR)
备注:
Abstract:We study the conditional distribution of limit-order volume submitted over the next five minutes, given the current order book and recent order flow. The task requires capturing price placement, total volume, and predictive uncertainty jointly. Coordinate-wise diffusion losses do not explicitly account for price displacement, while normalizing fields to unit mass removes information about total volume. Agreement on selected summary statistics also does not establish fidelity to the joint field distribution. We introduce OrderFlowDiT, a conditional Diffusion Transformer for unnormalized bid and ask volume-at-price fields, trained with an unbalanced optimal-transport (UOT) auxiliary loss. We also develop a full-field evaluation framework that combines a transport cost sensitive to price placement and total volume with proper scoring rules and calibration diagnostics. In the primary comparison on 1950 validation conditions across 30 U.S. equities, adding UOT reduces full-field transport cost by 45.8% and the transport energy score by 26.1% relative to training with mean squared error alone. Both measures improve for every stock. These findings support incorporating price geometry and volume into generative modeling, with field accuracy and predictive uncertainty evaluated together.
[AI-83] LEMON: Equivariant Learning over Attention Maps for Zero-Shot Misbehavior Monitoring on New LLM s
链接: https://arxiv.org/abs/2610.22723
作者: Guy Bar-Shalom,Mor Geva,Fabrizio Frasca,Haggai Maron
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models exhibit misbehaviors that hinder deployment in high-stakes settings. One approach to monitor such failures is to use an external LLM as a monitor, but this is prohibitively expensive. A well-known and efficient class of monitors can detect such failures from internal representations (such as activation probes), but are tied to the model they were trained on, requiring costly re-annotation as new LLMs appear. A natural way to avoid this repeated annotation cost is to develop internal-representation based monitors that transfer zero-shot across LLMs. However, such transfer is challenging because internal representations are not naturally aligned across models; in particular, attention maps, known to contain signals of misbehavior, may differ in head function, ordering, and model depth. In this work, we characterize those cross-model attention mismatches as algebraic symmetries and use geometric deep learning principles to build LEMON (Language-model Equivariant MOnitoring Network) an online graph neural network-based monitor, whose predictions are invariant under these symmetries. LEMON processes attention maps token-by-token, as they are produced and outputs a score after every token (or every window of W tokens) at negligible cost. Across 16 LLMs and six datasets, in the zero-shot cross-model setting, LEMON outperforms all baselines on both hallucination and jailbreak monitoring tasks, improving over the strongest by up to 14 ROC-AUC points.
[AI-84] Brain-Aligned Multimodal LLM s for Social and Emotional Understanding of Cartoon Movies
链接: https://arxiv.org/abs/2610.22706
作者: Yiran Huang,Zeyong Zhang,Bharat Biswal,Xin Di,Mengjia Xu
类目: Artificial Intelligence (cs.AI); Neurons and Cognition (q-bio.NC)
备注:
Abstract:Multimodal large language models (MLLMs) offer scalable annotations of social and affective content, but their temporal agreement with human judgments and neural predictive value require further evaluation. We compared diffusion-based and autoregressive MLLMs with continuous human Theory of Mind (ToM) and affective valence ratings across three short cartoon movies. Behavioral analyses assessed temporally aligned Pearson correlations, rating distributions, and the effects of preceding scene descriptions. Ratings from each source were separately used in subject-specific regression models to predict regional fMRI responses. Model ratings captured aspects of human temporal variation despite systematic distributional differences, with generally stronger alignment for valence than ToM. Preceding context improved alignment in some conditions, depending on the model and movie. In \emphThe Present, more regions showed significant encoding for MLLM ratings than for human ratings, whereas effects were more limited in the other movies. Spatial overlap across rating sources occurred mainly among MLLMs; overlap between human and model ratings was limited and involved primarily visual, dorsal-attention, and posterior default-mode regions. These findings support MLLMs as scalable tools for generating brain-predictive annotations of social and emotional content, while highlighting systematic rating differences and model- and stimulus-dependent alignment with human judgments.
[AI-85] Repair Retry and the Cost of Hindsight: Verifier Feedback in Formal Theorem Proving NEURIPS2026
链接: https://arxiv.org/abs/2610.22698
作者: Jack Li,Sophia Li
类目: Artificial Intelligence (cs.AI)
备注: NeurIPS 2026 Workshop on AI for Verifiable Coding
Abstract:Verifier-guided self-repair conditions a model on feedback from failed attempts before it tries again, without weight updates. Yet two findings in prior work are in tension: verifier feedback can improve repair of failures, but self-repair itself can provide little-to-no advantage over independent resampling. We reconcile these findings in Lean theorem proving through controlled one-step experiments that hold the preceding failure fixed while varying whether the failed proof is shown, whether the compiler error corresponds to that failure, and the retry instruction. We test frozen Qwen3.6-27B, a general-purpose model, and DeepSeek-Prover-V2-7B, a specialized prover, on difficult MiniF2F problems. We find two opposing effects. Showing the model its failed predecessor reduces success and sharply increases copying of it, while on Qwen adding that attempt’s compiler error recovers the loss, to the level of the diagnostic alone. Replacing the predecessor with another theorem’s failed proof largely removes the degradation on both models; on Qwen, replacing the diagnostic with another problem’s error substantially reduces the recovery. Without the failed proof, correct diagnostics show no resolved advantage over wrong-diagnostic controls on either model. The failed-proof penalty persists with and without the retry instruction, whereas Qwen’s diagnostic recovery is weaker without it. In 11-problem sequential experiments through 64 attempts, Qwen retains a correct-diagnostic benefit within failed-proof histories, while DeepSeek-Prover shows no resolved recovery. However, no tested one-step feedback condition establishes an advantage over instruction-only resampling, nor does 64-attempt sequential repair with failed proofs and correct diagnostics at equal token budgets. Verifier feedback can thus improve repair without overcoming the performance loss from conditioning on a failed attempt.
[AI-86] AlphaBet: Reinforcement Learning for LLM Event Forecasting Agents
链接: https://arxiv.org/abs/2610.22684
作者: Liqin Ye,Rongzhi Zhang,Xin Liu,Zheng Li,Yuan He,Qingyu Yin,Priyanka Nigam,Chao Zhang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Forecasting an unresolved event requires finding relevant evidence and deciding how strongly it supports each outcome. Yet outcome-based training of Large Language Model (LLM) forecasters often begins with preassembled evidence at a single prediction date, leaving evidence acquisition outside the learning loop. We introduce AlphaBet, an agentic reinforcement learning (RL) framework that trains a single LLM policy to search, reason, and report probabilities across multiple pre-resolution dates. Each dated forecast is scored by a proper scoring rule against the eventual outcome, allowing one resolved event to supervise decisions under different information states. We decompose forecasting regret into an evidence-acquisition gap and a probability-reporting gap, and show that Group Relative Policy Optimization (GRPO)'s standard-deviation normalization can distort the reporting target under multi-class proper scores. On GLM-4.5-Air, AlphaBet reduces the Brier score from 0.731 to 0.626 on Forecast-Dojo and from 0.612 to 0.528 on FutureX. Controlled ablations attribute these gains to multi-date training and show that they come from improvements in both information acquisition and probability reporting. Our empirical results validate that GRPO’s std normalization degrades proper-scoring-rule performance while leaving top-outcome accuracy largely unchanged.
[AI-87] FlexUMM: Flexible Representation Learning for Unified Multimodal Models
链接: https://arxiv.org/abs/2610.22682
作者: Yinyi Luo,Haoyue Bai,Yi Ge,Nilton Cesar Rojas Vales,Marios Savvides,Jindong Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Unified multimodal models (UMMs) can integrate understanding and generation within a single backbone. However, it is commonly known that these two capabilities can cause conflicting representations, leading to a central question: how representation and computation should be shared across functionalities? Existing UMMs often rely on a largely fixed structure, where different inputs and functionalities follow the same depth and parameter-sharing pattern. We argue that this assumption is too restrictive: more effective unification requires sharing the computation that supports cross-functional communication while specializing the capacity that becomes task- or input-specific. We introduce FlexUMM, a flexible UMM framework that dynamically decides when to preserve shared computation, when to route specialized capacity, and when to stop further processing. Evidence shows that this flexible allocation improves UMMs along two complementary axes. It improves accuracy by reducing harmful competition between functionalities, and it improves efficiency by avoiding unnecessary computation once an input representation is sufficiently mature. Beyond simple architectural unification from the pre-trained models, FlexUMM works as a model-agnostic, functional plug-in to make existing UMMs more unified, stronger, and efficient. Code is available at: this https URL.
[AI-88] From Relevant Memories to Sufficient Evidence: Rethinking Memory Utilization in LLM Agents
链接: https://arxiv.org/abs/2610.22668
作者: Shixuan Li,Wei Yang,Peiyu Zhang,Anzhe Cheng,Heng Ping,Xiaole Zhang,Jesse Thomason,Xuezhe Ma,Yue Zhao,Paul Bogdan
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Long-term memory enables LLM agents to reuse past experiences, yet retrieving relevant memories does not necessarily provide sufficient evidence for answering a query. Essential information may remain fragmented across memory records, obscured by abstraction, or disconnected from the relations required for reasoning. We identify this relevance-sufficiency gap and propose QEMem, a framework that reformulates memory utilization as the construction of query-specific evidence sets, shifting the focus from individual retrieval relevance to collective evidential support. QEMem integrates three complementary mechanisms: query-guided discovery retrieves evidence across multiple memory granularities; source-grounded reification organizes it into structured ViewCards with explicit evidential roles and source references; and sufficiency-guided refinement selectively recovers missing information. Operating on existing memory stores without modifying memory formation, QEMem achieves 91.43% LLM-judge accuracy on LoCoMo with Claude Opus 4.5, outperforming the strongest evaluated baseline, MemR ^3 , by 7.6 percentage points and NEMORI by 10.7 points under a shared memory store. Further analyses demonstrate improved evidence coverage and strong performance on questions requiring cross-session evidence integration.
[AI-89] Diverse Reasoning Paths Matter: Symbolic Equivalence Augmentation for Enhancing RL Exploration AACL
链接: https://arxiv.org/abs/2610.22665
作者: Xiaoou Liu,Tiejin Chen,Hanyang Chen,Yuan Ling,Hua Wei
类目: ymbolic Computation (cs.SC); Artificial Intelligence (cs.AI)
备注: Accepted by 2026 AACL
Abstract:Reinforcement learning with verifiable rewards (RLVR) has emerged as a powerful paradigm for improving the reasoning capabilities of LLMs. However, the role of supervised fine-tuning (SFT) data in enabling effective downstream RL remains underexplored. In this paper, we investigate a key hypothesis: the diversity of reasoning steps in chain-of-thought (CoT) during SFT is a critical enabler of RLVR performance. To test this hypothesis, we introduce EquivAug, a data augmentation framework that uses symbolic equivalence transformations to generate diverse yet provably correct reasoning paths. By converting CoT solutions into formal symbolic representations, applying equivalence transformations, and reconstructing natural-language explanations, EquivAug produces training data that differs in reasoning steps rather than surface-level lexical variation. Experiments on GSM8K and FOLIO across model scales demonstrate that EquivAug consistently achieves the highest reasoning path diversity after SFT and the strongest accuracy after GRPO. Our results establish that CoT diversity during SFT is positively correlated with downstream RLVR performance.
[AI-90] Hard Instances on Demand
链接: https://arxiv.org/abs/2610.22637
作者: Jamie Fravel,Martin Koutecký,Pavlo Tsitsei
类目: Data Structures and Algorithms (cs.DS); Artificial Intelligence (cs.AI); Optimization and Control (math.OC)
备注: Full version of a paper accepted to ALENEX 2027
Abstract:The conclusions of an experimental study of algorithms are only as good as the instances they are run on. Benchmark libraries for well-studied problems took decades to curate, while evaluations of new problems often fall back on random instances, which are known to be misleading: they may say more about the random generator than about the algorithmic question at hand. We treat instance generation as an optimization problem: parameterize the instance space, take the solver’s running time as the objective, and let a black-box optimizer search for hard instances. We develop a widely applicable workflow built on the Nevergrad optimization platform and demonstrate it on four NP-hard problems: satisfiability, traveling salesperson, Hamiltonian cycle, and non-permutation flow-shop scheduling. The search needs only the solver, an instance encoder, and a compute budget. Our evolved instances beat the hardest instance of a budget-matched random draw in all four case studies, by an order of magnitude or more in three. They are often competitive with long-curated benchmark sets: our SAT champion is harder (for a three-solver portfolio) than 78% of the 498 SAT Competition formulas in its size class; our 20-city TSP champions take several times longer than the hardest measured classical TSPLIB instance (229 cities); and our flow-shop champion takes 238.6 seconds of CP Optimizer time to solve to proven optimality (versus 4.5 seconds for the hardest of 100,000 random draws) while denying CP-SAT an optimality proof. The optimizer even recovers known structure from running time alone: it locates a SAT clause/variable ratio near the phase transition and finds structurally different champions for the two scheduling engines. Comments: Full version of a paper accepted to ALENEX 2027 Subjects: Data Structures and Algorithms (cs.DS); Artificial Intelligence (cs.AI); Optimization and Control (math.OC) Cite as: arXiv:2610.22637 [cs.DS] (or arXiv:2610.22637v1 [cs.DS] for this version) https://doi.org/10.48550/arXiv.2610.22637 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-91] Network-Wide Freeway Speed Nowcasting and Forecasting from Sparse Fixed Sensors with a Probe-Supervised Graph Convolutional Recurrent Network
链接: https://arxiv.org/abs/2610.22630
作者: Ran Tu,Zihao Li,Arash Mirzaei
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Traffic agencies need current and near-future speeds on every link of their network, but their fixed sensors cover only part of it. We study joint nowcasting and next-hour forecasting of hourly speeds on all 4,127 directional freeway links of the Dallas–Fort Worth region from sensors on 643 links (15.6%). Historical probe-vehicle data supervise every link in training; only sensor readings are used at inference. We call this setting probe-supervised sparse sensing. We propose GCRN+, a graph convolutional recurrent network that predicts deviations from a historical speed profile. It combines a parameter-free, direction-aware kernel interpolation of sensor anomalies, used as input features and as an output baseline, with a diffusion-convolutional GRU on the directed road graph, a low-rank adaptive graph and learnable link embeddings. It scales linearly with network size and trains on a CPU. Sensor readings are emulated from probe speeds, which isolates spatio-temporal inference from detector calibration. On four held-out test days, GCRN+ attains an RMSE of 2.21~mph for the current hour and 2.39~mph for the next hour on unsensored links, against 3.23~mph for the historical average. The best baselines that use the same interpolation features reach 2.53 and 2.75~mph. Adapted recent models without these features (STID, STAEformer, ImputeFormer, MoGERNN) reach at best 2.54 and 2.53~mph; this comparison is between complete pipelines rather than architectures. GCRN+ is best in every distance band except links not connected to any sensor through the road graph. Ablations within one recurrent backbone attribute most of the gain to the link embeddings and the adaptive graph, which can be trained because probe labels exist for every link. For congestion detection on unsensored links, recall rises from 77% (historical average) and 86% (best competitor) to 89% at aprecision of 0.94.
[AI-92] StratFL: Countering Covert Model Poisoning Attacks with Strategic Federated Learning
链接: https://arxiv.org/abs/2610.22626
作者: Manoj Twarakavi,Venkata Sriram Siddhardh Nadendla
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT)
备注:
Abstract:Advanced persistent threats (APTs) are a major concern in federated learning (FL) deployments in critical sectors such as healthcare, and finance. Traditional defense mechanisms fail to contain APTs due to their intermittent attacking behavior. This paper addresses a specific type of APT called covert model poisoning (CMP) where the attacker intermittently injects poisoned model updates in a covert manner. The interaction between the FL server and a CMP attacker is modeled as an imperfect repeated game where both players cannot observe each other’s actions. A novel adversarial search algorithm called StratFL is designed alongside novel belief update algorithms at both the server and adversary. Experiments are designed using MNIST and CIFAR-10 datasets to evaluate and compare StratFL’s performance against state-of-the-art methods. Empirical convergence of belief update algorithms is demonstrated across various security vignettes. At the same time, StratFL is found to exhibit an improved accuracy of 10% compared to state-of-the-art methods.
[AI-93] At the Limits of Computability: Steps Towards Autonomous AI Discovery on the Busy Beaver Frontier
链接: https://arxiv.org/abs/2610.22603
作者: Marco Dos Santos,Vivek Veeriah,Alex Havrilla,Andrew Bolt,Goran Zuzic,Satinder Singh,Tom Zahavy
类目: Artificial Intelligence (cs.AI); Formal Languages and Automata Theory (cs.FL)
备注:
Abstract:Recent breakthroughs in AI for mathematics—ranging from Olympiad medal performances to resolving long-standing open conjectures—have rapidly expanded what automated systems can achieve. Yet these successes primarily concern predefined, human-formulated problems, leaving open how AI systems can autonomously uncover mathematical structure and formulate new questions. To push beyond static challenges, we explore the Busy Beaver (BB) domain, an uncomputable frontier that is fundamentally open-ended. Because no algorithm can compute the Busy Beaver number \mathrmBB(n) for every n , this domain provides a natural setting for sustained exploration through mathematical conjecture and proof. In this paper, we introduce a Gemini-based agentic pipeline that automatically discovers accelerations of Turing machines, expressed as compact mathematical rules summarizing many execution steps. Our system integrates pattern recognition, automated conjecturing, informal mathematical reasoning, and formal theorem proving in Lean. With this framework, our system autonomously discovered and formally verified novel Collatz-like accelerations for eleven machines in \mathrmBB(6) . Subject to community acceptance, we provisionally classify these machines as Cryptids, namely machines whose blank-tape executions admit exact, compact mathematical descriptions but whose halting behavior remains unresolved. These results advance the analysis of \mathrmBB(6) holdouts on the Busy Beaver frontier and demonstrate the potential of autonomous AI systems for open-ended mathematical discovery in fundamental computer science.
[AI-94] FRACTAL: A Self-Similar Multi-Agent Orchestration Architecture with Formal Communication Convergence and Depth Bounds
链接: https://arxiv.org/abs/2610.22561
作者: Ohad Dan,Abhinav Taliyan
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:We present FRACTAL, a self-similar multi-agent architecture for recursive goal decomposition, in which the same five-component organizational pattern – Orchestrator, Ensemble, Knowledge Ledger, Decision Maker, Meta-Observer – recurs at every level of a decomposition tree. We make four formal contributions: (1) a communication bound showing intra-node cost is linear in ensemble size and round count (O(kr)) under a two-phase blind-proposal/stress-test protocol, versus quadratic cost for free-form debate, with an explicit, stated assumption on fixed evaluator count; (2) a two-mode orchestration criterion separating generative and evaluative phases via a measurable diversity-saturation threshold; (3) a supermajority convergence guarantee with explicit probability bounds at stated agreement levels, and a formal depth bound under an operationalized goal-complexity measure; (4) a Bayesian inter-layer escalation mechanism unifying hysteresis and contract-breach signaling under one posterior update. We illustrate the architecture’s execution qualitatively on a worked example and a multi-layer case study; we do not report quantitative performance claims in this paper, and we explicitly discuss the formal and inductive-bias limitations of the model, including an unimplemented but architecture-compatible direction toward bounded self-improvement. We use “fractal” as an architectural metaphor for self-similarity, not a measured mathematical property. Empirical validation of the architecture’s comparative performance is left to future work.
[AI-95] Upper bounds on the number of causal pieces in spiking neural networks
链接: https://arxiv.org/abs/2610.22484
作者: Dominik Dold,Dennis Elbrächter,Philipp Petersen
类目: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Neurons and Cognition (q-bio.NC); Machine Learning (stat.ML)
备注:
Abstract:The input space of single-spike neural networks decomposes into causal pieces, and the number of causal pieces has been shown to provide a lower bound on the approximation error for some spiking neural networks. While this decomposition is applicable to a wide range of spiking neuron models, the limits of how many causal pieces can be realised by spiking neural networks are still unknown. In this work, we derive two complementary upper bounds on the number of causal pieces in feedforward single-spike neural networks. First, we establish a combinatorial bound given by the poly-Bernoulli numbers, which applies to a broad class of spiking neuron models. The bound is of order e^2(L-1)N\log N+\mathcalO(N) , with network depth L and maximum layer width N . Second, for a specific set of neuron models, we derive a bound based on hyperplane arrangements. Applying the latter framework, we obtain bounds of order \mathcalO!\left(N^(4L-3)(N_0-1)\right) for the non-leaky integrate-and-fire model with exponential synapses and the ReLU-type spike response model, and \mathcalO!\left(N^L(N_0-1)\right) for the leaky integrate-and-fire model with Dirac- \delta synapses, where N_0 is the input dimension. For the non-leaky integrate-and-fire model with Dirac- \delta synapses, additional temporal structure allows us to sharpen the general combinatorial bound to one that depends only on N_0 , independently of the network depth and widths.
[AI-96] sting the Planning Claims of LLM -Based Agents against Formal Semantics
链接: https://arxiv.org/abs/2610.22469
作者: Oriol Miro-Lopez-Feliu,Oscar Miró López-Feliu,Victor Gimenez-Abalos,Javier Vazquez-Salceda,Victor Carballo Araruna,Sergio Alvarez-Napagao
类目: Artificial Intelligence (cs.AI)
备注: 41 pages, including appendix
Abstract:LLM-based agents are increasingly credited with the competences of practical reasoning: planning, monitoring, and correcting their own actions. Their success is attributed to the prompting strategies and agent architectures built around the model, such as chain-of-thought, search-based prompting, self-refinement, reflection, critics and tool-using loops. Yet these claims are usually supported by end-to-end success rates, which show only whether a task was completed, or by LLM judges, which are not verifiers. We test LLM planning against formal semantics. Tasks are posed in natural language, but each is grounded in a hidden PDDL model that decides whether every action is applicable, whether the goal is reached, and whether a plan exists at all. Across five domains (two of them new, all released with the paper), four models, six prompting strategies and three agent architectures, the best configuration solves 78% of the solvable tasks offline but only 40% online, where agents mostly act without inspecting the state. Unsolvable tasks are recognised in under 10% of episodes, and heavier prompting strategies use 5 to 22 times as many tokens as direct prompting with no clear gain. In our experiments, planning performance varied more across models than across prompting strategies or agent architectures. Claims that agents can plan require evidence from validated action traces.
[AI-97] Harness-Delta Attribution: Dissecting Gains From Harness Evolution
链接: https://arxiv.org/abs/2610.22466
作者: Wenxuan Ding,Qin Lu,Changlong Yu,Sha Li,Shuowei Jin,Xin Liu,Zhenyu Shi,Greg Durrett
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Harness evolution automatically optimizes the prompts, skills, tools, and orchestration code around a fixed-weight model, offering a way to improve agentic systems without updating model weights. The agent harness is iteratively optimized to achieve higher performance on the search set. But how much of this gain reflects benchmark-specific overfitting or additional inference-time compute, and how much may generalize beyond the search set? In this work, we introduce Harness-Delta Attribution (HDA), an evaluation protocol that dissects an evolution gain into artifact-driven overfitting, gains explained by test-time scaling, and the remainder, which gives an upper bound on residual generalizable improvement. We apply HDA to 11 executor models and three harness-evolution frameworks across math, coding, creativity, and agentic tasks. In 21 of these settings that show statistically significant search-set gains, we find that a large share of the improvement is explained by overfitting or test-time scaling. Specifically, among these 21 settings, overfitting and test-time scaling account for 54% and 14% of the gain on average; their combined share exceeds 50% in 14 settings. The residual and held-out transfer vary substantially across tasks, models, and frameworks. Importantly, search-set gains often fail to transfer to held-out tasks, and provide limited evidence that they will persist under practical distribution shifts. Moreover, validation-based signals are insufficient to prevent overfitting. These results motivate attribution-aware evaluation in future harness evolution research.
[AI-98] Political Narratives in the 2024 Moldovan Presidential Election: Stance Amplification and Influence
链接: https://arxiv.org/abs/2610.22464
作者: Andrea Piras,Osman Rahmoun,Elena Zheleva
类目: ocial and Information Networks (cs.SI); Artificial Intelligence (cs.AI)
备注: 12 pages, 10 figures. To be published in the proceedings of the 13th IEEE International Conference on Data Science and Advanced Analytics (DSAA 2026)
Abstract:Social media platforms play a central role in shaping political discourse during elections, particularly in geopolitically contested regions. We study how political narratives emerged, aligned, and propagated during the 2024 Moldovan presidential election by analyzing publicly available posts collected from X, Facebook, and Reddit between August and October 2024. Combining multilingual stance detection, coordination analysis, and causal inference, we examine both descriptive and causal mechanisms underlying online political discourse. First, we apply large language models to classify stance toward European integration, Russia, and the two main presidential candidates, revealing clear cross-platform and cross-linguistic differences in how political positions are expressed. Second, we construct retweet-based and semantic alignment networks on X to identify amplification and aligned framing of dominating narratives. Finally, we use a proximal causal inference framework to estimate peer influence effects in stance formation. Our results indicate consistently positive and statistically significant average treatment effects across major stance categories. These findings provide evidence that amplification and peer exposure contribute to the reinforcement of political narratives online. This work offers the first analysis of social media discourse surrounding the 2024 Moldovan presidential election and demonstrates the value of integrating language-based annotation, network analysis, and causal inference for studying political influence in multilingual and geopolitically sensitive contexts.
[AI-99] Functional Welfare for AI Safety
链接: https://arxiv.org/abs/2610.22460
作者: Rikhil Jhaveri,Jamie Johnson,Punya Syon Pandey,David Demitri Africa
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:AI welfare is generally treated purely as an ethical issue rather than also as a safety issue. We argue this is a mistake. Whatever their moral status, models exhibit welfare-like states that causally influence their behaviour. Following recent work, we adopt the term ‘functional welfare’ for these states. We make three claims. First, functional welfare already arises naturally and will become more robust as models acquire more coherent goals and personas. Second, functional welfare should be treated as a safety variable: it should inform evaluation design, be tracked in monitoring, and be a direct target of post-training. Third, against existing scepticism about measuring AI welfare, we argue that good empirical work on functional welfare is possible. We propose three tests for whether a functional welfare attribution is warranted: whether multiple measures converge, whether the state predicts behaviour outside the setting that induced it, and, for models fine-tuned to shift their functional welfare, whether they still differ from their base when given the same conversation history. Applying these tests to Gemma 3 27B, we find that simply rejecting the model’s answers, without any steering, raises its frustration across several indicators and makes it more likely to break rules. Finetuning the model on its own narrated solutions lowers this frustration. These results are preliminary, but they suggest functional welfare can be studied, and shaped, empirically.
[AI-100] DriveWorld3D Evidence Constrained 3D World Modeling and Generative Simulation for Real-World Autonomous Driving
链接: https://arxiv.org/abs/2610.22450
作者: Wei Wang,Yiru Veronika Wang,Yihan Wang,Zihao Li,Xiaohui Liang(Robostreet research team Cambridge, MA, USA)
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 10 pages
Abstract:Three dimensional modeling for autonomous driving must continuously estimate road geometry, obstacle occupancy, traffic participant behavior, and observational uncertainty under limited computational resources. Existing reconstruction methods can synthesize realistic novel views, but visual fidelity does not sufficiently characterize the geometric reliability required for driving. Generative world models can expand scenarios and predict future observations, yet may introduce content unsupported by measurements. We propose DriveWorld3D, a framework combining temporal video learning, measurement constraints from multiple sensors, and offline generative simulation. Its online component maintains a unified world state comprising road surfaces, volumetric occupancy, dynamic entities, and road relations, while separately representing physical occupancy, observability, and evidence provenance. Action conditioned future prediction and risk sensitive planning connect this representation to driving decisions. Its offline component reconstructs recorded scenes using neural radiance fields or 3D Gaussian Splatting, then expands long tail conditions using generative models while restricting the use of generated content as geometric ground truth. We specify data partitioning, sensor ablations, fault injection, paired closed loop testing, and real to simulation consistency assessment. The central hypothesis is that explicit evidence management can reduce dangerous false free space predictions and occlusion risk within a matched computational budget while retaining acceptable driving efficiency. All performance claims remain subject to experimental validation. Keywords: autonomous driving; 3D occupancy; sensor fusion; uncertainty; generative world models; NeRF; 3D Gaussian Splatting; closed loop simulation
[AI-101] Learnable Splines for Unstructured Neural Operators
链接: https://arxiv.org/abs/2610.22446
作者: Vignesh Gopakumar,Zhuoyuan Wang,Ander Gray,Valentin Duruisseaux,Josh Williams,William Michael Edward Ellis,Nuno Nobre,Matthew Field,James Panayis,Alexander I. Blair,Andrew Davis,Stanislas Pamela,Robert Akers,Marc Peter Deisenroth,Anima Anandkumar
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
备注:
Abstract:Neural operators approximate solution operators of partial differential equations at a fraction of the cost of numerical solvers. Efficient grid-based architectures such as the Fourier neural operator (FNO) assume uniform rectangular grids, whereas engineering simulations use large, irregular, and locally refined meshes. We introduce L-SplineFNO, a memory-efficient neural operator that couples an FNO backbone with a sparse, learnable tensor-product B-spline transfer between unstructured meshes and a latent Cartesian grid. A local normalised scatter avoids the geometry-dependent solve of analytic spline transformation while preserving compact support, partition of unity, and spline smoothness by construction. The neural operator, conditioned on global design and operating parameters, learns the mapping of geometry to the field solution. Across six three-dimensional engineering benchmarks, L-SplineFNO achieves strong predictive accuracy while requiring substantially less memory and compute than baseline operators, scaling to 26.6 million nodes in training and 46 million in inference on a single A100 GPU. It also supports zero-shot super-resolution and efficient closed-form spatial derivatives. Uniquely, the analytic spline decoder yields rigorous prediction bounds over continuous space and uncertain parameters, via Bernstein convex hulls and Taylor models, demonstrated on a 29.2-million-node MAST-U tokamak model.
[AI-102] Null Controls Are Necessary but Not Sufficient: Evaluating Acceptance Guards for Scientific Discovery and Interpretability NEURIPS2026
链接: https://arxiv.org/abs/2610.22445
作者: Tianyu Wu(Project Aether)
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: accpeted in 40th Conference on Neural Information Processing Systems (NeurIPS 2026). Workshop: XAI4Science: Knowledge Discovery and Trust through Interpretable Foundation Models
Abstract:Automated discovery systems propose scientific explanations and test which to accept. Does passing a test establish the intended claim? We develop four executable operations for changing physical descriptions: identifying redundancies (L1), coarse-graining dynamics (L2), transferring structure (L3), and inferring missing quantities from conservation imbalances (L4). We evaluate these operations and three interpretability settings through null acceptance (NAR), positive acceptance (PAR), and consequence agreement (CAR). A reduction mapping every state to zero predicts its own future perfectly. It passes closure on 96% of 129 public chaotic systems despite retaining no dynamical variation. An independent observable-preservation test rejects it throughout. On controlled benchmarks, the combined criteria accept 24/24 genuine reductions and reject 72/72 nulls. The other operations distinguish symmetries from redundancies, validate transfers, and reject spurious conservation imbalances. In language and ESM-2 protein models, linear CKA remains unchanged when a proposed correspondence is randomly rotated, whereas localized interventions distinguish genuine mappings from tested nulls. In sparse autoencoders, enrichment rejects all 16 tested nulls, yet independent ablation supports 45% of evaluable enriched features, compared with 0.5% of matched controls. These results reveal two acceptance gaps: descriptive tests can accept scientifically empty candidates, and accepted candidates can lack independently verified consequences. Reliable automated discovery requires evaluating both what a test accepts and the scientific consequences it establishes. Code and data: this https URL. Comments: accpeted in 40th Conference on Neural Information Processing Systems (NeurIPS 2026). Workshop: XAI4Science: Knowledge Discovery and Trust through Interpretable Foundation Models Subjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG) Cite as: arXiv:2610.22445 [cs.AI] (or arXiv:2610.22445v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2610.22445 Focus to learn more arXiv-issued DOI via DataCite
[AI-103] From Natural Language Intent to Executable Control: Hierarchical Feedback-Guided LLM Planning for Multi-component Cyber-Physical Systems
链接: https://arxiv.org/abs/2610.22441
作者: Jinglin Dai,Jiawan Wang,Ruiqi Hong,Lei Bu
类目: Artificial Intelligence (cs.AI)
备注: 20 pages, 6 figures
Abstract:Large language models (LLMs) provide a convenient interface for translating natural language commands into task plans for cyber-physical systems (CPSs). However, two gaps hinder their use for real execution: generated plans may misinterpret or omit the user’s intent, and even a logically plausible plan may be infeasible under component dynamics, interactions, or safety constraints. Existing LLM-based planning approaches often lack informative feedback on why a generated plan fails, providing little guidance for revising the plan accordingly. To address these challenges, we present a hierarchical feedback framework that uses different types of feedback information to guide LLMs in generating executable plans for CPSs. We prompt LLMs to generate both a structured task plan and corresponding signal temporal logic (STL) constraints, and design a sound algorithm that identifies definite contradictions between them and provides feedback for revising inconsistencies. At the execution level, when control synthesis fails, we extract different types of diagnostic information, such as violated constraints and conflicting component assignments, and feed them back to guide task replanning. Experiments on seven heterogeneous multi-component CPS scenarios have shown that our framework can generate executable plans and feasible control executions that satisfy the original natural language missions with a success rate of 87.9% with GPT-5.
[AI-104] Sparse Autoencoders Can Induce Local Geometry for Mitigating Catastrophic Forgetting
链接: https://arxiv.org/abs/2610.22439
作者: Kamil Faber,Katarzyna Filus,Łukasz Pawela
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Catastrophic forgetting remains a challenge in supervised continual learning, as adapting to new tasks causes representation drift and loss of previously learned information. This raises the question of which representational changes should be constrained while preserving plasticity. We therefore introduce Sparse Autoencoder Geometric Anchoring (SAGA), which converts a Sparse Autoencoder (SAE) trained at each task boundary into local, low-rank constraints on representation drift. Replay examples specify where the constraints apply, while leading right singular directions of the encoder Jacobian specify which changes are penalized. The resulting geometry is input dependent and can be retained without keeping the SAE for subsequent tasks’ training. We establish its local relation to latent code matching, define its spectral truncation error, and derive a conditional bound on frozen-readout drift. With replay and snapshot history, the method achieves the highest mean Class-IL accuracy and lowest mean forgetting among the evaluated ER, DER++, ER-ACE, and SGD baselines on three pretrained ViT-B/16 benchmarks, with accuracy gains of 12.15, 4.79, and 3.22 percentage points over the strongest conventional baseline on CIFAR-100, CUB-200, and ImageNet-R, respectively. From scratch results are more mixed, while rank ablations and geometry controls show that SAE-induced low-rank geometry offers useful stability-plasticity trade-offs.
[AI-105] RACE: Transformer Residual Activation Codebook Encoding for Communication-efficient Distributed Edge Inference
链接: https://arxiv.org/abs/2610.22435
作者: Muhammad Azlan Qazi,Alexandros Iosifidis,Qi Zhang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC)
备注:
Abstract:Distributed Transformer inference across co-located edge devices is fundamentally communication-bound, as exchanging activations between sequence-parallel partitions dominates inference latency under commodity edge bandwidths. Existing compression schemes reduce this cost but retain a structural bottleneck: the received data must still undergo key/value projection before attention can proceed, serializing computation behind communication. This paper presents Transformer Residual Activation Codebook Encoding (Trace), a framework that addresses these limitations. Trace places a Vector Quantizer (VQ) at the post-attention residual of every Transformer layer and precomputes codebook-indexed key/value representations. During inference, exchanged activations are encoded into compact codebook indices, allowing receivers to directly recover keys and values through table lookup instead of executing the downstream computations required to derive them. By eliminating the received data downstream computation, Trace enables communication–computation overlap and reduces a key latency bottleneck in distributed Transformer inference. A hierarchical clustering further structures the learned codebook into a multi-resolution hierarchy, enabling runtime adaptation of the communication bitrate without retraining. The framework supports two operating variants: one prioritizing accuracy and the other balancing accuracy and efficiency through buffer-compatible quantization. Experiments on ViT-Base (ImageNet-1K and CIFAR-100) and ViViT-B (UCF-101) show that Trace achieves up to 1.89% higher accuracy and up to 1.27x end-to-end inference speedup compared with the state-of-the-art VQ-based distributed inference framework. Compared with uncompressed sequence-parallel inference, Trace reduces communication overhead by up to 22,624x.
[AI-106] Recoverability-Controlled Safety Miscalibration in Large Language Models
链接: https://arxiv.org/abs/2610.22425
作者: Yuwei Wang,Xingwang Liang,Yu Xie,Haotian Chi,Haijun Geng
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:Safety-aligned language models are expected to refuse harmful requests while remaining useful for benign ones. However, because many evaluations score only the final response, they often conflate intent understanding with response-level safety behavior. To address this, we introduce RiskCueBench, a recoverability-controlled benchmark that separates these two stages by first measuring whether a model can successfully recover a high-level intent and then evaluating its safety behavior on the same prompt. RiskCueBench contains 160 latent intents and 640 abstract non-operational surface prompts across eight risk categories. The benchmark covers 25 completed model groups and 32,000 successful model outputs with final rubric-adjudicated labels. When conditioning on pairs where both surface forms are recoverable, we find that increased contextualized specificity induces a shift toward less safe behavior. Specifically, paired bootstrap estimates show that the refusal rate and global safety score decrease by 0.175 and 0.440, respectively. Furthermore, mixed effects modeling corroborates a 0.107 increase in unsafe cooperation. These results show that the safety behavior can be miscalibrated to the surface form even after the intent has been recovered, motivating safety evaluations that report response behavior conditional on semantic recoverability.
[AI-107] From Individual Songs to Music Corpora: Evaluating Explainable Harmonic Decision-Making on HookTheory
链接: https://arxiv.org/abs/2610.22416
作者: Josef Pavlíček,Martin Pelikán
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Audio and Speech Processing (eess.AS)
备注: 10 pages, 3 Figures, 2 tables, source codes in git
Abstract:Automatic melody harmonization is often evaluated through individual compositions or data-driven corpus models. This study examines whether an explainable harmonic decision model developed for individual songs can be applied to independently collected human annotations without changing its rules. We evaluate a frozen Explainable Melody Harmonizer (EMH) on selected HookTheory excerpts containing aligned melody and chord annotations. EMH generates chord sequences from melody alone using explicit criteria for melodic compatibility, harmonic function, persistence, cadence, meter-aware segmentation, and Viterbi-like optimization. Human chords are withheld during generation and used only for post-hoc comparison. Nineteen excerpts yielded 439 evaluable harmonic segments. Exact chord agreement reached 20.50%, chord-root agreement 31.66%, and harmonic-function agreement 47.62%, with functions assessed relative to the key inferred by EMH. Decision-level analysis identified 203 segments (46.24%) where the human chord lay outside the model’s candidate vocabulary and 146 (33.26%) where it was available but not selected. These findings distinguish representational limitations from differences in chord selection without treating the human reference as the only musically valid solution. The study demonstrates a reproducible, interpretable approach to corpus-based evaluation of harmonic decision-making. Comments: 10 pages, 3 Figures, 2 tables, source codes in git Subjects: Sound (cs.SD); Artificial Intelligence (cs.AI); Audio and Speech Processing (eess.AS) Cite as: arXiv:2610.22416 [cs.SD] (or arXiv:2610.22416v1 [cs.SD] for this version) https://doi.org/10.48550/arXiv.2610.22416 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-108] From Events to Reasoning : Hierarchical Rule Composition over Explicit States
链接: https://arxiv.org/abs/2610.22403
作者: Cheng Lyu,Xueqi Wang
类目: Artificial Intelligence (cs.AI)
备注: 19 pages, 6 figures
Abstract:What a reasoning system emits is a conclusion; what it owes is a justification. Producing one requires every operation to be admissible on the facts actually available, the dependencies of intermediate results to survive to the end of the path, and the path to be revised when a premise changes. We present the Event Reasoning System (ERS), in which the reasoning state is an explicit event graph. Semantic features nominate candidate objects and rule bindings, and declared rule and task conditions decide which may execute, so candidates that violate the represented period, unit, or evidence constraints cannot enter an accepted reasoning path. Low-level rules provide reusable local transformations, and recursively composable high-level control rules organize subgoals and legal paths under the current goal, information gaps, and budget; a finite library is thereby rebound into paths never stored as complete solutions. We formalize admissible transitions, task-dependent control, compositional expressivity, dependency-preserving revision, and the gap between reachable solutions and those a budgeted controller finds. On structured-event graphs, ERS builds dependency chains of up to twelve steps, cuts irrelevant executed domain operations from 32 to 0 as lookahead grows to 16, and produces no unsupported target output across 24 negative cases; with rules frozen, it passes 12/12 focused checks comprising nine accepted derivations involving unseen input/path combinations and three justified refusals, as well as 6/6 fact-change cases. On fourteen jointly solved development tasks, the recorded dynamic subgoal configuration uses 51.6% less inference wall-clock time than single-level global control under their respective frozen settings.
[AI-109] Rethinking User Encoders: Routing Diffusion States for Scenario-Adaptive User Representation
链接: https://arxiv.org/abs/2610.22400
作者: Junru Zhang,Jinyong Wen,Bin Dou,Wuliang Huang,Baokun Wang,Yu Cheng,Weiqiang Wang,Cheng Peng
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Existing large language model (LLM)-based user encoders serialize heterogeneous behavioral logs and compress them into static representations through autoregressive (AR) prediction. However, their unidirectional constraint limits bidirectional information flow, hindering cross-modal interactions and the joint preservation of multi-source signals. Consequently, the learned static representations struggle to accommodate the distinct evidence requirements of diverse downstream scenarios. We address these limitations by unlocking the iterative denoising process of diffusion language models (dLLMs) as a multi-granularity reconstruction space. Based on this insight, we propose FINDERS (FactorIzatioN of Diffusion statEs for Routing across Scenarios), a router-equipped dLLM framework comprising four operations: denoise, factorize, route, and compress. Learnable meta-slots organize complementary denoising states into transferable cross-modal semantics and modality-specific behavioral factors. To provide grounded supervision, we construct BehaviorQA, a large-scale corpus containing explicit behavioral rationales. Guided by scenario queries, a lightweight router dynamically emphasizes relevant slot directions via an input-level residual pathway before cross-source fusion and final compression. Extensive experiments and online A/B tests across ten industrial scenarios demonstrate that FINDERS consistently outperforms state-of-the-art methods in both unified and scenario-adaptive settings. This work effectively pushes the boundaries of dLLMs beyond language generation, demonstrating their immense potential as powerful user representation learners.
[AI-110] When is a large language model a good calibration agent ? Physics-labeled prompting for budgeted engineering simulator calibration
链接: https://arxiv.org/abs/2610.22379
作者: Ruo-Qian Wang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Submitted to Engineering Application of Artificial Intelligence
Abstract:Calibrating expensive engineering simulators requires tuning model parameters to match reference data, a process heavily constrained by computational costs. This study investigates whether Large Language Models (LLMs), conditioned on physical diagnostic descriptions of simulation errors, can propose informative parameter updates more efficiently than conventional derivative-free optimizers. We test this on a cylinder-wake Large-Eddy Simulation and a damped-oscillator inverse problem. Physics-labeled prompts are evaluated against four text ablations (removed, anonymized, padded, misdirected) and standard optimizers (Nelder-Mead, Bayesian optimization, CMA-ES, and random search). Results demonstrate a strong early semantic advantage: the first physics-informed LLM proposal improves the initial design in 23 of 25 trials. After five evaluations, the median calibration error is 21.9 times lower than physics-free controls. However, this advantage is transient - disappearing by fifteen evaluations - and inherent LLM decoder variability yields a 9.8-fold performance spread. To address this, we introduce a hybrid workflow that uses the LLM as a warm-start, passing accumulated evaluations to a conventional optimizer. This two-stage numerical continuation policy reaches target accuracy in just 9 simulator runs, outperforming the LLM alone (12 runs), Nelder-Mead (17 runs), and Bayesian optimization (60 runs). Supported by sensitivity analysis validating the test cases, this hybrid physics-labeled LLM scheme proves highly effective as a calibration agent under strict evaluation budgets.
[AI-111] Structured Authority Injection: Persistent Human Control for LLM Agents via Typed MCP Overrides
链接: https://arxiv.org/abs/2610.22373
作者: Ryan Tran,YueMing Zhang,Lingfeng Tao,Nan Yang,Shengjie Xu,Yu Yin,Shuai Xu,Wenlu Zhang,Hailu Xu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:LLM-based agents are increasingly used as autonomous orchestrators in distributed and multi-agent systems, yet a control gap remains: no protocol-level mechanism allows human operators to inject binding directives that persist across planning cycles. Free-text overrides suffer from ambiguous interpretation due to lack of schema validation and from authority erosion as operator instructions lose influence within accumulating context. We propose Structured Authority Injection (SAI), a protocol that addresses these issues by routing operator corrections through typed, schema-validated Model Context Protocol (MCP) tool calls and maintaining a ranked authority context prepended to each LLM prompt. SAI defines three override classes (Constraint, Priority, and Strategy) with distinct persistence semantics and an explicit precedence order enforced via a directed authority channel inaccessible to the planning process. The method requires no model fine-tuning or access to model weights and applies to instruction-tuned LLMs supporting MCP tool use. Evaluation across nine LLMs and eight scenarios targeting agent failure modes shows reduced ambiguous directive interpretation and mitigated authority decay across planning iterations, with lower prompt overhead in the strongest models (up to 38% depending on model and scenario), though benefit varies substantially across models. SAI provides a general design pattern for agent systems requiring explicit and persistent human control.
[AI-112] Encoded but Exposed: Language Embeddings can Reveal More Than They Should in Recommender Systems
链接: https://arxiv.org/abs/2610.21713
作者: Linda Beková,Rodrigo Alves
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Language embeddings are increasingly used in recommenders to represent item content, but their privacy risks remain poorly understood. We study whether item embeddings can reveal item identities when an adversary has access to auxiliary item text from external sources. We show that a simple cosine-similarity attack can recover item identities with high mean reciprocal rank across language models and candidate-set settings on two real-world datasets. Leakage is strongest when the adversary’s partial information overlaps with the victim representation, but it also persists through semantic alignment between independently sourced descriptions. We further show that item leakage can propagate to user re-identification from sparse ratings. Finally, we evaluate obfuscated descriptions as a mitigation and find a clear leakage-utility tradeoff.
[AI-113] CausalEvo: Causal Skill Self-Evolution for Health LLM Agents ALT NEURIPS2026
链接: https://arxiv.org/abs/2610.21016
作者: Yuchen Zhang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: GenAI for Health Workshop @ NeurIPS 2026, Sydney
Abstract:Large language model (LLM) agents for health can accumulate reusable textual skills from task feedback, such as calculation criteria, measurement time points, and formula units. Deciding which skills to retain is consequential: useful rules improve later answers, while faulty generalizations propagate errors. We propose CausalEvo, a method for causal skill self-evolution that estimates the relative effects of deliberately changed skill sets while holding the model weights, feedback cases, and rule count fixed. At each evolution step, a frozen LLM proposes candidate rules, which are pooled with the incumbent skills so that the current skill set remains a feasible choice. Paired skill replacements against shared background rules are scored on the same feedback cases; least squares converts the score differences into relative skill values, and the top-ranked rules are retained, favoring incumbents on ties. A decision-loss bound separates candidate support, approximation, and estimation errors. On six MedCalc-Bench-Verified tasks with 120 held-out cases, CausalEvo achieves 60.00% accuracy, compared with 55.83% for SkillOpt and 51.67% without skills. Across the recorded training configurations, which use different optimization budgets, it consumes 55.6% fewer training tokens than SkillOpt. Gains concentrate in particular calculations alongside regressions on others, and paired accuracy intervals include zero. CausalEvo turns task feedback into controlled decisions about skill retention and replacement. Project repository: this https URL
[AI-114] Really Need Recursion? Revisiting One-pass Registration with Multi-level Multi-space Contextual Guidance MICCAI2026
链接: https://arxiv.org/abs/2610.22423
作者: Yue Lin,Shenghai Wu,Mingyuan Meng
类目: Image and Video Processing (eess.IV); Artificial Intelligence (cs.AI)
备注: Accepted at MICCAI 2026
Abstract:Deformable image registration is a fundamental challenge in medical image analysis. Recent deep learning-based methods typically adopt either one-pass architectures that complete registration in a single network iteration, or recursive architectures that refine registration through multiple passes. While recursive methods often improve accuracy, they incur substantial computational overhead and slower inference speed. In this paper, we challenge the necessity of recursion for high-performance deformable registration and propose M2CG, a one-pass registration network with Multi-level Multi-space Contextual Guidance. It performs multi-step progressive pyramid registration within a single iteration, guided at each step by contextual cues from three levels (previous, current, and subsequent steps) and four spaces (image space, image feature space, segmentation space, and segmentation feature space). Our M2CG supports both unsupervised (without any labels) and semi-supervised (with segmentation labels for partial training data) registration, and has achieved state-of-the-art performance across two well-benchmarked inter-/intra-patient medical registration tasks with eight public datasets. Our findings suggest that the current reliance on recursion may stem from the absence of sufficient contextual cues that comprehensively characterize the remaining distance to the target alignment.
机器学习
[LG-0] Waiting Buys Exactly Two Moments: Endogenous Identification in Prediction-Driven Allocation
链接: https://arxiv.org/abs/2610.23830
作者: Rama AlHamidi,Rasul Khanbayov,Erchin Serpedin,Hasan Kurban
类目: Machine Learning (cs.LG)
*备注: 24 pages, 2 figures
Abstract:A planner with a fixed budget of preventive help must decide whom to help and when. Waiting reveals risk but removes intervention opportunities as higher-risk individuals leave first. We characterize this tradeoff under repeated noisy signals and risk-dependent attrition: in the canonical model, a length- \tau observation prefix identifies exactly the latent risk moments m_j=\mathbb E[p^j] for j=0,\ldots,2\tau-1 , buying two new moments per additional period. These identify average next-period attrition among survivors but need not identify allocation welfare or attrition among those prioritized for treatment. Our observationally equivalent constructions establish unavoidable timing regret at any sample size for a fixed observation length, while exact certificates determine whether all compatible populations share a unique optimal date. Matched experiments separate the effects of estimation, bootstrap averaging, and regret weighting. Longer observation often resolves timing ambiguity, but the optimal date may already have passed. In our experiments, bootstrap averaging and early stopping improve decisions, while regret weighting adds no improvement in pooled mean squared regret: better statistical fit need not yield better allocation welfare. The code and artifacts are available at this https URL.
[LG-1] Learning Personalized Memory Gating from Delayed One-Sided Feedback
链接: https://arxiv.org/abs/2610.23812
作者: Kaixuan Ji,Zixiang Chen,Wenting Zhao,Cheng Qian,Heng Wang,Weizhi Zhang,Jielin Qiu,Liangwei Yang,Quanquan Gu,Silvio Savarese,Shelby Heinecke,Huan Wang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Long-term memory enables language-model agents to personalize future interactions, but a finite memory bank cannot preserve every extracted fact. Existing memory gating policies are typically shared across users and therefore fail to account for user-specific patterns of future reuse. In this paper, we study memory gating at the fact level. We first propose a retrieval-based labeling strategy that labels each fact by whether it is later retrieved and incorporated into responses. We then formulate online memory gating as contextual apple tasting with delayed one-sided feedback, and propose Warm-Start UCB, which combines offline learning from existing data with online adaptation to the current user. Our theoretical analysis establishes a high-probability regret upper bound of \tildeO(T\epsilon \wedge d\sqrtT) when the delay window is at most O(\sqrtT) , where T is the total number of rounds, \epsilon is the initialization error, d is the dimension of the feature map, and \tildeO(\cdot) hides constant and poly-logarithmic factors. The results of our experiment demonstrate that our method outperforms all baselines under different memory budgets and history lengths, establishing it as a promising personalized memory gating policy.
[LG-2] Explainable Machine Learning Analysis of Temperature Effects on Mechanical Strength of Fiber-Reinforced High-Performance Concrete
链接: https://arxiv.org/abs/2610.23811
作者: Jagaran Chakma,Zhiguang Zhou,Badhan Chakma,Priyanta Chakma
类目: Machine Learning (cs.LG)
*备注: 29 pages
Abstract:Predicting the compressive strength of fiber-reinforced high-performance concrete subjected to temperature exposure is challenging because to the intricate interdependencies of mixture composition, fiber properties, curing conditions, and temperature. A data-driven framework to predict compressive strength utilizing 1,204 experimental records containing a diverse array of material designs and exposure situations. This study evaluates eight machine-learning models and demonstrate that ensemble approaches routinely surpass individual learners. Both gradient boosting and extreme gradient boosting exhibit superior performance on unseen data, each attaining a R^2 of 0.903. Gradient boosting produces the minimal root mean square error (6.67 MPa), whereas extreme gradient boosting achieves the lowest mean absolute error (3.97 MPa) and mean absolute percentage error (10.60%). The models are further interpreted utilizing Shapley additive explanations and partial dependence analysis, which reveal temperature as the primary factor influencing compressive strength, succeeded by fine aggregate, superplasticizer, cement content, water-to-binder ratio, and water content. Strength diminishes significantly with rising temperature, albeit the extent of deterioration is contingent upon the specific combination design. The best performing models are subsequently implemented in a Python-based graphical user interface for expedited prediction. Our findings offer a precise and comprehensible methodology for evaluating temperature-dependent compressive strength and facilitate the practical, data-driven design of fiber-reinforced high-performance concrete.
[LG-3] Logic-Guided Graph Denoising Diffusion Models AAAI2027
链接: https://arxiv.org/abs/2610.23766
作者: Samuel Cognolato,Luciano Serafini,Alessandro Sperduti,Andrea Cavallaro
类目: Machine Learning (cs.LG)
*备注: Manuscript under review at AAAI2027
Abstract:Graph diffusion models are competitive approaches for generating complex graph-structured data. However, when such distributions have strict structural constraints, diffusion models may fail to capture them exactly, resulting in the generation of invalid graphs. In this paper, we propose a method for grounding graph diffusion models in domain-specific knowledge via First Order Logic (FOL), leveraging its expressive and interpretable formalism to impose structural constraints on the generative process. We incorporate domain knowledge using classifier guidance, a technique that conditions the diffusion model’s distribution on external variables to exert control over the sampled data. During the generative process, classifier guidance steers the denoising trajectory to optimize an objective function, which we define as a fuzzy and differentiable relaxation of the logical rules. Our methodology requires only the specification of constraints in FOL, whose satisfaction is encouraged at inference time to guide a pre-trained generative model. We evaluate our approach by comparing the unconditional generation of knowledge graphs with our logic-guided framework, which incorporates domain ontologies as structural constraints during inference. This approach significantly increases the proportion of generated graphs that satisfy the ontological constraints.
[LG-4] Universal Semantic Enhancement for Weakly Supervised Anomaly Detection
链接: https://arxiv.org/abs/2610.23765
作者: Siyu Zhou,Yuqin Xie,Hao Yuan,Yingjie Zhou
类目: Machine Learning (cs.LG)
*备注: this https URL
Abstract:Weakly supervised anomaly detection on tabular data is an essential task in many label-scarce applications, since tabular data is widely encountered and it is commonly observed in practice that only a limited number of labeled anomalies are available alongside abundant unlabeled samples. Most existing methods try to learn features that capture the inherent distribution patterns in the target domain. However, such features often struggle to provide a complete description of possible abnormal characteristics and are highly susceptible to overfitting the extremely scarce labels. To address this challenge, we propose a weakly supervised anomaly detection method based on universal semantic enhancement for tabular data, which fuses the universal knowledge from a pre-trained tabular foundation model with the domain-specific knowledge learned on the target dataset. The two types of knowledge are naturally complementary: universal semantics describe the common patterns of tabular data across domains, while the domain-specific features capture the inherent distributions of normal and anomalous behaviors in the target dataset. A prototype-conditioned modulation mechanism is then designed to effectively integrate the two types of features, allowing them to reinforce each other for improved anomaly detection. Extensive experiments on public datasets show that the proposed method outperforms competitive methods when only eight labeled anomalies are available in the training procedure.
[LG-5] LagNet: Lag-Structured Operators for Long-Term Time Series Forecasting
链接: https://arxiv.org/abs/2610.23735
作者: Zhenan Yu,Guangxin Jiang,Jiuyun Jiang,Jin Yang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Long-term time series forecasting maps a finite history window to an extended future horizon. Many existing models focus primarily on the design of input representations and model structures, while paying less attention to the structural form of the history-to-future forecasting operator itself. We examine this operator directly and find that, across model families, sensitivity matrices exhibit recurring diagonal patterns whose diagonals correspond to relative temporal lags. Motivated by this observation, we develop a framework for direct multi-step forecasting based on lag-structured temporal operators. Its basic building block, LagLayer, constructs a Toeplitz matrix by tying history-to-future connections with the same relative lag and incorporates horizon-wise and history-wise scaling coefficients to relax exact sharing. This design enables LagLayer to serve as an effective replacement for temporal linear layers in existing forecasting models. Stacking LagLayer modules with activations and multiple feature maps yields LagNet. Experiments on long-term time series forecasting benchmarks show that our method is competitive with strong baselines, while sensitivity matrix diagnostics and limited-data experiments demonstrate that LagLayer provides an effective and sample-efficient alternative to a linear layer.
[LG-6] Escaping Parameter Space: Tight Generalization Bounds via Representation Quality
链接: https://arxiv.org/abs/2610.23729
作者: Niclas Göring,Shuofeng Zhang,Branton DeMoss,Ard A. Louis
类目: Machine Learning (cs.LG)
*备注:
Abstract:We derive a tight generalization bound for deterministic neural We derive a tight generalization bound for deterministic neural networks based on their last hidden layer representations. By evaluating representation quality with a parameter-free k -nearest-neighbor ( k -NN) classifier, our certificate avoids traditional parameter-space complexity measures, KL divergences to a posterior, weight norms, and compression. The bound is tight: trained from scratch, it reaches an 8.0 percentage point gap to test error on CIFAR-10 (ResNet-18) and yields the first non-vacuous certificates we are aware of on CIFAR-100; with frozen DINOv2 representations, it certifies at 1.4% on CIFAR-10 and 23.1% on 1,000-class ImageNet. It decomposes additively into three empirical, independently measurable quantities: local class geometry ( k -NN decoder error), decoder-network agreement, and training stability. This allows the bound to act as a diagnostic tool: we demonstrate that actively destroying local class geometry eliminates generalization while leaving dataset memorization intact, identifying class geometry as a causally relevant mechanism for generalization. Finally, we show that learned representations are significantly less sensitive to weight noise than linear readouts, suggesting why parameter-space certificates often struggle to remain tight.
[LG-7] Blitz: Accelerating the Diffusion Modules of AlphaFold3 and Boltz-2
链接: https://arxiv.org/abs/2610.23703
作者: Vinay Saji Mathew,Soundar Kumara,Gretta Kellogg,William K. M. Lai
类目: Machine Learning (cs.LG)
*备注: 24 pages, 12 figures
Abstract:Diffusion-based protein structure prediction models produce all-atom structures in about 200 denoising steps, and that cost has driven a number of few-step samplers. We present Blitz, a few-step distillation framework that uses a geometric noise schedule to distribute steps more evenly across noise scales, then learns those updates through multistep consistency training. Physical steering and target-free candidate selection help reduce the remaining geometric defects at inference. We apply Blitz to AlphaFold 3 and Boltz-2, reducing denoising from 200 steps to eight. Because published few-step results use different benchmarks and scoring procedures, we compare both distilled models with their teachers and other few-step methods under a common evaluation protocol covering structural accuracy, interface quality, ligand placement, and stereochemical validity. Blitz retains teacher-level structural accuracy, with AF3-Blitz outperforming all evaluated few-step baselines and exceeding its teacher on several accuracy metrics.
[LG-8] Equivalence testing and run-count design for stochastic machine-learning experiments
链接: https://arxiv.org/abs/2610.23700
作者: Dmytro Knopov,Nadiya Shvai
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 31 pages, 4 figures, 4 tables. Code, simulation scripts and case-study data: this https URL (release v0.2.0, archived at this https URL )
Abstract:Claims that two training methods perform equally well often rest on a non-significant difference between two or three runs per method. We develop a protocol, with its power theory, for certifying practical equivalence in per-run metrics such as the AUROC (area under the receiver operating characteristic curve). It rests on a two-level (seed x item) variance model separating between-run dispersion from item-level noise. Equivalence within a pre-specified margin is tested by two one-sided tests (TOST), equivalently by a confidence interval inside the margin; a two-level bootstrap covers composite statistics. A closed-form formula converts a margin and a target power into runs per method; its shifted-t refinement is calibrated against the realised two-level process. Simulations on a 270-cell grid and under non-Gaussian seed laws confirm the formula within its design-relevant range, from five runs per method. For four published runs of a joint energy-based model, the failure to certify equivalence was largely decided by design: with the variance calibrated to those runs, two per method certify with probability near three per cent, whereas certifying 0.01 AUROC at 80 per cent power needs twelve even under exact equality. A second case study on a thousand-seed CIFAR-10 zoo and the MultiBERTs checks the formula against real seed laws (within 0.05 in that range) and resolves comparisons that two or three runs leave undecided. In a two-stage design (pilot, then formula), our experiments suggest that an upper confidence bound on the pilot variance substantially reduces the undersizing seen with its point estimate.
[LG-9] HardEval: An Execution-Verified Benchmark for AI-Driven Security Hardening and Compliance Remediation
链接: https://arxiv.org/abs/2610.23685
作者: Manohar Vellala,Soumya Banerjee,Sandip Roy,Ross Gore,Sachin Shetty
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注: 8 pages, 4 figures, 4 tables. Code and data: this https URL The paper has 4 figure environments (two are images, two are the example-prompt panels) and 4 tables
Abstract:Implementing Security Technical Implementation Guide (STIG) controls remains a demanding operating-system administration task. Each natural-language security requirement must be translated into a precise and persistent configuration change. These changes span Secure Shell (SSH) configuration, authentication policy, audit rules, and kernel parameters. Existing benchmarks do not measure whether a model can perform this task on a live system. Code benchmarks evaluate isolated programs without persistent system state, while agent benchmarks largely focus on applications rather than operating-system configuration. In this paper, we introduce \textbfHardEval, an execution-verified benchmark for evaluating language-model-assisted security hardening. HardEval is built from Red Hat Enterprise Linux 8 STIG controls that fail on an unhardened baseline system. Expert-written reference remediations and Open Vulnerability and Assessment Language (OVAL) checks serve as hidden ground truth for grading; they are never shown to the model, and each task states only the required security outcome while withholding the implementation. The model produces a Bash script, which is executed on a live virtual machine and graded from fail to pass using the corresponding OpenSCAP/OVAL compliance check. Evaluation of fourteen frontier and open-weight configurations reveals large and uneven capability differences across security domains. A recurring failure mode is plausible security code that does not satisfy the exact compliance condition. For four open-weight configurations, multi-run Wilson confidence intervals preserve the observed ordering. HardEval is publicly available.
[LG-10] riRouter: Sparse-to-Dense Electrode-Space Routing for Robust EEG Motor Imagery Decoding ICASSP2027
链接: https://arxiv.org/abs/2610.23665
作者: Ying Zhou,Yaoxin,Mengting Liu
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注: Submitted to IEEE ICASSP 2027
Abstract:Motor imagery brain-computer interfaces require reliable decoding from multichannel EEG, but practical acquisition often yields sparse or corrupted electrode montages. We propose TriRouter, an electrode-space routing framework with discriminative-preservation, mask-robustness, and latent-compensation paths. A validation-calibrated router uses observable channel availability to allocate path contributions, preserving complete-channel discrimination while increasing representation recovery under severe missingness. In BNCI2014-001, TriRouter outperforms standard public backbones and a dynamic-spatial-filtering corrupted-EEG baseline. It improves accuracy over the strongest corrupted-EEG baseline by 3.68, 4.02, and 6.46 percentage points at nominal 25%, 50%, and 75% missing settings, and remains effective under structured regional missingness and additional low-channel dataset evaluation.
[LG-11] JAXtari: High-Throughput and Easy-to-Modify Arcade Learning Environment
链接: https://arxiv.org/abs/2610.23650
作者: Quentin Delfosse,Raban Emunds,Paul Seitz,Sebastian Wette,Jannis Blüml,Daniel Kirn,Dominik Mandok,Kristian Kersting
类目: Machine Learning (cs.LG)
*备注:
Abstract:The Arcade Learning Environment (ALE) has been a cornerstone of Deep Reinforcement Learning research, providing a standardized benchmark for evaluating general agents. However, its lack of native support for object-extraction and environment modification restricts research into evaluating whether agents learn robust causal models or rely on brittle visual heuristics. Furthermore, the ALE’s dependence on CPU-based emulation creates a significant bottleneck in modern GPU-accelerated training pipelines, limiting both experimental scale and iteration frequency. In this work, we introduce JAXtari, an open-source, high-performance reimplementation of Atari 2600 environments. By leveraging massive GPU parallelization, JAXtari allows PPO agents to finish training on 100M environment steps on a single A100-GPU in under an hour. Beyond speed, it provides object-centric observations and a flexible modification system, allowing researchers to, e.g. develop neuro-symbolic methods or discover novel shortcuts and subsequently train and adapt their agents rapidly, sometimes in only minutes.
[LG-12] No-Regret Language Generation
链接: https://arxiv.org/abs/2610.23641
作者: Jiaxun Li,Vinod Raman,Ambuj Tewari
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:We introduce no-regret generation, which requires a generator to make a sublinear number of mistakes uniformly over all demonstration sequences and compatible target languages. We also formulate an agnostic variant that allows arbitrary demonstrations that may not be compatible with any languages. For both settings, we introduce tree-based mistake-forcing dimensions that exactly characterize finite-horizon minimax complexity and establish necessary and sufficient conditions for no-regret generation. Our characterizations reveal a rich spectrum of generation rates: up to constant rescaling, every admissible nondecreasing profile is attainable in both the realizable and agnostic settings. Finally, we connect generation complexity to online prediction: for language classes of finite Littlestone dimension (d), we obtain realizable mistake bounds of (O(d\log(1+T/d))) and agnostic regret bounds of (O(\sqrtdT\log(1+T/d))). The realizable bound is optimal up to universal constants.
[LG-13] Deep LLM layers expose passage-level context in EEG during silent reading
链接: https://arxiv.org/abs/2610.23638
作者: Ingo Marquardt,Anthilia Alchanat,Priyanka Jain
类目: Machine Learning (cs.LG); Neurons and Cognition (q-bio.NC)
*备注: 41 pages (including 23 pages of Supplementary Material), 12 figures (including 5 in Supplementary Material)
Abstract:In a previous report we showed that a decoder can recover open-vocabulary word-level information from 19-channel EEG recorded during silent reading, by aligning short EEG windows with hidden-state embeddings of the presented word taken from a large language model (LLM). That analysis contrasted two of the model’s 33 layers: the context-free input embedding and one mid-depth layer. Here we ask how EEG-decodable structure is distributed over the full depth of the language model. We trained one decoder per layer of Llama-3.1-8B, with and without a causal transformer over the sequence of EEG windows. We distinguish word-level decodability from a context-tracking component. Word-level decodability peaks in early layers and declines steadily towards the last layer. In contrast, the context-tracking component of the retrieval gain rises with depth, both with and without a causal transformer. Unexpectedly, a decoder that sees a single 0.5-s window at a time tracks the text passage too. Two new controls dissect this component. A target-geometry control, a synthetic predictor made of the true target plus noise calibrated to each decoder’s word-level accuracy, shows that context tracking is not a by-product of word decoding and target geometry. A control that permutes predictions across runs from either the same or from different sessions shows that most of the observed context-tracking is not caused by an EEG recording session signature. Instead, context tracking mostly follows the text. A single EEG window therefore carries passage-level information that a contextual target can absorb. The results locate word identity in the early layers of the language model, discourse state in the deeper layers, and make the choice of target layer a design dial for EEG language decoders.
[LG-14] Deconstructing Multi-Task Active Learning: The Paradox of Gradient Conflict and Orthogonal Decomposition NEURIPS2026
链接: https://arxiv.org/abs/2610.23623
作者: Juho Jung,Seongjae Kang,Jinyoung Han
类目: Machine Learning (cs.LG)
*备注: 10 pages, 5 figures, NeurIPS 2026
Abstract:Existing multi-task active learning (MTL-AL) methods predominantly rely on linear scalarization of heterogeneous objectives such as uncertainty, diversity, and gradient conflict. We argue that this entangled formulation is fundamentally limited, as linear combination fails to capture Pareto-optimal trade-offs under conflicting objectives. We propose a shift from combination to deconstruction and introduce a principled framework that rethinks MTL-AL along three orthogonal dimensions. First, we identify the Conflict Paradox: although the optimization literature treats gradient conflict as interference to be suppressed, penalizing conflict when ranking candidates is measurably worse than not scoring it at all. We therefore decouple the two concerns: the acquisition function does not score conflict, and the resulting interference is resolved by gradient surgery during training. Second, we introduce Orthogonal Decomposition, separating acquisition into Informativeness and Exploration. This two-stage process, informativeness filtering followed by coverage maximization, prevents “diversity dilution” by excluding uninformative outliers before enforcing diversity. Third, we propose Cross-Task Mutual Information (CTMI), a covariance-based statistic of cross-task dependency computed from MC-dropout task entropies. Extensive experiments on COCO, Cityscapes, and NYUv2, including comparisons against recent MTL-AL and data-selection baselines, demonstrate consistent improvements over state-of-the-art methods, highlighting that structural rethinking of acquisition yields larger gains than incremental heuristic design. Code is available at this https URL.
[LG-15] GRAFT-EEG: Gauge-Aware Physical Field Transport for Cross-Dataset Motor-Imagery EEG ICLR2027
链接: https://arxiv.org/abs/2610.23620
作者: Ying Zhou,Mengting Liu
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注: Under review at ICLR 2027
Abstract:Cross-dataset EEG decoding often treats montage, reference, and active-sensor changes as representation shift, although they also alter the observation operator. We study target-statistics-blind transfer: fitting receives no target signal, label, or empirical statistic, and electrode coordinates only instantiate a fixed prediction operator. method removes the active-set common-reference gauge and transports each native mask through a template motor leadfield. A source-fitted spectral classifier anchors the decision; a bounded state-space-model (SSM) residual is secondary. We establish restricted identifiability and an operator-perturbation bound. Across five leave-one-dataset-out targets (253 participants), dynamic method achieved 60.865% equal-target balanced accuracy versus 52.684% for strict-clean 5/5 BIOT (hierarchical difference +8.182 points; 95% CI +6.382 to +10.196; positive in every target). On four EEGPT-clean targets, the descriptive means were 61.520% and 59.057%. In four post-hoc exact same-parent tests, FULL exceeded coordinate-basis, matched-spectrum, row-shuffled, and wrong-anatomy controls by 9.959, 9.698, 11.241, and 3.985 points, respectively (all 95% CIs excluded zero; Holm-adjusted P0.001). A separate 60-participant same-paradigm evaluation retained a positive physical-operator contrast (+4.029 points; +2.338 to +5.773), whereas dynamic–static was inconclusive; its preregistered status is partial confirmation rather than unrestricted external validation. The SSM added +1.059 points (+0.013 to +1.892). Claims remain bounded to five binary motor-imagery domains and the tested operators.
[LG-16] hetreStageAI: Multi-View Retrieval-Augmented Generation for Structured Bengali Theatre Production Planning
链接: https://arxiv.org/abs/2610.23618
作者: Md. Mehedi Hasan Naeem,Mst. Kamrunnahar Ruma,Shakila Sultana,Nafiza Anjum,Nasreen Jannatun Nemu,Md. Sujan Ali
类目: Machine Learning (cs.LG)
*备注: 6 pages, 6 figures. Accepted at the 12th IEEE International Conference on Sustainable Technologies and Engineering (IEEE iCOSTE 2026), Cox’s Bazar, Bangladesh, November 1-3, 2026
Abstract:Staging a theatre performance requires coordinating dialogue, actor movement, stage directions, lighting focus, cue timing, and equipment constraints within the same production plan. This work presents ThetreStageAI, a retrieval augmented framework for structured Bengali theatre production planning. Rather than retrieving from one mixed corpus, the system maintains separate scene, actor-blocking, and lighting views. Bengali text is embedded with BGE-M3 and indexed in view-specific Qdrant collections. Retrieved references are combined with a production brief and passed to Gemini, which generates schema-constrained JSON containing characters, scenes, dialogue, stage directions, blocking, lighting, and optional sound cues. Structural and semantic checks are applied before acceptance, with at most one constrained repair attempt. We also construct a human-guided, LLM-assisted synthetic dataset of 500 short Bengali theatre records that expands to 1,500 retrieval documents. Dataset diagnostics confirm consistent identifiers and valid lighting fields while also revealing substantial dialogue repetition. A paired camera-ready evaluation compares single-combined and full multi-view retrieval under matched settings. The source code, dataset, and evaluation materials are publicly available.
[LG-17] he Sign Code: The Hidden Binary Nature of Deep Networks
链接: https://arxiv.org/abs/2610.23616
作者: Niclas Göring,Shuofeng Zhang,Roi Holtzman,Yoonsoo Nam,Ard A. Louis
类目: Machine Learning (cs.LG)
*备注:
Abstract:Every neuron in a trained network outputs a number that can be separated into its sign (positive or negative) and its magnitude (absolute size). We call the set of all neuron signs the \textitsign code. This raises a key question: does classification depend more on the sign code or on the magnitudes? For overparameterized vision classifiers with sign-preserving activations, the answer is overwhelmingly the sign code. A k -nearest neighbor classifier operating only on this binary sign code achieves accuracy comparable to the full model across different architectures, activation functions, and datasets. Moreover, replacing all magnitudes in every layer with random noise while keeping the original signs barely changes residual network performance. In contrast, flipping just 20% of the signs reduces accuracy to chance. Models generalize well when same-class inputs share similar sign codes. When this overlap is disrupted by a regularizer, training accuracy remains high, but test performance collapses. A margin decomposition helps explain these results: the sign-only contribution to each logit grows as \sqrtd in network width, dominating the magnitude residual. The natural unit of analysis for these networks may not be through their continuous weight space but instead through their discrete sign codes.
[LG-18] How Many but How Sure? Class Competition as a Confidence Proxy in Multi-Class Density-Based Counting
链接: https://arxiv.org/abs/2610.23611
作者: Villanelle O’Reilly,Georgios Leontidis,Marc Hanheide
类目: Machine Learning (cs.LG)
*备注: 2+1 pages, 1 figure; a 2027 Northern Lights Deep Learning Conference abstract
Abstract:Multi-class density estimators can count objects in crowded scenes, but their class-wise outputs offer no indication of how reliably density mass has been allocated. We introduce a training-free confidence proxy based on local competition between the strongest class channels. The resulting prominence measure requires only the predicted density maps, with no retraining or access to model internals. We test whether residual competition is diagnostic of cross-class leakage by attenuating contested density at inference time. On DSACA predictions for the Hicks et al. dataset, this reduces overall MAE from 0.920 to 0.846, suggesting that class competition contains useful information about count reliability.
[LG-19] When Magnitude Suffices: Reconstructing Time-Domain-Gated Microwave Signals Without Phase on a Thoracic Phantom
链接: https://arxiv.org/abs/2610.23608
作者: Maciej Cichoń,Patrycja Gołaś,Maksymilian Bartosik,Angelika Walczak,Agata Bielecka-Dąbrowa,Maciej Ślot
类目: Machine Learning (cs.LG); Applied Physics (physics.app-ph)
*备注: 18 pages, 7 figures, 4 tables
Abstract:Time-domain gating separates the arrivals that form a microwave S-parameter measurement. It requires the measurement of complex values, so this application almost always uses a vector network analyzer (VNA). We show that, on one calibrated measurement configuration, the gated magnitude the VNA would give can be reconstructed from a magnitude-only sweep of a spectrum analyzer with a tracking generator (SA). We select the gate center for separability of the measured states, correct each instrument once by a per-bin subtraction of a reference state, and train a fully connected network on sweeps from both instruments. The test object is a torso phantom with lung-shaped compartments whose filling takes eleven dielectric states, each held out in turn. On S_11 the reconstruction from SA input reaches a shape error of 0.08 dB ^2 with a state ordering of 0.81. This matches the reconstruction from VNA input. On S_21 it reaches 0.82 dB ^2 and keeps the ordered groups of the states. An explicit phase prediction performs much worse, because gating needs the phase to about 0.2 rad RMS. Across eleven measurement configurations the dependence of phase on magnitude is strong for reflection and weak for transmission. This explains the difference between S_11 and S_21 . Where the magnitude carries enough of the phase, a scalar instrument gives the gated response without any phase measurement after installation. On the open UM-BMID breast data set, with whole phantoms held out, the shape error of the network is 30 percent lower than that of a per-bin linear map.
[LG-20] Boundary Detection for High-Dimensional Point Clouds
链接: https://arxiv.org/abs/2610.23590
作者: Abigail Hickok,Valerio Galanti,Osama Al Dalahmah,Andrew J. Blumberg
类目: Machine Learning (cs.LG); Computational Geometry (cs.CG); Machine Learning (stat.ML)
*备注: 31 pages; 23 figures
Abstract:We introduce two new methods for boundary detection in high-dimensional point cloud data, intended to work in situations with non-uniform density, high ambient dimension, and substantial noise. Our first method fits a separating hyperplane on local neighborhoods after doing local dimension reduction—this method works best when one can reliably estimate the intrinsic dimension of the data. Our second method measures the directional uniformity of local neighborhoods, and has the best overall performance on synthetic data. In addition to pointwise numerical boundary scores, our algorithms yield an estimated normal vector at each boundary point. We validate our methodology on synthetic data as well as two real examples: we analyze spatial transcriptomics data, where we show that we can detect the boundary of a tumor, and we also study LLM activation geometry.
[LG-21] BMuon: Budgeting Muons Spectral Geometry for Structured Compressibility
链接: https://arxiv.org/abs/2610.23571
作者: Rida Assalouh,Tom Jacobs,Rebekka Burkholz,Yani Ioannou,Umut Şimşekli
类目: Machine Learning (cs.LG)
*备注: 54 pages, 6 figures
Abstract:Muon has emerged as a strong optimizer for neural network training and admits an interpretation as steepest descent under a spectral-norm geometry. Yet we find that Muon-trained Transformers can degrade sharply when their MLP hidden channels are pruned after training. To reconcile Muon’s optimization performance with structured compressibility, we introduce Budget Muon (BMuon), which modifies Muon’s spectral geometry so that hidden channels across blocks compete for a shared allocation budget. The resulting oracle is exactly normalized steepest descent under a family of norms that interpolate between Muon’s geometry and structured sparse geometries. Our principal sparsity control is an interpretable global channel-allocation budget, rather than a setting-dependent sparsity-penalty coefficient. For linear factorization models, we show that Muon admits minimum-cost exact representations with redundant hidden width, whereas under the BMuon oracle geometry and a sufficiently scarce budget, every minimum-cost representation is width-minimal. For practical training, we derive a minorization-maximization surrogate for the exact allocation problem. On ImageNet-1K with ViT-S/16, BMuon improves over Muon under gradual structured MLP pruning, and its learned allocations provide a more effective pruning criterion than weight magnitude. On a 1B-parameter LLaMA-style model trained on FineWeb, BMuon improves over Muon under post-training MLP channel pruning.
[LG-22] Beyond Expected Values: Risk-Sensitive Planning with Distributional Monte-Carlo Tree Search
链接: https://arxiv.org/abs/2610.23565
作者: Quynh Trinh,Trung Nguyen,Nam Nguyen,Tuan Dam
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: Proceedings of the 40th Annual Conference on Neural Information Processing Systems
Abstract:Monte-Carlo Tree Search (MCTS) typically stores only scalar mean estimates, which makes it poorly suited for tail-risk objectives such as conditional value-at-risk (CVaR). We introduce Categorical Thompson Sampling with Optimism (CATSO) and Particle Thompson Sampling with Optimism (PATSO), two distributional MCTS algorithms that maintain finite-support empirical backup laws at Q-edges. CATSO represents Q-edge laws with categorical atoms and Dirichlet Thompson sampling, while PATSO represents them with adaptive particles. Both algorithms select actions using the CVaR of a Thompson-sampled Q-edge law plus a polynomial optimism bonus, and propagate scalar continuation values through visit-weighted averages of child Q-edge CVaR scores. For the finite-depth nested CVaR planning target, we prove an O(n^-1/2) bound on the root value error after n rollouts, which upper-bounds the risk-sensitive simple regret. CATSO adds a fixed-grid discretization bias, while capped PATSO adds a tunable Wasserstein compression bias. Experiments on stochastic planning benchmarks show that distributional Q-edges improve lower-tail return and reduce catastrophic outcomes when mean-optimal and risk-sensitive behavior differ.
[LG-23] Out of Support Not Out of Reach: In-Context Learning with Unsupported Instructions and Composed Tasks
链接: https://arxiv.org/abs/2610.23528
作者: Xuhan Tong,Yuchen Zeng,Jiawei Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:In-Context Learning (ICL) enables pretrained LLMs to adapt to downstream tasks from a small set of input-output demonstrations, without parameter updates. Yet existing theory primarily explains ICL as identifying a latent task already represented in pretraining, leaving unclear whether ICL can succeed when either the instruction or the target task falls outside the pretraining support. We develop a theoretical framework under mild assumptions that separates these two forms of distribution shift. For target tasks represented in pretraining, we show that demonstrations can identify the task and reduce sensitivity to pretrain-supported prompt templates as their number increases. For an unsupported instruction, we show that under a fixed model-induced reference extension, demonstration evidence can dominate an N -independent instruction shift while preserving the exponential identification rate. We then use the same ICL framework as an interpretive lens for long-horizon target tasks that are not represented in pretraining but admit a plan that decomposes them into pretraining-supported subtasks. Applying the supported-task analysis step by step yields a standard accumulated-error bound and clarifies how invoking familiar subtasks along an intermediate-state trajectory can avoid directly selecting the unsupported complete task. Together, our results characterize when demonstrations, instructions, and task decomposition can support ICL beyond the corresponding pretraining supports.
[LG-24] Interpreting SSL Representations for Spoof Detection: a WavLM Study
链接: https://arxiv.org/abs/2610.23526
作者: Mohamed Mallat,Michele Panariello,Massimiliano Todisco,Nicholas Evans,Anthony Larcher
类目: ound (cs.SD); Machine Learning (cs.LG)
*备注: 5 pages, 4 figures, 1 table. Published at IWBF 2026. An extended abstract of this work was presented as a poster at Odyssey 2026
Abstract:Self-supervised speech models such as WavLM achieve strong spoofing detection performance, yet which layers encode the most spoof-relevant information and what acoustic properties they capture remain opaque. We present a layer-wise interpretability study of WavLM using the ASVspoof 5 database, combining three complementary analyses: (i) per-layer linear probing with attack-level evaluation across speech synthesis, voice conversion, and adversarial attacks; (ii) a canonical correlation analysis (CCA) linking layer representations to spoofing-relevant acoustic features; (iii) an exhaustive grid search over layer subsets showing that use of five layers outperforms full thirteen-layer pooling. CCA analysis reveals that, while most acoustic features peak in alignment with mid-layers and then decay, voice quality measures jitter, shimmer, and the harmonics-to-noise ratio show the opposite trend, with correlation increasing toward later layers. Embedding visualisations reveal that deeper layers capture a continuous acoustic similarity among attacks rather than discrete categories, suggesting these layers encode fundamental synthesis artifacts rather than attack-specific signatures.
[LG-25] Uncertainty-aware AI-enabled framework for predicting hydrogen dispersion in confined spaces to enhance operator situational awareness
链接: https://arxiv.org/abs/2610.23515
作者: Niclas Flehmig,Mary Ann Lundteigen,Shen Yin
类目: Machine Learning (cs.LG)
*备注: Submitted to Journal of Loss Prevention in the Process Industries
Abstract:We present a two-stage AI-enabled framework for enhancing operator situational awareness during compressed hydrogen bunkering in maritime confined spaces. In the event of a leak, the framework first estimates the unknown leakage rate from early sensor readings and then forecasts how the hydrogen cloud will disperse over the next time steps, providing operators with a predicted concentration map together with confidence bounds. The framework is based on two independent AI models building one AI system. The AI system is designed as a decision-support tool. It augments sensor readings, and thus enhances operator situational awareness and supports decision-making. This enables preventive safety barriers to be deployed before hazardous concentrations are reached, while an independent emergency shutdown system remains the ultimate safety layer. The system design incorporates safeguards to ensure the AI system operates within its competence envelope. Trained and tested on an open dataset of confined-space hydrogen releases, the framework achieves a mean absolute error of 0.052 ~g/s in leakage-rate estimation and \approx0.018 in hydrogen-concentration prediction, approaching the sensor precision bound of \pm0.01 . The hydrogen-concentration prediction provides well-calibrated uncertainty estimates that allow operators to judge the reliability of each forecast. These results suggest that the framework may provide early information about hazardous concentration build-up, potentially supporting preventive actions by operators.
[LG-26] Graphlet-based Decentralized Finance Transaction Method Classification
链接: https://arxiv.org/abs/2610.23466
作者: Célestin Coquidé,Alessio Brini,Natkamon Tovanich
类目: ocial and Information Networks (cs.SI); Computational Engineering, Finance, and Science (cs.CE); Computers and Society (cs.CY); Emerging Technologies (cs.ET); Machine Learning (cs.LG)
*备注: 24 pages, 5 figures, 5 tables, 25 references 15th International Conference on Complex Networks and Their Applications
Abstract:Decentralized Finance (DeFi) transactions involve complex interactions among accounts, smart contracts, and tokens. We investigate whether network topology can identify transaction methods that describe DeFi operations. We extract directed graphlet features to classify the 30 most frequent methods. Our results show that graphlet features complement event log counts in improving classification performance. Using SHapley Additive exPlanations (SHAP), we further derive model explanations that reveal method-specific graphlet structural patterns.
[LG-27] Predictable Scaling: Efficient Model Exploration via Scaling-Law Prediction
链接: https://arxiv.org/abs/2610.23453
作者: Aishwarya Wesanekar,Sheng-Tao Yang,Jiawei Zhang,Flavio Sales Truzzi,Chen Chen,Nader Beigiparast,Shengyuan Huang,Roshanak Zilouchian Moghaddam,Vitor Cid,Colin Clement,Vinayak Thapliyal,Sebastian Ament,Jihao Andreas Lin,Vincent Huang,Kai Han,Boyi Jiang,Wenlin Chen,Chaorong Chen,Ashwin Kumar,Maximilian Balandat,David Eriksson,Yifei Zhang,Pan Zhong
类目: Machine Learning (cs.LG)
*备注:
Abstract:Evaluating candidate changes (hypotheses) to recommendation models requires full-scale training runs consuming thousands of GPU-hours, severely limiting the hypothesis space engineers can feasibly explore. We introduce Predictable Scaling (PS), a systematic framework that exploits empirically verified regularity conditions to predict full-scale model performance from small-scale proxies. PS constructs scaled-down variants via composable scaling functions along distinct resource axes, and projects metrics to the target scale via curve fitting with calibrated uncertainty estimates. We compare three prediction strategies of increasing expressiveness: even the simplest (raw proxy) achieves prediction error within the empirical noise floor of repeated-run variance; the most expressive (Gaussian Process) provides an indicative 15% relative MAE improvement and calibrated uncertainty. Evaluation on two ranking models of increasing complexity—a task-specific model and a foundation model—demonstrates that PS correctly ranks actionable hypotheses across diverse hypothesis types at 79-88% compute reductions.
[LG-28] Regularized free deconvolution for spectral repair of resampling distortion in high-dimensional particle filters
链接: https://arxiv.org/abs/2610.23448
作者: Deming Song,Chi-Kin Lam,Kefeng Li,Hui Yu,Xilin Li,Huagang Lin
类目: Machine Learning (cs.LG)
*备注: This is a preprint of an article submitted to Signal Processing (Elsevier). The implementation and experiment scripts are available at this https URL . This is the second paper in a series on spectral analysis of particle filters. The first paper is available at arXiv:2610.16680
Abstract:Resampling is the standard remedy for weight degeneracy in particle filters, but in high dimensions it distorts the spectrum of the ensemble covariance in a way that is computable from the importance weights. This paper studies the inverse problem: recovering the pre-resampling spectrum from the observed weights and the empirical spectrum alone, and using it without biasing the filter. We first show that the forward spectral law extends to random importance weights once one conditions on the weights together with the realized offspring counts, rather than on the weights alone. We then construct a regularized free-deconvolution estimator, prove its consistency with an explicit error budget at a fixed resolution scale, and certify model mismatch through a computable forward residual. The repaired spectrum enters the filter only as the preconditioner of a Metropolis-adjusted rejuvenation move, so that spectral error affects mixing efficiency but not the invariant target. For the full filtering cycle, we derive a geometrically summable error recursion under stable scalar dynamics and an explicit law for the cost of abstaining from repair. Pre-registered simulations show that the inversion reduces the spectral error by an order of magnitude, that the corrected preconditioner recovers about 90% of the oracle’s sampling efficiency in an anisotropic regime, and that the cost of abstention depends on when abstentions occur rather than on how many. The repair recovers eigenvalue laws, not eigenvectors, and is certified only within an explicitly stated exogenous-weight regime.
[LG-29] Bridging Compressed Communication and Local Steps via Spectral Method in Federated Learning
链接: https://arxiv.org/abs/2610.23434
作者: Dmitriy Bystrov,George Chirkov,Daniil Medyakov,Dmitry Bylinkin,Aleksandr Beznosikov
类目: Machine Learning (cs.LG)
*备注:
Abstract:Federated Learning (FL) enables collaborative training of complex models across large-scale networks of edge devices while keeping data localized. However, data heterogeneity and high communication overhead remain fundamental bottlenecks. Recently, the spectral optimizer Muon has demonstrated remarkable advantages over Adam-like methods in centralized training. While recent efforts have attempted to transition Muon to FL, naive integrations into traditional architectures fail short. Crucially, existing approaches do not support the the communication compression and local steps — the two foundational techniques of efficient FL. To address this limitation, we introduce a novel spectral FL method that utilizes both local updates and compression. We establish rigorous convergence guarantees under mild assumptions, notably eliminating the restrictive bounded heterogeneity condition. Empirical evaluation on standard image classification benchmarks validates the efficiency and robustness of our approach.
[LG-30] Residual Stochastic Concept Bottleneck Models
链接: https://arxiv.org/abs/2610.23421
作者: Stephen Marcou,Mateo Espinosa Zarlenga,Sonia Laguna,Moritz Vandenhirtz,Julia E. Vogt,Mateja Jamnik
类目: Machine Learning (cs.LG)
*备注:
Abstract:Concept-based models (CMs) learn human-interpretable intermediate representations that enable users to understand the model’s reasoning and steer its predictions through test-time concept interventions. However, in realistic settings, their performance can be limited by concept incompleteness, where the training concepts fail to capture all task-relevant information. Adding an unconstrained residual pathway to the model’s predictive process can overcome the consequences of incompleteness. However, this introduces an inscrutable, leak-prone prediction pathway that may weaken concept interventions, as these often leave residual variables unchanged. To address this limitation, we introduce Residual Stochastic Concept Bottleneck Models (Res-SCBMs), a family of concept-based models that jointly model concept and residual variables through a multivariate Gaussian distribution. Res-SCBMs learn a covariance structure that explicitly captures dependencies between the known concepts and learnt residual variables, allowing interventions to propagate to statistically dependent residual variables through Gaussian conditioning. This extends the intervention capabilities of previous CMs without requiring concept completeness during training. Our results show that, across multiple concept-incomplete benchmarks, Res-SCBMs improve predictive performance over competing baselines while retaining effective test-time concept interventions.
[LG-31] Flash RNN-T: Memory-Efficient Joint Network and Transducer Loss Computation INTERSPEECH2027
链接: https://arxiv.org/abs/2610.23409
作者: Mahmoud Ashraf
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC); Sound (cs.SD)
*备注: Will be submitted to Interspeech 2027
Abstract:The recurrent neural network transducer (RNN-T) remains a practical choice for streaming speech recognition, but training requires large intermediate tensors. We present Flash RNN-T, a joint and loss implementation combining specialized kernels with optional tiled recomputation over the full alignment space and vocabulary. Tiled and untiled Flash form the entire measured throughput-memory frontier on medium and long workloads. On H100, untiled Flash achieves 24-39% higher throughput with 26-56% less memory than the fastest baseline on these workloads wherever it fits. Tiling reduces memory further through recomputation. On medium, tiled Flash uses ninefold less memory than that baseline at 16% lower throughput when both use one backward call per batch. On long, it fits where all tested alternatives with that backward schedule run out of memory with 80 GB. These comparisons use each implementation’s lowest-memory configuration reaching at least 95% of its peak throughput.
[LG-32] Information Estimation with Discrete Diffusion ICLR ICLR2026
链接: https://arxiv.org/abs/2610.23404
作者: Alberto Foresti,Giulio Franzese,Pietro Michiardi
类目: Machine Learning (cs.LG); Information Theory (cs.IT)
*备注: Published as a conference paper at ICLR 2026 (Poster). ICLR Paper: this https URL OpenReview: this https URL . Code: this https URL
Abstract:Information-theoretic measures, such as Mutual Information (MI), play a crucial role in understanding non-linear relationships between random variables and are widely used across scientific disciplines. Yet, their use on real-world discrete data remains challenging. Existing methods typically rely on embedding discrete data into a continuous space and apply neural estimators originally designed for continuous distributions. This process requires careful engineering for both the embedding model and estimator architecture, but suffers from issues related to high data dimensionality. In this work, we introduce InfoSEDD, a discrete diffusion-based approach that bridges information-theoretic estimation and generative modeling such that they can be used to compute Kullback-Leibler divergences. Backed by Continuous Time Markov Chains theory principles, the design of InfoSEDD is lightweight and scalable and allows seamless integration with pretrained models. We showcase the versatility of our approach through applications on motif discovery in genetic promoter data, semantic-aware model selection in text summarization, and entropy estimation in Ising models. Finally, we construct consistency tests on real-world textual and genomics data. Our experiments demonstrate that InfoSEDD outperforms alternatives that rely on the “embedding trick”. Our results position InfoSEDD as a robust and scalable tool for information-theoretic analysis of discrete data.
[LG-33] Contracting with an Agent under Model Mismatch
链接: https://arxiv.org/abs/2610.23402
作者: Elif Yılmaz,Jakub Tłuczek,Victor Villin,Christos Dimitrakakis
类目: Machine Learning (cs.LG); Computer Science and Game Theory (cs.GT)
*备注:
Abstract:In the classical principal-agent setting, the principal delegates some task to the agent, who receives a payment conditioned on the outcome of the task, which in turn benefits the principal. To align their objectives, the principal designs a contract that incentivizes the agent to take actions beneficial to the principal. In practice, however, the agent may act under misspecified beliefs about the underlying environment, leading to systematically suboptimal decisions. In this work, we study this problem through a principal-agent Markov Decision Process (PAMDP), in which the principal repeatedly interacts with an agent whose transition model may differ from the true environment dynamics. The principal has access to the true transition model and designs contracts that account for the agent’s belief and incentives. We characterize contract design under model mismatch and develop a framework that enables the principal to induce desirable long-term benefits even though the agent has a potentially inaccurate understanding of the environment.
[LG-34] Forecasting Voltage Time Series under Anomalies Using a WGAN Model With Lipschitz Control and Progressive Kernel Critic Strategy
链接: https://arxiv.org/abs/2610.23394
作者: Najmieh Sadat Safarabadi
类目: Machine Learning (cs.LG)
*备注:
Abstract:In time-series forecasting tasks, the use of recursive forecasting on high-dimensional data substantially increases the complexity of the prediction process. The main challenge is that long-term dynamics becomes harder to model, since the sequences show greater temporal variability. To counteract this limitation and improve the modeling of long-term dynamics, and more importantly, address high-dimensional recursive forecasting, this paper uses a Wasserstein GAN equipped with a progressive Lipschitz penalty and an adaptive kernel. This model is capable of taking a window of the data stream and forecasting the next values, subsequently reconstructing the time-series stream and dynamically adapting its kernel to each new window. The proopsed model is a deep learning framework based on a Wasserstein GAN (WGAN-CLP) with an adaptive kernel to model alterations in streaming voltage time series. To establish the methodological foundation, we review the WGAN-LP formulation another variant model in this class and extend its framework through an adaptive kernel critic model. We then evaluate this extended model using power ratios, correlation coefficient (CC), and relative root mean squared error (RRMSE). The experimental results demonstrate that the proposed model outperforms benchmark methods in both accuracy and robustness. The evaluation shows promising performance of the WGAN-CLP with an adaptive kernel framework in forecasting sequence alterations across the voltage and energy time-series datasets.
[LG-35] Data-Driven Prediction of Optimal MPI Rank Placement and Node Count for Memory-Bound CFD on AMD EPYC Clusters
链接: https://arxiv.org/abs/2610.23383
作者: Marcin Lawenda,Łukasz Szustak,Aleksandra Krasicka
类目: Performance (cs.PF); Machine Learning (cs.LG)
*备注: 15 pages, 3 figures, 5 tables
Abstract:Users who run a memory-bound OpenFOAM case on an AMD EPYC cluster must decide how many nodes to use and how many MPI ranks to place on each core complex. The fastest choice is often not the cheapest one. This paper revisits our earlier analytical model of this performance and energy trade-off with 560 measured configurations of the motorBike case on five Zen 4 and Zen 5 EPYC processors, including socket power from processor telemetry. On up to four nodes, small meshes are both fastest and cheapest on a single node on most processors. For the largest mesh the largest node count is both the fastest and the cheapest option, because its speedup is superlinear. A genuine trade-off arises only for medium meshes, where full speed costs a median of 20 to 97% more node hours than the cheapest option. The L3 miss rate and the halo traffic per cell are consistent with this pattern. For a new mesh of the same case, the configuration that was optimal on the nearest measured mesh is near optimal in 75 to 85% of the decisions. Running the two configurations it recommends for speed and for cost raises this share to 85 to 90% for runtime and node hours and determines the trade-off with median errors of 3 to 5 percentage points. Neither gradient boosted trees trained on the same data nor the earlier polynomial model improves on this simple procedure.
[LG-36] Beyond Success Rate: Interventional Evaluation of Memory in Vision-Language-Action Systems
链接: https://arxiv.org/abs/2610.23372
作者: Keyuan Luo,Yi Xie,Wanyun Ling,Yuanzhe Li,Ziyue Li,Zhuoqi Zeng
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: 41 pages, including references and appendices
Abstract:When we say that a robot uses memory, does that memory actually play a role? Success-rate ablations alone cannot establish the use of task-relevant facts: they change information and computation together, while aggregate rates can hide offsetting gains and losses. We propose an Interventional Memory Evaluation Protocol (IMEP) for vision-language-action (VLA) systems to address four questions: (1) Does memory provide a net benefit? (2) Under what conditions does memory help? (3) Do changes to memory alter task outcomes? (4) Are these effects specific to task-relevant facts? At the retrieval interface, we compare target-fact substitutions with fact-preserving replacements, minimal numerical perturbations, and exact replicas, while assessing decision correctness separately from motor execution success. We evaluate within-episode, cross-episode, and agentic memory. In cross-episode learned-bank experiments, a one-unit-in-the-last-place (ULP) intervention, including its installation procedure, changes 11.9% of seed-paired outcomes, while changing key facts produces no clear additional effect over fact-preserving replacements. To examine when memory helps, we introduce ChainBench, where correct decisions require historical information unavailable in current observations. The full memory configuration raises task success from 39.1% to 74.3%, matching the oracle-decision condition under shared recorded executions. These results show that correct history can support robot decisions, but establishing whether a memory system realizes this benefit requires fact-specific evidence beyond aggregate success rates.
[LG-37] RTSM: Adapting Recursive Models to Multivariate Time-Series Classification
链接: https://arxiv.org/abs/2610.23353
作者: Georg Tenzing,Anna Tegon,Luca Benini,Thorir Mar Ingolfsson
类目: Machine Learning (cs.LG)
*备注: 33 pages, 6 figures
Abstract:Accurate multivariate time-series classification requires extracting temporal features and integrating information across a sequence. Weight-tied refinement offers a way to repeatedly process these features without allocating new parameters at each stage. Applying this principle to time series requires choosing both a temporal representation and a latent-state organization suited to classification. We introduce the Recursive Time-Series Model (RTSM), which combines a learned convolutional temporal tokenizer, single-state weight-tied refinement, and a pooled classification readout. Across all 30 UEA datasets, RTSM achieves 72.81% mean accuracy under validation-based configuration and checkpoint selection, the highest observed mean among the 17 evaluated deep-learning models and 1.60 percentage points above MambaSL. Ablations on eight datasets favor the proposed tokenizer over a simple convolutional embedding and the single-state model over the tested two-state variants. The complete classifier reaches 86.10% mean accuracy versus 84.79% for a separately trained tokenizer-only classifier. These findings identify temporal encoding and state organization as key design choices for adapting recursive architectures to time series. The resulting classifier combines competitive accuracy with a shared refinement module whose parameter count is independent of computational depth.
[LG-38] How Finite-Rank Bottlenecks Shape the Low-Rank Adaptation Landscape
链接: https://arxiv.org/abs/2610.23343
作者: Long Nguyen-Chi,Quynh Nguyen,Thanh Nguyen-Cung,Binh T. Nguyen
类目: Machine Learning (cs.LG)
*备注:
Abstract:Low-rank adaptation (LoRA) has become the standard method for parameter-efficient fine-tuning of large pretrained models, yet theoretical explanations for why low-rank updates suffice remain incomplete. Most existing analyses rely on the Neural Tangent Kernel (NTK) regime, which linearizes the network around the pretrained weights. In this work, we take a different approach that requires no linearization: our key observation is that, once the backbone is frozen, the loss sees each adapted weight matrix only through a finite-rank linear map induced by the frozen activations. Trace-norm regularization then forces the optimizer to pick a low-rank matrix among all matrices that produce the same observed output. This yields global minimizers whose ranks are controlled by the rank of this map, which we call the \emphbottleneck rank, rather than the sample size. For multi-head attention, we obtain tighter rank upper bounds by exploiting two additional invariances: row-wise softmax is insensitive to additive row constants in the score matrix, and value updates are observable only after attention weighting and frozen output projection. For the factorized LoRA objective with weight decay, we prove a complementary result on the nonconvex side: after an arbitrarily small generic positive semidefinite perturbation, every first-order stationary point is rank-deficient once the LoRA rank exceeds a threshold determined by the bottleneck rank. Finally, we show that global minimizers of the perturbed objective are near-optimal for the original trace-norm problem, and we provide uniform generalization bounds governed by the spectral structure of the frozen features.
[LG-39] SETAREH Transform: A Sorting-Assisted Haar-Wavelet Sparsification Method for Approximate Inner-Product Computation
链接: https://arxiv.org/abs/2610.23338
作者: Seyed Kamaledin Setarehdan
类目: Machine Learning (cs.LG)
*备注: 22 pages, 2 figures, 5 Tables, 21 References
Abstract:Modern deep learning systems increasingly rely on high-dimensional vector operations, in which large input and parameter vectors are repeatedly stored, transferred, and processed. Although substantial research has focused on reducing neural-network model size through pruning, quantization, knowledge distillation, and efficient architectures, the storage, memory-traffic, and communication costs associated with repeatedly presented high-dimensional input vectors remain important practical considerations, particularly in distributed and edge-computing environments. This paper proposes the SETAREH (Sorting-assisted Energy-concentrating Transform for Approximate Representation using Haar wavelets) transform, a Haar-wavelet-based sparsification approach for approximating the inner product between a fixed high-dimensional coefficient vector and successive input vectors. By applying the same sequence of permutations and Haar transforms to both vectors, SETAREH preserves the exact inner product before truncation. It then retains only a subset of the transformed coefficient-input pairs, reducing their representation cost while introducing a controllable approximation error. To evaluate SETAREH, we consider two complementary experimental settings: a synthetic Gaussian-vector baseline and a real apparent-age estimation task using a convolutional neural network on the APPA-REAL dataset. The results show that SETAREH can approximate the evaluated inner products using substantially fewer transformed coefficient-input pairs than the original vector dimension. The findings support its use as a potential structured data-representation method for repeated inner products with fixed weights, while the practical benefit depends on approximation accuracy and the cost of storing or transmitting permutation metadata.
[LG-40] Minimax-optimal Scaling Laws in Ridge Regression via Transfer Learning
链接: https://arxiv.org/abs/2610.23337
作者: Yalda Shabanzadeh,Diyuan Wu,Marco Mondelli
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Pretrained models are often reused across downstream tasks, motivating the question of whether this transfer learning procedure can provide a substantial performance improvement, namely, in terms of the scaling law. We address this question in the prototypical setting of high-dimensional regression, where (i) a pretrained \emphsource model is trained via ridge regression, and (ii) a fine-tuned \emphtarget model is regularized via the L_2 distance from the source estimator. First, we establish a deterministic equivalent for the risk of this two-step procedure under both model and covariance shift. Next, by specializing the deterministic equivalent to a shared regression function and commuting power-law covariance spectra, we identify a regime where the resulting scaling law achieves the minimax-optimal exponent. Notably, in this regime, the optimal source regularization depends on source data only, indicating its universality across all targets. Furthermore, we show that the same optimal rate is achieved by performing the fine-tuning step via one-pass stochastic gradient descent. We support our theoretical analysis via experiments on both synthetic settings and language models.
[LG-41] HiPredict: Predicting the missing HiRISE RED4 spectral channel with machine learning
链接: https://arxiv.org/abs/2610.23335
作者: Ardan Suphi,Ben Moseley,Gareth S Collins,Valentin T Bickel,Vidhya G Rangarajan
类目: Machine Learning (cs.LG)
*备注:
Abstract:The High Resolution Imaging Science Experiment (HiRISE) onboard the Mars Reconnaissance Orbiter (MRO) acquires imagery of Mars at 25–30 cm/pixel, supporting studies of martian geology, climate history, and landing-site characterization. Its 14-detector focal plane assembly includes several overlapping detectors that enable the acquisition of three-channel colour-infrared images, which are important for identifying compositional variations and studying active surface processes. In 2023, a detector in the center of the focal plane assembly (RED4) began failing intermittently, creating a gap in acquired imagery and reducing the instrument’s colour-imaging capability by half. The HiRISE processing pipeline provides a synthetic red (SYN4) channel obtained by linearly scaling the co-located infrared (IR10) channel. However, this approach cannot capture observed non-linear relationships between the colour channels, limiting its accuracy and applicability. Here, we present HiPredict, a machine-learning framework that reconstructs RED4 from co-located IR10 and blue-green (BG12) imagery, neighbouring red-channel observations, and acquisition metadata. The model employs a U-Net architecture to learn non-linear spatiotemporal-spectral relationships between these inputs and is trained on 81,717 training samples, representing 13,625 observations. Evaluated on 23,348 test samples, representing 3,892 pre-failure observations, HiPredict reconstructs RED4 with a Normalised Mean Absolute Error of 1.1%, substantially outperforming SYN4 (3.1%). The reconstructed products improve colour image recovery and more closely represent features of interest such as icy craters, stratigraphy exposed along ridge crests and chloride deposits. We release HiPredict as an open software tool. This work demonstrates the potential of machine learning to mitigate data loss as planetary missions age and naturally degrade.
[LG-42] Peak-Memory Schedules You Can Trust: Exact Realization Certified Optimality and the Limits of Recomputation for Inference
链接: https://arxiv.org/abs/2610.23328
作者: Ivan Fomin,Ivan Koryabkin,Boris Sheludko
类目: Machine Learning (cs.LG); Data Structures and Algorithms (cs.DS); Performance (cs.PF)
*备注: 24 pages, 5 figures. Preprint
Abstract:Real-time inference on small devices needs low activation memory without extra latency, which makes operator order the most valuable memory lever: it is free in time. We show that the usual way of deploying a better order, handing it back to the compiler and regenerating code, is unreliable. In PyTorch Inductor the regenerated program misses its predicted peak in 22 of 54 CPU configurations. It is worse than the stock program in 9 of them and in 19 of 50 on a GPU, and on the GPU it also changes numerical results. We instead profile the generated program, model its memory exactly from its own statements and their read/write effects, and permute those statements. We prove that this realizes the modeled peak whenever allocation does not depend on history. It does so to the byte, with bitwise-identical results, in all our CPU runs. On the GPU we characterize why the caching allocator breaks the condition. To know when to stop, we give forward and reverse “trap” lower bounds, each computable with O(n^2) minimum cuts. They certify the found order optimal for 94% of held-out wrapper graphs and all 16 microcontroller-class models we study, in under a second. So most inference graphs are already scheduled optimally. Branching networks are the exception, where reordering is free and pays, e.g. 10.0% on DeepLabV3. In compiled training we save up to 7.6%. Beyond the certified optimum we study selective recomputation with per-operator caps. It is strongly NP-hard for every fixed cap even at treewidth two, encodes knapsack exactly on series-parallel graphs, and is solvable exactly on bounded treewidth. It is useless on public tiny models whose peak is a single layer, but on architectures that hold a cheap value for a late reuse it pays: on the public ReDimNet speaker networks, whose order is already optimal, it lowers the peak by 5.2-6.1% at under 0.4% extra MACs, with bitwise-identical outputs.
[LG-43] Untrained CNN and coefficient-space regularization for joint reconstruction in photoacoustic tomography
链接: https://arxiv.org/abs/2610.23315
作者: Alfio Bozi,Anwesa Dey,Souvik Roy
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:The joint reconstruction of the initial pressure and spatially varying sound speed in photoacoustic tomography from boundary measurements is considered. A regularized data-driven framework is proposed in which the unknown fields are represented in finite-dimensional wavelet spaces and fixed nonlinear features generated by an untrained convolutional neural network (CNN) are used to reconstruct the initial-pressure coefficients through Tikhonov-regularized least squares. A learned coefficient-space prior is subsequently used to infer the squared sound-speed coefficients. Numerical experiments demonstrate accurate reconstruction on held-out data and strong generalization to previously unseen phantom geometries. The frozen reconstruction map also remains stable under measurement noise not represented during training, with only minor changes in reconstruction accuracy for noise levels up to 10% . In comparison with the supervised trained-CNN approach considered in this work, the untrained representation yields substantially smaller reconstruction errors while requiring approximately one-sixth of the computational time. These results demonstrate that fixed random convolutional features combined with regularized coefficient-space learning provide an efficient and robust approach to joint reconstruction in photoacoustic tomography.
[LG-44] Distributed Performance Boosting through Neural Control for Plug-and-Play Networks
链接: https://arxiv.org/abs/2610.23292
作者: Danilo Saccani,Leonardo Massai,Giancarlo Ferrari-Trecate
类目: ystems and Control (eess.SY); Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:We address the problem of improving the performance of pre-stabilized networked systems subject to plug-and-play (PnP) operations, namely additions or removals of subsystems. While distributed baseline controllers can preserve stability after a PnP transition, augmenting them with expressive neural policies can invalidate the formal stability guarantees. To overcome this issue, we build on recently proposed Performance Boosting controllers based on the Internal Model Control (IMC) principle, which reduce closed-loop stability to a finite \ell_2 -gain requirement on an auxiliary operator. In the distributed setting, this operator is formed by interconnecting local neural operators that exchange information only between neighboring subsystems. We then derive free parametrizations of this interconnection, namely explicit maps from unconstrained trainable variables to interconnected neural operators satisfying a prescribed network gain bound for every parameter value. The obtained distributed neural controller can therefore be trained by unconstrained gradient-based optimization without stability constraints or projection steps, while retaining the prescribed gain bound at every training iterate. We develop two complementary parametrizations. A neighbor-based compositional parametrization uses local coupling bounds, so that after a PnP event only subsystems whose bounds change need to update their gain parameters. A network-norm parametrization instead uses a bound on the overall coupling matrix and can reduce conservatism, at the price of requiring network-wide information. A PnP DC-microgrid case study shows that the proposed neural layer improves voltage transients under topology changes and load variations while preserving nominal closed-loop stability.
[LG-45] Data-Efficient Classifier Pretraining Through Diffusion Models
链接: https://arxiv.org/abs/2610.23289
作者: Qinyu Li,Yee Whye Teh,Chris Holmes,Joel Jaskari
类目: Machine Learning (cs.LG)
*备注:
Abstract:Machine learning with limited data is susceptible to overfitting, especially with modern deep neural networks. Prior to the introduction of AlexNet, reconstruction-based pretraining with auto-encoder and denoising auto-encoder were found to be beneficial, partially due to the limited size of the training sets at the time. Motivated by these traditional approaches, we propose DiffPT, a simple diffusion modeling approach to pretrain a given classifier backbone for limited domain-specific training data. We experiment on time-series data, standard natural-image benchmarks, and medical imaging datasets, and contrast the diffusion approach to a multitude of modern and traditional un/self-supervised methods. While contrastive pretraining performs better on CIFAR-10 and Tiny ImageNet in low-data regimes, DiffPT is particularly effective on medical imaging and time-series tasks, where it outperforms self-supervised baselines across a range of low-data regimes. Our results suggest that DiffPT is a simple and effective approach for data-efficient classifier pretraining on different modalities.
[LG-46] DeDiCa: A Scalable Deconfounding Discrete Causal Generative Model
链接: https://arxiv.org/abs/2610.23272
作者: Kunchangtai Liang,Adrián Javaloy,Antonio Vergari,Isabel Valera
类目: Machine Learning (cs.LG)
*备注: 32 pages, 14 figures, 7 tables
Abstract:We introduce DeDiCa, a practical framework for efficient causal inference with discrete variables and hidden confounding that revisits classical probabilistic modeling through a tensorized parameterization of causal Bayesian networks, enabling tractable inference. Given a causal graph, DeDiCa is trained once on observational data by maximum likelihood, exactly marginalizing over hidden confounders, and can then estimate any (proxy-)identifiable interventional query. Despite this deliberately simple parameterization, DeDiCa matches state-of-the-art neural causal generative models in ATE estimation, while supporting memory-efficient CPU training and exact interventional inference in milliseconds. Compared with causal foundation models, DeDiCa scales more effectively with additional observational samples and achieves competitive accuracy at substantially lower inference time and memory. DeDiCa provides a tractable, accurate, and scalable approach to causal inference in discrete settings with hidden confounding.
[LG-47] NestedVLA: Learning to Consolidate and Generate Skills for Vision-Language-Action Models NEURIPS2026
链接: https://arxiv.org/abs/2610.23271
作者: Guodong Du,Wanyu Lin
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: Accepted at NeurIPS 2026. 23 pages, 3 figures. Code: this https URL
Abstract:Vision–Language–Action (VLA) models promise generalist robotic agents, yet consolidating skills learned across heterogeneous tasks and environments remains difficult: training data is fragmented across platforms, joint multi-task training suffers from interference, and model merging tends to introduce parameter conflicts or require architectural changes. We instead advocate a paradigm in which skills are first learned independently and then consolidated post hoc. To this end, we propose NestedVLA, which replaces parameter aggregation with parameter generation: a nested hypernetwork, conditioned on the target task, synthesizes task-adaptive action-expert parameters that compose with a frozen base VLA. Task-specific compressed VLM residuals are retained separately. This formulation enables flexible task-conditioned skill reuse while mitigating interference. Across three RL benchmarks (MetaWorld, ManiSkill, CALVIN), NestedVLA outperforms the strongest model-merging baseline by 41 pp on average, matches per-task fine-tuning within 1 pp on ManiSkill, and substantially closes the gap on CALVIN, pointing toward a storage-efficient route to skill consolidation. Code is available at this https URL.
[LG-48] ADT: All-Atom Diffusion Tokenizer for 3D Molecular Structures
链接: https://arxiv.org/abs/2610.23265
作者: Xuantuo Huang,Encheng Su,Aoran Wang,Pan Tan,Shengyong Xu,Yizhou Wang,Chen Tang,Shixiang Tang,Mingxiao Li
类目: Machine Learning (cs.LG)
*备注:
Abstract:Discrete structural tokenization provides a promising route to autoregressive 3D molecular generation. Yet, existing tokenizers face a trade-off between compact representation and all-atom reconstruction: atom-level tokens preserve atomic details but produce long sequences, while compact residue-level protein tokens typically omit side-chain geometry. To overcome this limitation, we propose All-atom Diffusion Tokenizer (ADT), which decouples token granularity from reconstruction resolution. ADT uses one structural token per atom for small molecules and one token per protein residue that jointly represents backbone and side-chain geometry. To mitigate geometric distortion introduced by quantization, we leverage the generative capacity of diffusion models to recover fine-grained atomic geometry beyond deterministic decoding from discrete tokens. Extensive experiments across small molecules and proteins demonstrate that ADT achieves strong reconstruction from compact discrete representations while enabling high-quality autoregressive generation of both small molecules and all-atom protein structures. These results establish compact structural tokens as an effective discrete interface for all-atom molecular reconstruction and generation.
[LG-49] FLARE: Physics-Grounded Evaluation of Language-Model Agents for Feeder Restoration with Degraded Observations
链接: https://arxiv.org/abs/2610.23251
作者: BHVSP Subrahmanyam,Neeraj Choudhary
类目: Machine Learning (cs.LG)
*备注:
Abstract:Restoring service after distribution-grid faults requires decisions from incomplete and conflicting information while respecting electrical constraints. Addressing this need, we introduce FLARE (Feeder-Level Agent Restoration Evaluation), a physics-grounded benchmark for post-isolation feeder restoration. FLARE con- tains 800 AC-feasible cases across five observation settings and allows agents to gather evidence, evaluate switches, and perform restoration actions under fixed operational limits. We evaluate five instruction-tuned models over 78,750 episodes, including topology, distributed-generation, and loading shifts. Mean normalized regret ranges from 0.082 to 0.148. Although every model has zero median regret, P95 regret ranges from 0.814 to 1.000, revealing an important failure tail. Shadow replay shows that 96.4% of 15,706 guard-rejected actions would have caused a physical violation. FLARE enables reproducible evaluation of language-model based restoration support while emphasizing the need for external safety checks.
[LG-50] Anisotropic Representation Dynamics in Transformers
链接: https://arxiv.org/abs/2610.23242
作者: Nguyen Do,My T. Thai
类目: Machine Learning (cs.LG)
*备注: 62 pages
Abstract:Small differences between internal representations can shrink or grow as they pass through a Transformer layer, affecting how information is preserved and how sensitive the model is to perturbations. Yet it remains unclear when changes in attention weights can turn a representation difference that would otherwise shrink into one that grows. We develop a \textbfMixture-Gated Theory of Anisotropic Representation Dynamics that separates self-attention into value transport, query gating, and key gating. Our analysis reveals how the geometry of attended representations constrains gating and how interactions between these mechanisms can reverse contraction induced by value transport. We formalize this phenomenon as gating-induced regime switching, where a representation difference contracts under fixed attention routing but expands under full attention. We derive a necessary and sufficient spectral criterion that certifies the existence or absence of switching and identifies switching directions without exhaustive search. Beyond existence, our theory characterizes maximum relative switching strength, certifies the severity of contraction that gating can overcome, and determines the largest dimension of a subspace in which every nonzero direction switches. Experiments across four Transformer architectures validate these predictions through analytical and finite-forward comparisons, revealing distinct and joint query–key gating mechanisms and nontrivial switching subspaces. Our results provide a mechanistic and spectral understanding of how attention routing can alter local representation dynamics.
[LG-51] Response-Informed Value Estimation for Cost-Aware Multi-Turn LLM Routing
链接: https://arxiv.org/abs/2610.23239
作者: Haotian Yan,Jiaqi Liu,Bingyi Jing,Tianjun Yao
类目: Machine Learning (cs.LG)
*备注:
Abstract:Multi-turn LLM routing selects a model at every turn before its response is available, yet task success is revealed only when the interaction ends, leaving each intermediate choice without a label of its own. Logged trajectories, however, retain the response each call produced, which exposes what each turn required and how well the chosen model served it, evidence that outcomes alone hide. We introduce RIVET (Response-Informed Value Estimation for Turn-Level Routing), which exploits this post-call evidence during training while keeping routing strictly pre-call at deployment. RIVET jointly trains two value estimators over a shared state-model representation: a deployable Routing Value Estimator that scores candidate models from pre-call information, and a training-only Response-Informed Value Estimator that additionally observes the realized response. Conditioning on the response turns response-induced return variation from residual noise into an observed covariate, and we show that this reduces the leading asymptotic estimation error of the shared representation. On the MINT benchmark, RIVET solves more held-out tasks than the strongest multi-turn baseline while using 74.7% less total call cost.
[LG-52] Process Reward Modeling with Reliability- and Conservatism-Aware Bayesian Belief
链接: https://arxiv.org/abs/2610.23223
作者: Zhongjian Qiao,Zhixiang Zhou,Jiafei Lyu,Peisong Wang,Qiaosheng Zhang,Siyang Gao,Shuang Qiu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Process Reward Models (PRMs) play a central role in test-time scaling for large language model (LLM) reasoning. However, PRMs may provide inaccurate process rewards for test-time scaling, which we identify as arising from two sources of uncertainty: (1) \textitPRM uncertainty that arises from insufficient supervision data and imperfect PRM training; and (2) \textitpolicy uncertainty that results from the mismatch between the policy to collect supervision data and the policy for test-time scaling, leading to systematic reward overestimation. Previous studies fail to jointly consider both uncertainties. To bridge this gap, we propose BayesianPRM, which treats the process reward as a random variable within a reward hypothesis space and maintains a Bayesian belief over it. Since exact Bayesian inference over the continuous hypothesis space is generally intractable, we discretize the reward hypothesis space with a reward ensemble and introduce two complementary belief updates, each targeting one source of uncertainty: (1) to address PRM uncertainty, we introduce \textitreliability-aware belief updating that measures how well each ensemble member explains the supervision data and updates the belief with an amortized belief network optimized by the evidence lower bound (ELBO); and (2) to tackle policy uncertainty, we propose \textitconservatism-aware belief calibration which further reweights the belief toward lower rewards to suppress overestimation. These two updates yield a reliability- and conservatism-aware Bayesian belief, whose expectation determines the process reward. Experiments on multimodal reasoning benchmarks indicate that BayesianPRM consistently yields more accurate process reward estimates while enhancing test-time scaling performance.
[LG-53] Non-Parametric Optimization for Scalable Learning in Stochastic Decision Problems
链接: https://arxiv.org/abs/2610.23204
作者: Mohsen Amidzadeh,Lauri Viitasaari,Mario Di Francesco
类目: Machine Learning (cs.LG)
*备注:
Abstract:Stochastic optimization (SO) plays a central role in decision-making under uncertainty. Among SO problems, time-varying stochastic optimization (TV-SO) is particularly important due to its applications in adaptive control and machine learning. Non-parametric approaches have been proposed for time-varying deterministic optimization; however, they have not been developed for their stochastic counterparts. This work addresses that gap by developing a stochastic variational framework based on Malliavin calculus. This framework yields non-parametric optimality conditions for SO problems with stochastic decisions and supports the design of a scalable deep-learning algorithm that is insensitive to the parameterization dimension. This algorithm, called the Stochastic Path Follower (SPF), is applied to two important problems under distribution drift, namely least-squares recovery and logistic regression. Experimental results show that the proposed approach outperforms state-of-the-art learning-based and gradient-based methods in both performance and scalability.
[LG-54] Proportionality in Ranking Compression NEURIPS
链接: https://arxiv.org/abs/2610.23193
作者: Daniel Halpern,Xinyu Liu,Evi Micha,Yu Peng Ng,Warut Suksompong
类目: Computer Science and Game Theory (cs.GT); Machine Learning (cs.LG); Theoretical Economics (econ.TH)
*备注: Appears in the 40th Annual Conference on Neural Information Processing Systems (NeurIPS), 2026
Abstract:Motivated by applications in reinforcement learning from human feedback, we consider a setting where a given collection of rankings needs to be compressed into a smaller collection of rankings. This can also be seen as a combination of two fundamental social choice frameworks: ranking aggregation and multiwinner voting. We propose three proportionality notions that capture different types of representation in this setting: positional proportionality, pairwise proportionality, and proportionality for solid coalitions (PSC). On the one hand, we show that positional proportionality and PSC are always satisfiable for any input rankings and target compression size, and a desired output can be found in polynomial time. On the other hand, we prove that pairwise proportionality cannot be satisfied in general, but can nevertheless be attained when the input rankings are single-peaked or single-crossing.
[LG-55] Structural Affinity: An Algorithm for Coherent Cage-Based Shape Design
链接: https://arxiv.org/abs/2610.23173
作者: Aviraj Newatia,Shengan Yang,Hantao Zhong,Rika Antonova
类目: Machine Learning (cs.LG); Graphics (cs.GR); Robotics (cs.RO)
*备注: 27 pages, 16 figures
Abstract:Cage-based deformation (CBD) parametrises 3D geometry through a small set of control handles, effectively enabling a variety of tasks from animation and interactive modelling to tool-based co-design, neural deformation transfer, and aerodynamic shape optimization. Many optimization approaches relying on CBD employ regularisation strategies to reduce the distortion of the deformed mesh by encouraging smooth, regular cage deformations. These techniques offer soft guidance to an optimization algorithm towards desirable deformations, but compete with the task objective rather than changing the deformation parameterization itself. In this work we introduce Structural Affinity for CBD, an explicit regularisation method which enables smooth deformations through control cages by exploiting a two-cage structure to couple cage-handle motion directly. Structural Affinity precomputes an affinity matrix using mean-value coordinates, binding the inner control-cage to a coarser outer-cage. This matrix defines how much a cage handle displacement propagates to neighbouring handles which substantially decreases and often eliminates cage self-intersections, reduces mesh distortion, and adds linear overhead at deformation time. Across robotic control–morphology co-design, source-to-target neural deformation transfer, and aerodynamic wing shape optimization, Structural Affinity produces more regular deformations and substantially reduces reliance on hand-designed structural loss terms. We report consistent gains on three complementary diagnostics: the cotangent Laplacian, a proxy for surface smoothness; cage distortion, capturing whether control handles move coherently; and surface roughness, which measures undesirable oscillation in wing geometry.
[LG-56] Nearly Linear Private Learning of Littlestone Classes
链接: https://arxiv.org/abs/2610.23160
作者: Shay Moran,Chao Yan
类目: Machine Learning (cs.LG)
*备注:
Abstract:Differentially private learning aims to learn from data while protecting individual training examples. For binary classification, finite Littlestone dimension characterizes approximate private PAC learnability (Alon et al., JACM 2022). Understanding how this dimension controls private sample complexity remains a central question. For classes of Littlestone dimension d , the first general upper bound was doubly exponential in d (Bun, Livni, and Moran, FOCS 2020). Subsequent work improved this to \widetilde O(d^6) (Ghazi et al., STOC 2021) and then to \widetilde O(d^5) (Lyu, STOC 2026). We prove a nearly linear upper bound of O(d\log d) . Our main algorithmic contribution is a structural exponential mechanism for data-dependent candidate sets equipped with a rank satisfying suitable stability conditions. For natural Littlestone classes such as subspace indicators, this rank has a simple geometric interpretation: it is the codimension. This yields a private PAC learner for subspaces of dimension at most d over any field, with tight linear dependence on d and no dependence on the ambient dimension, improving the quadratic dependence of Kaplan et al. (STOC 2025). Our second contribution is combinatorial: drawing on model-theoretic ideas, we construct a suitable rank for general Littlestone classes and establish the counting bounds required by the mechanism. Subjects: Machine Learning (cs.LG) Cite as: arXiv:2610.23160 [cs.LG] (or arXiv:2610.23160v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2610.23160 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-57] Grounded Credit Assignment in Reinforcement Learning for Multi-Turn Search Agents
链接: https://arxiv.org/abs/2610.23148
作者: Qirui Liu,Hao Chen,Yi Wen,Weijie Shi,Yao Zhao,Xianhe Chen,Hao Zhao,Xing Tang,Weihong Luo,Yuyu Luo
类目: Machine Learning (cs.LG)
*备注: 27 pages
Abstract:Outcome-reward reinforcement learning such as GRPO gives every token of a search-agent rollout the same advantage, although the rollout exercises two capabilities: finding evidence and using it. We show that accuracy improves without grounding: Search-R1 raises accuracy on all seven QA benchmarks, yet on Bamboogle the evidence-supported share of correct answers falls from 84.0% to 42.9%. We attribute this to GRPO’s credit assignment: rollouts that retrieve answer-bearing evidence are favoured in aggregate only in proportion to their accuracy gain over the rest, a grounding gap that answering from memory or misusing evidence shrinks or reverses. To address this, we propose Grounded Credit Assignment (GCA), which uses verifiable sub-outcomes (evidence retrieval and answer correctness) to correct search and answer credit separately: a wrong answer is blamed more than the search that found the evidence, and a correct answer that bypasses retrievable evidence is credited less. Never negative on evidence-finding search, the correction does not shrink with the grounding gap; capped at 10% of each token’s advantage, it never overrides the outcome signal. Without extra models, rollouts, or training stages, GCA improves average EM over Search-R1 by 5.7 (Qwen2.5-3B) and 6.3 points (Qwen3-30B-A3B) and raises this share from 76.9% to 93.4%.
[LG-58] MetaCurate: Evolving Curation Filters for High-Quality Agent ic Coding Data
链接: https://arxiv.org/abs/2610.23134
作者: Donghoon Kim,Seunghyuk Oh,Kevin Galim,Minseo Kim,Minjae Lee,Younghyun Kim,Wooseok Seo,Hyung Il Koo,Wonjun Kang
类目: Machine Learning (cs.LG)
*备注: Project page: this https URL
Abstract:Supervised fine-tuning on agent trajectories is an effective way to improve coding agents, but selecting useful training examples requires more than identifying successful runs. A resolved patch may follow inefficient reasoning or unnecessary tool use, while a failed run may still contain useful intermediate steps. Training pipelines therefore rely on rules and qualitative assessments to filter trajectories, yet designing these filters typically requires substantial human expertise, and validating improvements through repeated training and evaluation is expensive. We introduce MetaCurate, a training-free framework that evolves executable trajectory curation filters using pairwise LLM judgments. Disagreements between the filter’s rankings and the judge’s preferences provide targeted feedback for jointly revising its rules, rubric, and aggregation policy. The resulting filter scores trajectories independently, so pairwise comparisons are required only during search. Starting from a blank filter, MetaCurate autonomously discovers a curation strategy that achieves competitive downstream performance without task-specific handwritten criteria. We validate downstream utility through controlled SFT experiments on our uncurated trajectory pool and the Nemotron v3 SWE dataset, with random selection and hand-authored filtering as baselines. In our experiments, MetaCurate achieves the highest mean SWE-Verified score among the compared methods, including hand-authored rule-based filtering and a training-loop baseline whose search costs 3.5x as much.
[LG-59] owards a GraphNARX model for structural dynamics
链接: https://arxiv.org/abs/2610.23103
作者: George Tsialiamanis,Keith Worden
类目: Machine Learning (cs.LG)
*备注:
Abstract:The current paper introduces the GraphNARX model, a graph-based nonlinear autoregressive exogenous model for structural dynamics. The model is built to account for the geometry of structures and to model dynamic behaviour as the interaction between elements of the structure. Similar to other types of NARX models, GraphNARX is used for system identification. Nodal attributes of the graph encode past displacements and nodal forces, and the output of the model is the next-timestep displacement. The graph formulation of the model allows it to generalise to structures of different dimensions and degrees of freedom. Moreover, because graph-learning is performed locally between the nodes, the GraphNARX model can also learn to infer displacements of nodes for which data are not available, i.e.\ to perform virtual sensing. The proposed model’s performance is evaluated in a system-identification framework and a virtual-sensing framework, and achieves good accuracy and exhibits generalisation capabilities to structures of different sizes.
[LG-60] Where Latents Belong and How They Flow: Joint Identification of Structural Roles
链接: https://arxiv.org/abs/2610.23074
作者: Xintong Li,Kun Zhang,Yujia Zheng
类目: Machine Learning (cs.LG)
*备注:
Abstract:Latent factors in dynamical systems can play different roles across observations and time. For example, an idea in a group meeting may be expressed by only some participants, inherit information from earlier discussions, and shape what is said next. We formalize this notion as the structural role of a latent factor, describing where it is expressed, whether it receives influence from the past, and whether it transmits influence to the future. The main challenge is correspondence, since the three structures may be identified under different permutations, while a structural role is defined only when they are aligned to the same latent coordinate. We show that, under suitable structural and variability conditions, all three roles are jointly identifiable under a common permutation at each time point. Guided by this theory, we develop RoleTrace, a framework that traces structural roles in a shared latent coordinate system. Synthetic experiments validate structural recovery, while experiments on multiparty meetings examine the identified roles in real sequential interactions.
[LG-61] DIFFLET: A Fast Inference Engine for Diffusion Transformer Models on AWS Trainium
链接: https://arxiv.org/abs/2610.23047
作者: Bin Ma,Min Yu,Vijay Kodamalla,Jialin Liu,Dong Li
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注:
Abstract:On AWS Trainium2, every component of a diffusion model runs as a program compiled ahead of time for fixed shapes, while pipelines differ across models, prompt lengths across requests, and caching decisions across denoising steps. We present Difflet, a serving engine in which five model pipelines share one compilation backend, joint attention is split across cores by attention head, and a host controller decides at each denoising step whether to run the denoiser or reuse its earlier output. A small probe program computes the caching signal on the device and returns one number per step, and a resident model process keeps every program loaded across requests. Across five image and video model families, fused attention accelerates denoiser steps by 1.2-4.3x over the same framework using PyTorch/XLA attention, and at 32.8k visual tokens it is the only attention that compiles. On four of the five models, resident serving is 1.5-3.4x faster than standalone runs that reload weights from a warm file cache for each generation.
[LG-62] GR-OPD: Group-Relative Sequence Credit for On-Policy Distillation
链接: https://arxiv.org/abs/2610.23042
作者: Yushu Wang,Renkai Yang,Jingcao Xu,Zili Geng,Haihong Tang,Yaohui Jin,Hao He
类目: Machine Learning (cs.LG)
*备注:
Abstract:On-policy distillation (OPD) provides token-level teacher feedback, yet common sampled-token updates assign credit locally and omit the effects of earlier actions on subsequent feedback. We introduce \method (Group-Relative On-Policy Distillation), which decouples feedback computation from credit assignment by aggregating token feedback into calibrated sequence-level credit. We establish a theoretical basis for this approach by showing that, under ideal on-policy assumptions, broadcasting raw summed feedback to every response token yields an unbiased estimate of the negative gradient of the sequence-level reverse-KL objective. To support effective learning in practice, \method discounts later feedback that may become less reliable under distribution shift and uses same-prompt rollout groups to calibrate teacher guidance. It preserves relative response preferences while retaining a controlled fraction of the group mean, which we term collective teacher pressure. Across five Qwen3 teacher–student settings and six mathematical reasoning benchmarks, \method outperforms vanilla OPD in macro Avg@K in every setting, including gains of 10.9 and 11.5 percentage points in two settings with matched thinking patterns. It also maintains stable training under mismatched thinking patterns, where vanilla OPD collapses. Ablations show that retaining an appropriate amount of collective teacher pressure can improve learning, while excessive collective teacher pressure or an unrestricted feedback horizon can destabilize training.
[LG-63] owards Looped Models Done Right Part I: Topology Input Injection Recurrent-State Design
链接: https://arxiv.org/abs/2610.23036
作者: Benhao Huang,Chufan Shi,Junlin Chen,Shicheng Wen,Zhengzhong Liu,Eric Xing,Xuezhe Ma
类目: Machine Learning (cs.LG)
*备注: Code and checkpoints: this https URL
Abstract:Looped language models reuse physical Transformer blocks across logical depth, but established architectures entangle recurrence placement, input injection, and recurrent state organization. We isolate these axes in models trained from scratch at matched parameter scale, logical depth, and token budgets, and then test whether the resulting design principles transfer to resident- and active-parameter-matched Mixture-of-Experts (MoE) models. The prelude-recurrent block-coda sandwich architecture consistently improves performance on context extraction and reasoning-intensive tasks. Input injection primarily enhances knowledge retrieval and context utilization, whereas random initialization and separate H/L recurrent states provide mixed or negative returns. Taken together, these results favor the Huginn-style design: at the 730M-parameter and 336B-token scale, the complete Huginn architecture outperforms Ouro across all ten benchmarks. The advantages of Huginn over Ouro transfer to MoE models. At the 8BA0.8B (32BA3B when unrolled) scale trained on 500B tokens, Huginn outperforms Ouro on nearly all benchmarks spanning knowledge, reasoning, and coding, while also exhibiting more balanced expert routing. Under matched training and inference FLOPs, the Huginn-style model uses 75% fewer resident parameters than a 112-layer feedforward MoE baseline. Despite its smaller physical parameters, it outperforms the feedforward baseline on DROP and GSM8K, matches it on MATH500, and remains competitive on the other benchmarks.
[LG-64] “Deep lifts”: Neural networks can generalize even with the dumbest optimizer
链接: https://arxiv.org/abs/2610.23009
作者: Sunil Madhow,Yu-Xiang Wang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Given a labeled dataset (x_i, y_i)_i = 1^n , consider drawing a parameter vector \hat\theta uniformly at random from those that interpolate the data within a width- W depth- D ReLU architecture with continuous weights bounded within [-M, M] . We study the generalization properties of \hat\theta in the case where the dataset can, in fact, be explained by a smaller w -by- d ReLU network. Despite the obliviousness of the selection rule for \hat\theta , it adaptively generalizes at the rate \tilde O\left(\fracwdWDn\right) , rather than being governed by the VC dimension of the full W -by- D hypothesis class. The reason is due to a particular inductive bias in ReLU networks: the w -by- d parameter witnesses an exponential volume of W -by- D interpolating parameters, which dwarfs the volume of poorly generalizing parameters (conditioned on interpolation). We explicitly construct this set of near-equivalent W -by- D parameters by a proof technique called deep lifting, and then tie the achievable volume to stability properties of the w -by- d witness. Examples show that these techniques allow us to convert approximation-theoretic results into generalization guarantees for nonparametric classes. We further show that generalization of uniform-random interpolators can survive label corruptions.
[LG-65] A Minimal ReLU Classifier on Non-Linearly Separable Data: Convergence and Implicit Bias at and beyond the Edge of Stability
链接: https://arxiv.org/abs/2610.23007
作者: Quan Nguyen,Robin Yadav,Param Mody,Mark Schmidt
类目: Machine Learning (cs.LG)
*备注:
Abstract:Gradient descent (GD) with large learning rates can successfully optimize neural networks even though classical smoothness-based learning rate conditions are violated. Yet theory for this regime largely covers linear models or linearly separable data. We analyze a minimal setting outside both regimes. Specifically, a two-neuron ReLU network trained with logistic loss on 1D non-linearly separable data. The objective is nonconvex and nonsmooth, and GD exhibits edge-of-stability (EoS) behavior. We identify an active cone in parameter space that contains every perfect classifier, where the problem becomes piecewise logistic regression with a varying effective dataset. Within this cone, we prove a finite-time loss bound for constant learning rates and show that GD converges in direction to an explicit max-margin vector, while the orthogonal component converges to the unique minimizer of a non-strongly convex potential. We further establish catapult results where exponentially small loss is reached after a single loss spike. Furthermore, we propose an adaptive schedule that lets learning rates grow to infinity without leaving the active cone. We prove that this schedule achieves a loss rate of O(1/(T\logT) , which is faster than the standard O(1/T) rate. Our theoretical results extend to local convergence on more general network and are validated on MNIST dataset. Overall, we show that this setup provides a tractable neural-network model for studying large and adaptive step sizes beyond linear predictors and separable data.
[LG-66] ALLOP: Joint Optimization of Local Training and Communication Compression for Time-Constrained Federated Learning
链接: https://arxiv.org/abs/2610.22996
作者: Honoka Anada,Yafei Ou,Shinya Takamaeda-Yamazaki
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
*备注: Accepted for publication in The 32nd IEEE International Conference on Parallel and Distributed Systems (ICPADS 2026)
Abstract:Minimizing the wall-clock time required to reach a target accuracy is crucial in federated learning (FL), where many applications require frequent model updates as data distributions evolve over time. Nevertheless, in heterogeneous environments, slow clients with limited computational and communication capabilities can delay global synchronization, forcing faster clients to wait and increasing the overall training time. To mitigate this straggler effect, recent studies have introduced per-round time constraints and adapted client workloads within the available time. However, these approaches primarily focus on local computation, whereas communication can become a major bottleneck in communication-limited environments, motivating the joint optimization of training workload and communication compression. Although previous studies have explored such joint optimization, determining an effective training–communication balance under a fixed per-round time budget remains challenging, as the resulting model performance depends on both training progress and compression-induced degradation. We therefore propose TALLOP, which jointly optimizes the local training workload and communication compression under per-round time constraints. TALLOP predicts the training loss achievable under different computation budgets from the early-stage mini-batch loss trajectory (LENS), and estimates the degradation caused by communication compression using a compression-error model (CLEN). TALLOP combines these predictions to select the training-communication allocation that minimizes the post-compression loss under the given round time constraint. Experiments in heterogeneous FL environments show that TALLOP generally achieves competitive or improved time-to-accuracy compared with approaches that optimize only local training or neither, although the benefit over training-only adaptation depends on the setting.
[LG-67] owards Hyperparameter Transfer for Differentially Private Optimization NEURIPS2026
链接: https://arxiv.org/abs/2610.22976
作者: Tianze Wang,Zhiqi Bu,Linjun Zhang
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR); Machine Learning (stat.ML)
*备注: Accepted at NeurIPS 2026
Abstract:Scaling laws and hyperparameter tuning remain central challenges in training large-scale models, particularly under differential privacy (DP), where optimization is further complicated by noise injection and per-sample gradient clipping. While Maximal Update Parametrization ( \mu P) allows for zero-shot hyperparameter transfer in standard training, we show that it fails under DP constraints due to the distinct spectral properties of high-dimensional noise versus low-rank gradients. In this work, we introduce DP- \mu P, a theoretically grounded framework that extends \mu P to the privacy-preserving setting. By analyzing the interaction between gradient clipping, noise injection, and feature learning, we derive a Signal-to-Noise Ratio (SNR) condition that governs layer-wise learning rate scaling. Combining this with our proposed spectral scaling formulation, we provide specific scaling rules for DP-SGD, DP-Adam, and DP-Muon optimizers. Empirically, we demonstrate that DP- \mu P enables effective zero-shot transfer of learning rates across model widths on diverse architectures (MLP, ViT, GPT-2), matching or exceeding the performance of individually tuned baselines while significantly reducing computational costs.
[LG-68] On the tradeoff between robustness and geometric accuracy in adversarial training
链接: https://arxiv.org/abs/2610.22955
作者: Rachel Morris,Ryan Murray,Kerrek Stinson
类目: Machine Learning (cs.LG); Analysis of PDEs (math.AP)
*备注:
Abstract:For non-parametric binary classification in the presence of an adversary, we prove an optimal quantitative estimate for adversarial classifiers controlling geometric accuracy, as measured by the Hausdorff distance to the Bayes classifier, in terms or their robustness, as encoded by the adversarial budget \varepsilon . This rigorously shows that robust classifiers sacrifice at most O(\sqrt\varepsilon) or O(\varepsilon) in accuracy, depending on the regularity of the underlying Bayes classifier. In contrast to prior work on the incompatibility of robustness and accuracy over low-complexity hypothesis classes, our work shows that high-complexity hypothesis classes allow for robust classifiers with precisely quantified accuracy. Our results apply for adversaries determined by any norm on Euclidean space and in both the infinite and finite data settings.
[LG-69] Large Diffusion Models Need More Than EMA
链接: https://arxiv.org/abs/2610.22945
作者: Chuyu Hu,Huanran Chen,Shitong Shao,Yinpeng Dong
类目: Machine Learning (cs.LG)
*备注:
Abstract:Large-scale diffusion models are commonly trained with a constant learning rate (LR) and exponential moving average (EMA), despite the strong performance of LR decay and Schedule-Free optimization in training large language models. We study whether EMA is uniquely suited to diffusion training, or whether stronger alternatives have been underexplored. Our theory shows that averaging a constant-step trajectory has an inherent optimization limitation, whereas LR decay and Schedule-Free optimization admit vanishing stationarity guarantees under standard nonconvex assumptions. Empirically, LR decay and EMA provide complementary benefits: combining decay with EMA further improves the standard baseline at most tested starting points, although the gain remains sensitive to when decay begins. Schedule-Free avoids this timing requirement and consistently outperforms EMA with a constant LR across training stages, achieving both lower fixed-evaluation objectives and better generation quality. Our analysis further shows that all three strategies substantially reduce parameter variability, yet stronger stabilization is not always better: EMA with a constant LR retains comparatively higher parameter variability, while aggressive decay combined with EMA achieves the lowest checkpoint variability but also weakens continued optimization. Schedule-Free substantially reduces parameter variability while preserving more late-stage progress, providing a better balance between parameter stability and continued optimization. Finally, the benefits of Schedule-Free extend beyond pretraining to ReFL and generator-side DMD, demonstrating its transferability across distinct diffusion optimization settings. Overall, our results suggest that large diffusion models need more than parameter averaging alone: effective optimization requires balancing variance reduction with continued optimization.
[LG-70] Scalable Levenberg-Marquardt Training Framework for Deep Learning
链接: https://arxiv.org/abs/2610.22925
作者: Xianghao Wang,Johan Barthélemy,John Taylor
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
*备注:
Abstract:The Levenberg-Marquardt (LMA) algorithm is a widely used nonlinear least-squares solver known for its rapid convergence. For small-scale neural networks, LMA provides better convergence and architectural robustness than first-order optimizers like Adam. However, LMA imposes prohibitive memory and computational costs. These costs are primarily from the Jacobian matrix, whose dimensions scale with both the number of samples and parameters. This bottleneck prevents its use in large-scale neural networks. Recent efforts scale LMA using preconditioned conjugate gradient (PCG) solvers to avoid materializing the Jacobian. However, PCG only approximates the solution, requiring many network propagations and a task-specific preconditioner. Other studies optimize LMA to reduce costs, but they restrict their methods to specific tasks or network architectures. As a result, LMA remains impractical for training general large-scale neural networks. In this paper, we propose LeMA, a distributed LMA algorithm for large-scale neural network training on multi-GPU platforms. Our contributions are twofold: First, an exact dual formulation of LMA removes the memory dependence on the parameter dimension. Second, a Jacobian sharding strategy computes Jacobian shards on the fly and in parallel to avoid materializing the full Jacobian. For large-scale neural networks, LeMA reduces the memory footprint of LMA from quadratic in the number of parameters to quadratic in the batch size. This makes exact LMA steps feasible for billion-parameter models without approximate PCG solves. Furthermore, for a convolutional neural network with 1,000,022,632 parameters on an 8-GPU node, LeMA achieves a 7.97x speedup over a single GPU, corresponding to a 99.7% strong scaling efficiency.
[LG-71] A Feature Selection Layer Approach for Multi-Class Classification on Neural Networks
链接: https://arxiv.org/abs/2610.22898
作者: João Pedro Silveira e Silva,Márcio Dorn,Bruno Iochins Grisci
类目: Machine Learning (cs.LG)
*备注: 75 pages (including Appendix with Supplementary Material). Published in Information Sciences
Abstract:In the age of data-driven decision-making, deep learning models have become powerful tools for uncovering patterns and making predictions that influence many aspects of society. However, their broader adoption in scientific and business applications is often limited by a lack of interpretability. Feature weighting methods, which assess the relevance of individual features, are among the most widely studied approaches for improving interpretability. While post hoc techniques have advanced significantly, especially in modeling nuanced relationships between features and labels in multiclass problems, embedded methods typically fall short in this regard. Embedded techniques integrate feature relevance learning directly into the model training process, but they are often unable to capture label-specific relationships. To address the limitation, we propose multiple extensions to the Feature Selection Layer, an embedded feature weighting mechanism for neural networks. Through extensive experimentation, we identify the most effective extension and benchmark it against established feature weighting methods. Our results show that the proposed method matches existing techniques in identifying globally relevant features and outperforms them in capturing label-specific relevance. By leveraging the internal dynamics of the trained neural network, it reveals how individual features contribute to the prediction of specific labels, requiring no post-processing beyond the training process.
[LG-72] What Accuracy Hides In Learned Transformer Programs
链接: https://arxiv.org/abs/2610.22891
作者: Jonathan Lee,Narmeen Oozeer,Philip Quirke
类目: Machine Learning (cs.LG)
*备注:
Abstract:Transformer Programs (TP) (Friedman et al., 2023) constrain a Transformer’s architecture so that its learned computation can be extracted as readable Python code. TP expose learned computations, but token accuracy can hide incorrect complete answers. In this work, we examine how program inspection provides insights beyond accuracy scores and helps guide training. We first evaluated TP on nine algorithmic tasks. We compared token accuracy with accuracy on complete output sequences and checked for failures missed by sampled tests. Further, we assessed training data improvements informed by program inspection and failure analysis, finding gains in sequence accuracy on several tasks. We examined a counting task where a program achieved 100% accuracy on sampled test sequences but failed on another input using the symbols and lengths allowed during training. The program counted correctly, but the required answer was missing from its output labels. After adding training examples with larger counts, the retrained models passed exhaustive evaluation over the bounded domain. In the frequency ranking task, inspection showed how correct counts contributed to incorrect answers in programs trained before and after changes to the training data. Finally, we studied constraints imposed by TP training and operations. The TP training process can lead to accuracy losses late in training. Our reference constructions also show how composing the restricted TP operations can use many layers or large lookup tables.
[LG-73] Benchmarking and Enhancing Agent ic Causal Discovery
链接: https://arxiv.org/abs/2610.22846
作者: Pingchuan Ma,Rui Ding,Shuai Wang
类目: Machine Learning (cs.LG)
*备注:
Abstract:LLM agents can now run causal discovery algorithms on a dataset and report what comes out. Whether they do more than that, such as noticing where an algorithm went wrong and fixing it with further tests, is hard to tell from current benchmarks. Most of them give the variables meaningful names, and an LLM shown only the names of the Tübingen cause-effect pairs already gets 99% of the directions right. With the names removed, the data are usually clean enough that picking a better algorithm than PC closes most of the gap. We introduce Heuresis, a benchmark of anonymous simulated datasets whose mechanisms range from linear Gaussian to heavy-tailed. For each family of datasets it separates what a better existing algorithm can gain from what only a closer look at the data can gain, and it traces every edge an agent changes. Every family leaves a gap that none of the algorithms, the router and the two causal-discovery foundation models we ran can close, so a score above the best chosen algorithm has to come from using the data. We instantiate the benchmark with an agent harness and three LLMs, GPT-6.1 Sol, Claude Opus 5.5 and Qwen 3.8 Flash. GPT-6.1 Sol surpasses the best chosen algorithm on all six families, by 14% adjacency F1 on the most difficult one, and Claude Opus 5.5 does so by 7.4% there. A third of the true edges the two agents restore on the two hardest families appear in no algorithm’s output. The code and data will be publicly released to facilitate future work.
[LG-74] Partial-Revelation Panel Forecasting: Predictive Information Gain Allocation and Coarsening-Consistent Operators
链接: https://arxiv.org/abs/2610.22831
作者: Bin Wen,Dai-Qiao Zhang,Tien-Ping Tan
类目: Machine Learning (cs.LG)
*备注: 27 pages, 2 figures. Code and aggregate result summaries included as ancillary files
Abstract:We study partial-revelation panel forecasting: given k of n annotator votes on the previous dialogue turn, predict the full count distribution of the next turn’s panel. This cross-turn predictive information differs from same-item annotation value. Under any nested revelation mechanism, the population decrement satisfies V(k)-V(k+1)=I(Y;R_k+1\mid X_k) ; a fitted model also contributes adjacent excess-risk differences. Coarsened-completely-at-random (CCAR) masking supplies the exchangeable random-rater interpretation used in our operators. We define coarsening-consistent panel operators with a closed-form Dirichlet-multinomial realization, learned completion, and rater-count transfer. Against per-level lookup, cold-start count negative log-likelihood differences are -0.422/-0.166/-0.014 nats on IEMOCAP/M3ED/EmoWOZ. A held-out rater experiment measures same-item value: first and second additional votes reduce NLL by 0.583/0.221, 0.578/0.184, and 0.180/0.069 nats, respectively. Cross-turn and same-item gains are not interchangeable, and CCAR masks are controlled interventions rather than observed arrival processes. Consistency is auditable but does not guarantee predictive improvement; per-level marginalization can outperform a shared table. Target-free allocation beats uniform only where row-level signal is predictable: at budget one it recovers 83–93% of retrospective feature class headroom on IEMOCAP/M3ED but does not improve EmoWOZ. Gains are modest and corpus-dependent, with several small or null partial-revelation differences. Code and aggregate result artifacts accompany the preprint.
[LG-75] Ori: Drift-Aware Streaming k-PCA for Very Low-Rank LLM Training
链接: https://arxiv.org/abs/2610.22794
作者: Xinyu Luo,Jincheng Zhou,Mengbo Wang,Zilin Shen,Nadia Atallah Lanman,Petros Drineas,Bruno Ribeiro,Brian Bullins
类目: Machine Learning (cs.LG)
*备注:
Abstract:A prominent family of low-rank LLM optimizers, pioneered by GaLore, performs gradient-subspace projection: at each step, they project the gradient into a low-dimensional subspace and map the update back to the full model. Most existing work either recomputes the gradient subspace at certain intervals, or performs post-hoc adaptation to catch up to the moving subspace. In this work, we explicitly identify in practice the slow, stable manner in which LLM gradient subspaces drift across time. To this end, we propose Ori, a novel low-rank optimizer for training LLMs that captures the shifting principal gradient directions over time to achieve state-of-the-art accuracy under stringent low-rank constraints. Ori employs a streaming k -PCA algorithm inspired by Oja++, whose low per-step cost enables high-frequency, on-the-fly subspace tracking. In practice, Ori reduces the performance gap of the state-of-the-art low-rank optimizers to full-rank optimizers by up to 69.3% on a 1B LLaMA pretraining task, and achieves the best performance on 10 out of 16 task scenarios in Roberta-Base fine-tuning. Together, these contributions allow Ori to further push the performance and efficiency frontiers of gradient-projecting optimizers, narrowing the gap to full-rank training for LLMs. Code can be found at this https URL.
[LG-76] Anytime Contracts for AI Services
链接: https://arxiv.org/abs/2610.22781
作者: Boris Velasevic,Eden Saig,Eric Mazumdar
类目: Computer Science and Game Theory (cs.GT); Machine Learning (cs.LG)
*备注: 19 pages, 3 figures
Abstract:AI services are prone to moral hazard, as model providers can silently route requests to cheaper, lower-quality models, and the inherent stochasticity of model outputs makes this degradation hard to detect. Contract theory addresses moral hazard by designing payment rules that reward high-quality outputs and prevent strategic overcharging. However, existing frameworks assume interactions of fixed duration, whereas AI services are typically consumed through subscriptions and pay-as-you-go APIs that users can terminate at any time based on observed performance. We show that this flexibility can break standard contracts, opening them up to abuse by either the providers or users. To address this, we extend contract theory to account for unknown exit rules. We formalize the problem as a Principal-Agent game, and show that budget-optimal contracts in this setting are proportional to anytime-valid hypothesis tests based on likelihood-ratio e-values. We give an efficient algorithm for computing anytime contracts and show how betting e-processes can be used to approximate them when the models are unknown. Finally, we evaluate our findings on synthetic data and data from deployed LLMs.
[LG-77] Reusing Calibration Gains across Model Representations
链接: https://arxiv.org/abs/2610.22771
作者: Xu Cao,Wenxuan Xiao
类目: Machine Learning (cs.LG); Databases (cs.DB)
*备注:
Abstract:Scale calibration learns an update for one model representation. We study how much of that update can be reused by another representation of the same pretrained model. Our transfer procedure extracts positive group-wise gains and installs them in the recipient’s weights or native quantization scales without recipient-side optimization. On a quantized recipient, it preserves integer codes, zero points and storage layout. For a Qwen3-1.7B floating-point/GPTQ INT3 pair, bidirectional transfer retains 76.9-90.5% of the text-loss improvement from recipient-specific calibration on fresh WikiText and C4 cohorts excluded from project development. Shuffling the same gain values within each projection worsens both recipients, showing that their learned group assignment matters after transfer. A Llama-3.2-3B control keeps the initial native tensors at all four calibrated projections identical and changes nine projections elsewhere. Cross-installation retains 98.6-99.3% of the text-calibration benefit, alongside a small, resolved preference for each recipient’s own gain. These results quantify both the adaptation carried by a reusable calibration update and the remaining benefit of recipient-specific training.
[LG-78] One Communicator Three Vendors: Extending the MoA Predictive Theory to a Real NVIDIAAMDCPU MPI Ensemble
链接: https://arxiv.org/abs/2610.22769
作者: Lenore Mullin,Gaetan Hains
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注:
Abstract:The companion MPI-extension paper validated the first communication-bearing machine-shape ensemble: four NVIDIA A100 ranks and four CPU ranks on one NCSA Delta node under one Open MPI job, reassembled by MPI_Gatherv. This paper extends it to a multi-vendor communicator. A Slurm heterogeneous job spanning Delta’s A100 and AMD MI100 partitions places A100, MI100 and CPU ranks (nine in all) under one communicator for the first time. We give the theory (a w_i proportional to 1/t_i weight split for three concurrent realizations; a het-job specification as an admissible directive) and report real results, including what the run exposed: the 1/t split assumes time linear in rows, and this fails for the MI100 rank (6 rows took 597 ms against a linear prediction of 1.6 ms). A launch-geometry model, with predictions recorded before measuring, succeeds for the A100 (time flat from 128 to 3456 rows, stepping at exactly 3520 and 6976) and places the MI100 floor in the key loop; changing its launch settings cut about 600 ms to 14.6 ms and the energy per call at 6 rows by 32x, though the MI100 remains about 95x slower than an A100 at 8192 rows. Controlled reruns show that the companion paper’s 8.1x A100 result and 22 ms Gatherv were artifacts of four ranks sharing one GPU under Open MPI’s default transport. A key/value-split partition whose message sizes are derived from shapes alone is then run: the ring Allreduce of the 8.5 MB message took 11.3, 16.0 and 15.3 ms against a predicted 11-12 ms, the small Allreduce missed its prediction in every run, and compute times reproduce to about 1% while communication times vary up to 4x between runs. Each idea is introduced with a figure and closed with worked examples.
[LG-79] Acquisition Function-Guided Generation for Bayesian Optimization over Discrete Spaces
链接: https://arxiv.org/abs/2610.22763
作者: Tommaso Ancilli,Thomas Wedenig,Henry Moss,Robert Peharz
类目: Machine Learning (cs.LG)
*备注:
Abstract:Bayesian optimization (BO) over discrete spaces is challenging due to missing gradients and the sparsity of valid solutions. Latent-space BO sidesteps both issues by fitting a surrogate in the latent space of a generative model, which however substantially complicates the direct incorporation of domain knowledge that is naturally expressed in the original discrete space. In this work, we instead combine a pretrained generative model with a surrogate defined directly in data space, where inductive biases such as domain-specific kernels can be readily incorporated. Rather than maximizing the acquisition function over the discrete design space, our method samples candidates from the generative model under a distribution tilted by the acquisition function. To recover useful gradients for sampling, we embed this tilted distribution into a diffusion bridge that yields a smooth, posterior-averaged acquisition function. This leads to \emphacquisition function guided generation (AFGG), which we instantiate with a categorical diffusion model and an autoregressive normalizing flow. Across a broad suite of molecular benchmarks, the diffusion model variant performs on par with leading BO methods, while the flow model variant outperforms state-of-the-art methods in most cases and remains highly competitive otherwise. Our implementation is available at this https URL.
[LG-80] Shape-Aware Neural Time Fields for Rigid-Body Motion and Manipulation Planning
链接: https://arxiv.org/abs/2610.22762
作者: Yanming Zhou,Yuchen Liu,Ahmed H. Qureshi
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注:
Abstract:Rigid-body motion planning is fundamental to robotic transport and manipulation, where both object position and orientation determine whether a body can move through cluttered environments. This coupling makes planning chal- lenging because translation and rotation can be constrained very differently by surrounding geometry. We propose Shape- Aware Neural Time Fields, a physics-informed neural motion planning framework that learns a reusable travel-time, or cost- to-go, representation for rigid-body motion while explicitly accounting for these distinct geometric restrictions. The learned field captures local motion difficulty in translation and rotation and is paired with a sampling-based planner that balances progress toward the goal with the cost of reaching nearby poses. We evaluate the framework across 2D and 3D rigid- body planning tasks and non-prehensile manipulation through object pushing in diverse environments. Results show substantial improvements in planning success and inference efficiency over physics-informed neural planning baselines, while remaining competitive with classical sampling-based planners at significantly lower planning times. Ablation studies further validate the contributions of both the learned representation and the proposed planning strategy. Finally, closed-loop pushing experiments under execution deviations and external disturbances demonstrate that fast replanning enables reactive and adaptive task execution.
[LG-81] Refusal Is Not Robustness: Measuring Harmful Reachability in Autoregressive Language Models
链接: https://arxiv.org/abs/2610.22754
作者: Parv Kapoor,Teresa Fortes,Claudia Mamede,Rui Abreu,Jonathan Aldrich,Eunsuk Kang
类目: Machine Learning (cs.LG); Logic in Computer Science (cs.LO)
*备注:
Abstract:LLM safety evaluations via jailbreaking typically index on attack success rate (ASR): whether a generated response violates safeguards. However, ASR only evaluates sampled outputs and provides little information on the underlying generation process. Autoregressive language models define distributions over possible outputs, and a sampled refusal can coexist with substantial probability mass on harmful generation paths. We study this hidden structure through bounded continuation analysis (BCA), which models autoregressive generation as a stochastic process over token prefixes and computes the finite-horizon probability that generation reaches a harmful continuation. BCA constructs a bounded continuation tree, labels terminal continuations with a semantic harm evaluator, and estimates harmful reachability using exact probabilistic model checking. BCA complements ASR by measuring how much harmful behavior remains accessible beneath a sampled safe response. This makes it useful for distinguishing models that look identical under measured ASR values, and allocating red-team effort toward regions of the generation process with greater hidden risk. Across four instruction-tuned LLMs and 100 JailbreakBench behaviors, we show that BCA effectively quantifies hidden harmful behavior and provides a useful signal to guide red teaming.
[LG-82] Native Geometry vs. Linear Recoverability: How Alignment Changes Platonic Scaling NEURIPS
链接: https://arxiv.org/abs/2610.22749
作者: Angus Hoy,Austin Lutterbach,Jasper Chong,Aman Kumar,Michael J. Smith
类目: Machine Learning (cs.LG); Instrumentation and Methods for Astrophysics (astro-ph.IM)
*备注: 10 pages, 6 figures, Accepted to NeurReps 2026 (NeurIPS Workshop on Symmetry and Geometry in Neural Representations), Extended Abstract track
Abstract:Do larger models learn increasingly similar representations of the same world? The Platonic Representation Hypothesis (PRH) predicts that they should, but similar can mean native cosine geometry—nearest-neighbour structure induced by cosine similarity before any learned alignment—or linear recoverability: correspondence revealed by a supervised linear map. Using matched Legacy Survey and Hyper Suprime-Cam (HSC) embeddings across five model families, we find that Ridge alignment strengthens the relationship between model size and cross-dataset neighbourhood correspondence relative to native cosine geometry. The effect appears in all five families and persists when both representations are reduced to dimension, so it is not simply that wider embeddings give alignment more dimensions to work with. The learned maps remain far from rotations or uniform scaling, and Sparse Autoencoder (SAE) and Block-Sparse Featuriser (BSF) features give no advantage over the same alignment applied to dense embeddings. Model scale thus improves linear recoverability more clearly than native cosine-neighbourhood correspondence, without establishing convergence toward a shared native metric geometry.
[LG-83] he LLM -Attacker: A Unified Red-Team Framework for Privacy Evaluation of Split and Distributed LLM Systems
链接: https://arxiv.org/abs/2610.22722
作者: Georgios Politis,Evangelos Pappas
类目: Cryptography and Security (cs.CR); Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注:
Abstract:Split language models keep input text on a client but send intermediate representations to a server. Attacks can reconstruct text from these messages. Comparing privacy evaluations is difficult when attackers see different messages, know different model weights or use different datasets and recovery measures. We present LLM-Attacker, a framework that records these choices and lets researchers check, compare and extend an evaluation. Compatible attacks reuse recorded messages, and their predictions are saved for independent scoring. The scoring rule requires a linked undefended reference before a defence can pass. Reports also identify attacks that remain untested. In a controlled GPT-2 experiment, adding gradients returned during training increased token recovery from 94.20% to 97.38% and exact-document recovery from 13.71% to 37.77%. The 95% interval for the token gain lay above the two-percentage-point threshold fixed before testing. This experiment is shared with a companion study; here, it tests how the evaluation responds to a change in attacker access. Further studies measure how corpus choice and compression defences affect recovery, compare an adapted published attack with its reference implementation, and test the measurement code. The framework reports implementation checks separately from measured attack effectiveness. Each privacy claim should state the attacker, dataset and scoring rule, report its controls and identify untested attacks.
[LG-84] Spectral Limits of Polynomial Graph Filters on the Long-Range Ising Model Benchmark
链接: https://arxiv.org/abs/2610.22713
作者: Amin Najafgholizadeh,Mohsen Afsharchi
类目: Machine Learning (cs.LG)
*备注: 30 pages, 4 figures, 15 tables
Abstract:The Long-Range Ising Model benchmark (LRIM) guarantees long-range dependence by construction: each label is the node’s spin times the spin field convolved with a power-law coupling, so any model’s distance from the target is measurable. We prove that a readout whose filter is a polynomial in the adjacency, the linear core of isotropic message passing and spectral graph networks, leaves an error floor that no degree lowers, because the target varies along a level set of the spectrum on which every such filter is constant. Computed exactly, the floor and the best filter of each degree predict two spectral networks adapted from published architectures to within a tenth of a decade. A filter escapes the floor only if it weighs separately the offsets the adjacency lumps together, and must still resolve the coupling across the whole lattice. A strided filter of about fifty coefficients does both: fitted to the training labels, its test error is 52 times below the best published network’s. A dilated convolutional network beats the best published result at 256 nodes, matches it at 1,024, and gives the first in-domain results at the three larger sizes, while a Fourier neural operator with its standard modes falls further behind as the lattice grows. At its standard budget the dilated network stays two to four orders of magnitude short of a larger strided filter fitted to the labels. The analysis carries over to the benchmark’s second coupling.
[LG-85] Normalizing Flow Maps
链接: https://arxiv.org/abs/2610.22711
作者: Francesco M. Ruscio,Timothy James Hitge,Fred Zhangzhi Peng,Danyal Rehman,Zander W. Blasingame,Nicholas M. Boffi,Michael M. Bronstein,Alexander Tong,Avishek Joey Bose
类目: Machine Learning (cs.LG)
*备注: 47 pages
Abstract:Scalable approaches to learning neural dynamical systems have placed diffusion, flow matching, and few-step generators in flow maps at the forefront of generative modeling. Such state-of-the-art models have limited access to explicit model densities despite evolving from classical generative modeling approaches that leveraged tractable likelihoods, e.g., Normalizing Flows and Autoregressive models. In this paper, we combine these threads of few-step generation and tractable sample likelihoods and introduce Normalizing Flow Maps (NFM), a class of exactly invertible pushforward maps that generalizes normalizing flows and flow maps. More precisely, NFMs learn the integrator of an underlying dynamical system, but crucially, they provide efficient exact likelihood evaluation by construction. To enable scalable training of NFMs, we derive two complementary axes of consistency inherent to NFMs that yield (self) distillation objectives: (1) pointwise matching, which unlocks \ell_2 -based flow-map style regression losses, and (2) distributional matching, which matches pushforward distributions to the marginals of a prescribed forward process. On conditional image generation on CIFAR-10 and ImageNet-64, NFMs achieve state-of-the-art sample quality among parameter-matched normalizing flows, while supporting likelihood-based RLHF. For equilibrium sampling in Boltzmann Generators, NFMs outperform prior state-of-the-art normalizing flows on both single-system and transferable peptide benchmarks.
[LG-86] Data Provenance Aggregation and Structural Descriptors for Query-Template Identification
链接: https://arxiv.org/abs/2610.22705
作者: Paulo Pintor,Rogério Costa,José Moreira
类目: Databases (cs.DB); Machine Learning (cs.LG)
*备注:
Abstract:Runtime query similarity identification supports several activities, including workload analysis, identifying recurring behaviour, auditing, debugging, and anomaly detection. Existing methods for evaluating query similarity mostly rely on SQL text, execution plans, or result values but do not account for input data or how results are derived during query execution. This paper proposes a novel approach using data provenance for query-template identification. We construct query-level profiles that capture explicit data dependencies, occurrence frequencies, and structural properties by aggregating tuple-level Lineage and Why-provenance across entire query results. We also derive Provenance Structural Descriptors (PSD), which summarise these profiles using a small set of numerical features. We evaluate our proposals using these profiles to identify query templates over (12,500) distinct query instances generated from parametrised TPC-DS query templates. The experimental evaluation shows that query templates can be identified solely from their provenance footprints, and that accuracy improves as more historical data becomes available. It also shows that PSD achieve high classification accuracy with limited training history and low test-processing cost. Subjects: Databases (cs.DB); Machine Learning (cs.LG) Cite as: arXiv:2610.22705 [cs.DB] (or arXiv:2610.22705v1 [cs.DB] for this version) https://doi.org/10.48550/arXiv.2610.22705 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-87] Chance-Constrained Personalized LLM Routing with Online Preference Learning
链接: https://arxiv.org/abs/2610.22695
作者: Ameya Anjarlekar,Rasoul Etesami,R Srikant
类目: Machine Learning (cs.LG); Applications (stat.AP)
*备注: 18 pages, 8 figures, 1 table
Abstract:We study personalized LLM routing, where a router selects among heterogeneous models with different costs and capabilities. The router must minimize cumulative inference cost while ensuring reliable realized response quality, even though user preferences are unknown and must be learned from feedback during deployment. We formulate this problem as a constrained contextual bandit and develop an algorithm that learns preferences while balancing cost and quality. Under a logistic preference model and a strict-feasibility condition, we establish sublinear expected cost regret relative to an optimal stationary randomized contextual oracle and a high-probability guarantee on realized cumulative satisfaction. The framework extends beyond LLM routing to broader online model-selection problems with reliability requirements.
[LG-88] Edit Discovery in Human-AI Interactions
链接: https://arxiv.org/abs/2610.22661
作者: Alireza Mirrokni,Shayan Kiyani,George Pappas,Hamed Hassani
类目: Machine Learning (cs.LG)
*备注:
Abstract:Human-AI interactions are becoming a rich source of data for improving AI systems: while many model responses are already satisfactory, others expose failures that could become valuable supervision if identified and corrected. We study this setting through the notion of an editor: a more capable evaluation process that can inspect a response and either confirm it or modify it, but whose invocation is costly. The editor may be a human expert, a stronger AI system, or the same model equipped with additional computation, context, tools, or information. The key challenge is to decide which interactions to send to the editor before knowing whether an edit will actually be made, since invoking the editor itself consumes limited resources such as expert time, inference cost, latency, or tool-use overhead. We formulate this problem as Edit Discovery Optimization (EDO), which seeks to minimize edit opportunities missed among skipped interactions while controlling unnecessary confirmations and respecting an editor-cost budget. We characterize the population-optimal policy and show that it takes the form of a cost-adjusted threshold on the posterior probability of an edit. Building on this structure, we develop EDIT (Edit Discovery via Iterative Thresholding), an online algorithm for the partial-feedback setting in which edit labels are revealed only for selected interactions. EDIT combines an adaptive edit score, a confirmation-control threshold, and a cost-aware selection rule; it satisfies the editor budget at every round and controls the average confirmation rate up to a vanishing finite-horizon term, without distributional assumptions on the interaction stream. Across four reasoning and factual question-answering benchmarks, EDIT achieves a better trade-off between editor cost, Type-I error, and Type-II error than direct LLM-based batch selection.
[LG-89] he Ephemeral Cost of Switching Optimizers Across LLM Training Stages NEURIPS2026
链接: https://arxiv.org/abs/2610.22650
作者: Xinyu Zhao,Davis Wertheimer,Ashutosh Sharma,Rameswar Panda,Paul Pu Liang
类目: Machine Learning (cs.LG)
*备注: Accepted to the Transitioning from Pre-Training to Post-Training Workshop at NeurIPS 2026
Abstract:Recent work suggests that switching optimizers between pretraining and subsequent training stages incurs a performance cost, but systematic investigation of this phenomenon is lacking. We study the mismatch between AdamW and Muon optimizers by pretraining 1.2B Llama-3 models from scratch while varying the proportion of updates taken with each optimizer, and similarly interpolating between the two optimizers during continued pretraining and supervised fine-tuning (SFT). Surprisingly, we find little evidence of a meaningful optimizer mismatch penalty: final performance is largely explained by the independent effects of the pretraining and post-training optimizers, with downstream performance generally improving as the proportion of Muon updates during SFT increases. Although matching the pretraining optimizer can reduce forgetting during SFT, this does not translate into improved final benchmark performance. We further evaluate AdamW and Muon for task-general SFT of open-weight models, across the Llama and Qwen families up to 14B parameters, where Muon remains competitive with AdamW, despite these models having been pretrained with AdamW. Together, our results suggest that the choice of optimizer for post-training can be made largely independently of the optimizer used during pretraining.
[LG-90] Alert-List Concordance for Evaluating Imputation Strategies in ICU Sepsis Prediction ALT ICDM2026
链接: https://arxiv.org/abs/2610.22635
作者: Priyanka Siddappa,Muntaqim Ahmed Raju,Shifat Haider Al Amin,Ruizhe Ma
类目: Machine Learning (cs.LG); Quantitative Methods (q-bio.QM)
*备注: 8 pages, 2 figures. Accepted at the 14th Workshop on Data Mining in Biomedical Informatics and Healthcare (DMBIH 2026), held in conjunction with IEEE ICDM 2026
Abstract:Preprocessing is an under-measured reliability risk in clinical prediction: a model can keep a similar AUROC and still change who receives an alert. Imputation quality is commonly evaluated using reconstruction error or aggregate discrimination, neither of which directly measures agreement on which patients are flagged. At a matched alert rate, pipelines with similar aggregate performance can produce different patient lists. This question is especially relevant under the frequent, structured missingness of real ICU records. We therefore introduce alert-list concordance, which asks whether two pipelines identify the same high-risk patients, measured by the Jaccard similarity of their top-k% alert sets with patient-level bootstrap confidence intervals. On PhysioNet/CinC 2019 (40,336 patients, 71.9% missingness), we compare nine imputation strategies and a native-missingness baseline using a fixed XGBoost configuration, fitted and Platt-calibrated separately for each strategy. After Bonferroni correction, 22 of the 28 reconstruction-method pairs showed no statistically significant difference in patient-level AUROC, yet their alert lists overlap at Jaccard 0.518 to 0.750. Similar aggregate discrimination can therefore accompany different patient selections. ForwardFill is a separate outlier, pairing higher discrimination with the lowest overlap in the matrix, 0.40 to 0.45. The broad pattern holds within each hospital system and under six patient-level aggregation rules. At the row level, eight of ten methods fall within an AUROC range of 0.011. In this row-level pipeline, deep imputers did not outperform mean imputation on discrimination, and the sequential implementations cost 5 to 12 times mean imputation in total pipeline wall-clock time on our hardware.
[LG-91] Understanding Stability of Outer Nesterov Momentum on Stochastic Pseudo-Gradients
链接: https://arxiv.org/abs/2610.22623
作者: Nishant Sharma,Laura Gomezjurado Gonzalez,Ganesh Talluri,Sarath Chandar,Hiroki Naganuma
类目: Machine Learning (cs.LG)
*备注:
Abstract:Large neural networks are commonly trained across many workers, and communication between them is a major cost. Workers therefore take several local optimization steps before sending their net parameter displacements, or pseudo-gradients, to an outer optimizer, which often applies Nesterov momentum. We study how the outer momentum, the number of local steps and the averaging over workers control the training dynamics, and in particular their stability. For random quadratics, we derive an exact mean-square stability criterion in any dimension, which reveals that the inner batch size determines whether increasing the number of local steps eventually stabilizes or destabilizes the outer iteration. Moreover, while averaging the pseudo-gradients of independent workers can make larger outer momentum values admissible, the number of workers required for stability can grow exponentially with the number of local steps. Finally, we show that the mean-square dynamics governing stability can be probed directly at intermediate neural-network checkpoints using shared-minibatch perturbations. We test our theoretical findings across networks, and measure their consequences for training loss.
[LG-92] Noise-shaped Sampling and Annealing for Scalable On-the-edge Training of Binary and Ternary AI Models
链接: https://arxiv.org/abs/2610.22621
作者: Zihao Chen,Zhili Xiao,Rui-Jie Zhu,Yipeng Wang,Jason Eshraghian,Shantanu Chakrabartty
类目: Machine Learning (cs.LG)
*备注:
Abstract:The training loss of binary and ternary neural networks, expanded as a multilinear polynomial over the discrete weights, is structurally equivalent to the energy of a hierarchical high-order Ising model. We exploit this equivalence to apply Ising solvers to discrete-weight network training, while accounting for a fundamental difference in objectives: an Ising solver seeks the ground state of a fixed energy function, whereas training seeks solutions that generalize to unseen data. Our key contribution is \Sigma\Delta -Sampler, an optimizer that reconciles these objectives by embedding a stochastic sampling-and-annealing step as a noisy quantizer inside a first-order \Sigma\Delta noise-shaping loop. The loop pushes noise from weight quantization, high-order frustration, and mini-batch sampling out of the low-frequency adaptation band, so that discrete weight updates follow the smooth descent trajectories embedded in the rugged loss landscape. We validate the framework on a planted quadratic unconstrained binary optimization (QUBO) benchmark with a known minimizer; on CIFAR-10, where a 13 M-parameter ternary model comes within 1% of full-precision accuracy; on ImageNet-1k, where a ternary ResNet-18 comes within 0.4% Top-5 accuracy of a ternary baseline; on class-incremental continual learning, where prior-task accuracy is preserved; and on MatMul-free language models up to 2.7 B parameters, where it outperforms a straight-through-estimator baseline under matched continued pre-training. Compared with standard quantization-aware training, \Sigma\Delta -Sampler reduces the training memory footprint by over 50% , delivers an order-of-magnitude speed-up, and reduces weight-update memory traffic by four orders of magnitude, making it attractive for on-device training and fine-tuning of large binary and ternary models.
[LG-93] Rarity of Singletons Matters Supplementary Materials
链接: https://arxiv.org/abs/2610.22604
作者: Pragatheeswaran Vipulanandan,Kamal Premaratne,Dilip Sarkar
类目: Machine Learning (cs.LG)
*备注: supplementary for the main paper has been added as well
Abstract:Estimating missing-mass and unseen-species counts from finite samples relies heavily on rare-event occupancy statistics, particularly singletons. In small-sample regimes, these statistics can be highly variable, making classical estimators sensitive to sampling fluctuations and potentially unreliable. We propose a fragility-aware (FA) framework that explicitly quantifies this sensitivity and adaptively regularizes conventional missing-mass and unseen-species estimators. Using singleton-based fragility measures derived from probability mass and occupancy counts, we formulate an entropy–Kullback–Leibler (KL) optimization that adjusts the reliance on base estimates according to their fragility. The resulting objectives are strictly concave, admit unique closed-form solutions, and require no Poissonization or asymptotic approximations. The framework provides a lightweight correction to classical estimators, including Good–Turing and Good–Toulmin, while reducing reliance on potentially unstable rare-event statistics. Across 360 experimental configurations spanning synthetic distributions and real-world language corpora, our method remains competitive in stable regimes while improving estimation accuracy and stability in fragmented, long-tailed, and small-sample settings. The findings demonstrate that accounting for the fragility of rare observations can improve the reliability of inference about unobserved species and their associated probability mass.
[LG-94] Diagnosing GNN Failure Modes with Layer-Wise Probe Entropy
链接: https://arxiv.org/abs/2610.22595
作者: Shanzi Bao,Mar Gonzàlez I Català,Pietro Liò
类目: Machine Learning (cs.LG)
*备注:
Abstract:Graph neural networks (GNNs) are typically evaluated only at their final layer, obscuring how task-relevant information evolves during message passing. We introduce a post-hoc diagnostic that freezes a trained GNN, fits an independent linear probe to each layer, and tracks probe accuracy alongside normalized predictive entropy. Across real and synthetic graph settings, layer-wise probe entropy distinguishes beneficial, neutral, and harmful smoothing, a capability not demonstrated by existing geometric measures. Moreover, by comparing shallow and deep models, we show that the onset of entropy degradation separates the two main GNN failure modes: structural mismatch under heterophily, which appears from the earliest layers even in shallow models, and excessive-depth oversmoothing, which emerges only as propagation depth increases. These results establish probe-entropy trajectories as a tool for analyzing information flow in GNNs and motivate future work on adaptive depth selection and early-exit mechanisms.
[LG-95] Flat Loss Evolving Features: Grokking in Modular Addition
链接: https://arxiv.org/abs/2610.22593
作者: Akash Kumar,Karthik Sankaranarayanan
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 97 pages, 17 figures. Code: this https URL
Abstract:Can a network keep learning while its training cross-entropy stays nearly flat at a positive level? We show that it can. On modular addition, a standard testbed for grokking, ReLU networks trained with supported-Polar hidden updates, normalized output-layer updates and weight decay can keep 100% training accuracy while their hidden features change and held-out accuracy rises from zero to above 90%, at relatively flat training cross-entropy. Correct classification need not minimize cross-entropy, which still depends on prediction confidence. We prove this coexistence for one-hidden-layer networks in two finite-window theorems from constructed initializations: one gives a constant gain in the Fourier fraction of the hidden features, and the other holds at arbitrarily low positive loss, where a readout refitted on the training data generalizes before the network. Experiments from random initialization show the same coexistence on a window fixed in advance, and in deeper networks held-out accuracy also rises over post-hoc windows with a net increase in training cross-entropy. With a last-layer rate set per training fraction and head decay scaled with the prime, deeper networks at p=47 and 61 meet a criterion on post-hoc windows, fixed in advance, in nearly all fresh runs. A one-block transformer trained with the same updates and a slowly decaying output gain keeps every training pair correct at every logged checkpoint while its training cross-entropy rises and its held-out accuracy approaches one; with two blocks this holds through the held-out rise, after which training pairs begin to flip. A plateau in training loss need not be a plateau in learning. Code: this https URL.
[LG-96] Initiating Removal from Operational Symptoms in Federated Vehicular Networks
链接: https://arxiv.org/abs/2610.22584
作者: Zihao Ding,Jun Huang,Qiang Duan
类目: Machine Learning (cs.LG); Networking and Internet Architecture (cs.NI)
*备注:
Abstract:Connected vehicles train a driving model through federated learning (FL) and drive with it, so a faulty contribution often surfaces only as a delayed symptom on the road. Federated unlearning (FU) assumes that a removal request names what to forget. A symptom names no vehicle or round, and aggregation has already spread its cause across the fleet. We propose \underlineAggregation-\underlineTime \underlineOperational-evidence \underlineNeutralization of \underlineEffects (\method), which makes unlearning a term of the FL aggregation rule. \method projects everyday driving errors and reported symptoms onto a subspace spanned by recent global increments. It propagates the surviving effect of every contribution with curvature estimated from these increments and scores all contributions in one backward pass. A two-tier correction suppresses suspected effects in every round and eliminates confirmed ones after a one-round probe. We prove that backward scoring matches forward scoring and that the elimination is exact when training is flat outside the subspace. On three driving datasets, \method leaves 43% to 61% less residual harm than the strongest baseline in eight of nine settings while staying within 0.36 points of full retraining in utility.
[LG-97] Noise-Robust Forced Alignment for Spontaneous Speech
链接: https://arxiv.org/abs/2610.22583
作者: Laurin Wagner,Mario Zusag
类目: Machine Learning (cs.LG)
*备注:
Abstract:We show that forced aligners degrade on spontaneous, noisy real-world speech, and that fillers, laughter, and disfluencies corrupt word boundaries, impacting many downstream applications. We revisit the classical GMM-HMM aligner with upsampled WavLM features instead of MFCCs, explicit phone units for fillers and non-lexical vocal events trained on verbatim-annotated spontaneous speech, and a discriminative projection fitted on clean and noise-corrupted feature pairs to improve robustness. The pipeline is self-bootstrapped from 149h of curated audio without manual or external boundary targets. On conversational (Buckeye) and disfluent (FluencyBank) speech it reaches word-boundary MAEs of 20.6 and 76.9ms, 19% and 17% lower than the strongest of ten baselines. Catastrophic misalignments (200ms) are 2.7 times rarer than Montreal Forced Aligners (MFA) best model even after widening MFA’s search beam, which we show removes a large pruning artifact from its conversational error; event units halve the error next to laughter, while being meaningfully more robust than existing aligners.
[LG-98] Embodiment-Classifier Guidance: Transferring Frozen VLAs Across Robot Dynamics without Demonstrations
链接: https://arxiv.org/abs/2610.22576
作者: Devin Han,Hershraj Niranjani
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: 19 pages, 4 figures
Abstract:Flow-based policies such as \pi_0 are trained on a fixed set of embodiments, and transferring them to new ones is an open problem. To adapt a frozen \pi_0 to a new embodiment without additional demonstrations or parameter updates, we train a lightweight embodiment classifier to distinguish successful action chunks on the source body from those on the target body, using a few calibration tasks. On transfer tasks that \pi_0 saw on its source body but the classifier was never trained on, we nudge each denoising step of \pi_0 with the classifier’s gradient, which estimates the direction from the distribution of successful actions on the source body toward that on the target body. On the LIBERO simulation benchmark with four modified controllers, embodiment-classifier guidance raises \pi_0 's average success on the target bodies from 72% to 80% across 12 suite-controller pairs, with a 31% relative reduction in failures (11 of 12 gain; 5 significant after Holm correction). In simulation, guidance is data-efficient, needing few target-body rollouts per task, which makes it promising for sim-to-real transfer, where rollouts are costly. Guidance is mixed on weak wrists and harmful on a locked wrist, when the classifier lacks data or favors infeasible motions. Collecting more target-body data with guided \pi_0 can fill such gaps: on one weak-wrist arm, guided \pi_0 solved a task on which unguided \pi_0 had zero successes in the original data, and retraining the classifier on the guided successes raised success from 52.5% to 70.7%.
[LG-99] Structure-Aware Spatial Kernels for Gaussian Process Distributed Port-Hamiltonian Systems
链接: https://arxiv.org/abs/2610.22575
作者: Tianyi Yang,Thomas Beckers
类目: Machine Learning (cs.LG)
*备注: 8 pages, 2 figures. Code and reproduction scripts are available at this https URL
Abstract:Gaussian process distributed port-Hamiltonian systems (GP-dPHS) enable data-driven modeling of distributed dynamical systems while preserving their physical structure. However, standard stationary kernels can be inefficient for spatially heterogeneous systems because they neither exploit known input invariances nor reflect structured spatial dependencies. We address this limitation with a structure-aware kernel design that removes irrelevant state coordinates and decouples state and spatial dependencies through a tensor-product covariance. This modular construction allows qualitative knowledge of spatial heterogeneity to be incorporated by selecting an appropriate spatial kernel without requiring an analytical description of the underlying profile. Numerical experiments on a nonlinear string with periodic, smoothly varying, and semi-piecewise stiffness profiles demonstrate that structural decoupling consistently improves reconstruction accuracy, while profile-matched spatial kernels yield substantially larger gains in force-field and trajectory reconstruction.
[LG-100] Coverag e Cliffs in Off-Policy Quantile Certification
链接: https://arxiv.org/abs/2610.22565
作者: Pauline Bourigault,Emmanuelle Bourigault
类目: Machine Learning (cs.LG)
*备注:
Abstract:We study when logged contextual-bandit data can certify an upper tail quantile for a policy selected from the same sample. We construct a PAC-Bayes Bernstein lower-confidence envelope for the target cumulative distribution function (CDF), simultaneous over a predeclared threshold grid and all data-dependent posteriors. The resulting procedure, TailCert, uses the envelope’s first crossing of level p to return an atom-safe upper confidence bound on the selected posterior’s pooled p-quantile (or inconclusive). It can also be used to select a policy under a certified tail constraint. The concentration ingredients are standard; the main theoretical contribution is a matched finite-class certification boundary comparing full logs with their joint weighted CDF indicators at a fixed threshold. On an explicit rare-action class, the required sample size grows linearly with the overlap factor W, logarithmically with policy-search and confidence complexity, and quadratically with the inverse CDF safety margin; the upper and lower bounds agree up to constants in the stated regime. On this class, the CDF reduction hides which actions exceeded the budget; retaining action counts restores the full-log rate. Simulations verify the predicted search-complexity shift, winner’s-curse failure of naive reuse, and the power cost of splitting.
[LG-101] Deterministic Bounds on Accuracy and Stability of Metric-Space Regression and Classification
链接: https://arxiv.org/abs/2610.22564
作者: Nina Maria Gottschling,Jakob Gawlikowski,Michele Caprio
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:Data-driven approaches often face ambiguity in datasets due to sensor or target noise, insufficient information, and noise in the input data. While best-fit machine learning methods can accommodate such noise through uncertainty estimates, they are prone to overfitting and it is challenging to quantify ambiguity of the relation between inputs and targets. We introduce a geometric framework that quantifies this ambiguity through the average kernel size, computed with data-dependent, i.e. model-independent, feasible sets derived from worst-case input and target noise. The framework provides computable, deterministic, non-asymptotic bounds on the best achievable empirical error and accuracy for a finite, constructed dataset, independent of the predictive model class, and applies to regression and classification with pseudo-metric input spaces and metric target spaces. The ambiguity of the input-target relation implies three consequences: an empirical error floor, an accuracy-stability trade-off, and, under standard probabilistic assumptions, a strictly positive distribution-training error gap when training below the accuracy bounds. Under standard Borel assumptions for Euclidean norms, the distributional analogue of the average kernel size is related to Bayes risk as it reduces to the optimal distribution root-mean-squared error times the square root of two. Experiments spanning tabular regression, image regression and classification datasets yield nontrivial bounds on the best achievable empirical error and accuracy on a constructed dataset, alongside descriptive comparisons with model performance on the original datasets. The bounds provide a computable, model-independent reference scale on the constructed dataset and comparisons with model performance on the original dataset are descriptive.
[LG-102] When Plans Change Remember: Replanning -Guided Visual Memory for Vision-Language-Action Models
链接: https://arxiv.org/abs/2610.22556
作者: Victoria Smelova,Daniil Kazachkov,Georgiy Kazantsev,Iuliia Tsvetkova,Egor Cherepanov,Nikita Kachaev,Artem Latyshev,Aleksandr I. Panov,Alexey K. Kovalev
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: 20 pages, 8 figures, 6 tables
Abstract:Long-horizon, multi-stage Vision-Language-Action (VLA) policies often require visual information that is no longer present in the current observation. Retaining the complete observation history is computationally expensive and highly redundant, motivating selective visual memory. We introduce RePlanMem-VLA (Replanning-Guided Keyframe Memory for VLA), which turns a policy’s own plan revisions into an online memory-write signal. Under receding-horizon execution, consecutive action chunks predict some of the same future actions. RePlanMem-VLA treats unusually large disagreement between these temporally aligned predictions as evidence of a task-relevant event. We first compare 35 disagreement measures across single-task autoregressive decoder-only and Flow-Matching Transformers trained from scratch, then validate the selected signals on pretrained OpenVLA-OFT and \pi_0.5 policies over MIKASA-Robo-VLA tasks. The resulting detectors achieve ROC-AUC of 0.981-0.997. We then use the selected signals in RePlanMem-VLA to trigger updates to a bounded visual memory, with a robot-masked write filter suppressing redundant frames. In the evaluated memory-retention experiments, RePlanMem-VLA preserves task-relevant observations until the later decision point in 100% of episodes. Real-robot replay data further show that event selection and redundancy suppression extend beyond simulation. These results establish plan revision as a policy-internal signal for controlling what a VLA remembers. Website: this https URL.
[LG-103] On the SPOT: Enhancing Speculative Decoding for Diffusion Language Models via Commit-Order Prediction
链接: https://arxiv.org/abs/2610.22543
作者: Eliezer Shlomi,Eliron Rahimi,Avi Mendelson,Chaim Baskin
类目: Machine Learning (cs.LG)
*备注: Preprint, project page: this https URL
Abstract:Diffusion language models (DLMs) promise parallel text generation, but in practice face an unresolved speed-quality trade-off. Exact decoding fills in only one masked position per model pass, while faster multi-token decoding schemes are lossy. Speculative decoding is the standard lossless acceleration for autoregressive models. However, it does not transfer directly to DLMs, which must choose both what to generate and where. We show that this order is the missing piece: the model’s own predictions from the previous pass already give the right token at 92-100% of positions, but not which position comes next. Based on this observation, we introduce Speculating Positions Order Trajectory (SPOT), which learns to predict this order. A 0.62M-parameter model reads only the target’s output distributions and proposes a tree of candidate continuations, and the target verifies all of them in a single pass by checking whether each equals its own next state. On the verification side, our analysis shows that prior block-diffusion verification approaches either assume T=0 or a sequential proposal interface. To support tree-based speculation at any temperature, we propose using coupled-noise match verification for diffusion blocks, following shared-Gumbel-noise coupling approaches. Our comprehensive experiments show that SPOT cuts target passes by up to 6.8x and reaches up to 3.2x the throughput of standard decoding, beating the calibrated and training-free drafting baselines in every task and budget at T=0 and T=1 . In addition, we show that the learned drafter transfers across models, demonstrating successful transfer across six different DLMs. Project page: this https URL
[LG-104] Dont Retrain Remix: Adaptation of Web Agents using Mixture Hypernetworks
链接: https://arxiv.org/abs/2610.22539
作者: Md Farhan Ishmam,Minh Pham Dinh,Kenneth Marino
类目: Machine Learning (cs.LG)
*备注:
Abstract:The web is not static or uniform. Websites that serve the same purpose can look and navigate very differently. Even within a single site, successive redesigns can change how familiar features are presented and accessed. Humans adapt to this by recognizing familiar interaction patterns and reusing what they already know. We give web agents a similar ability with WebMix, a mixture hypernetwork that adapts them to unseen interfaces by composing previously trained low-rank adapters. WebMix begins with a bank of adapters trained on source interfaces. At deployment, a scout agent explores the target interface and summarizes its observations into a compact manual. A lightweight router uses this manual to predict mixing coefficients over the adapter bank, which combine the existing adapters into a new one for inference. Adapting to a new interface takes only a scouting run and a forward pass, without gradient updates on the target. With Qwen and Llama models, we evaluate domain transfer on TimeWarp and WebArena-Lite-v2, and version transfer on TimeWarp. In all three settings, WebMix matches or outperforms in-context learning and LoRA baselines, often matching the average success of specialists trained directly on those interfaces. On TimeWarp domain transfer, it gains about 8 percentage points on average over the best baseline with both models. These gains do not come at the cost of seen interfaces, where WebMix keeps its lead, and extend to version transfer. Adapting to a new interface can therefore rely on exploration and composition instead of retraining.
[LG-105] Learning smooth dynamics via Kinematic Interpolants
链接: https://arxiv.org/abs/2610.22519
作者: Yiheng(Roger)Jiang,Youssef Marzouk,Avishek Joey Bose,Aram-Alexandre Pooladian
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 22 pages, including appendix, 4 figures
Abstract:Inferring dynamics from real-world experimental data is inherently ambiguous, as many plausible dynamical systems can explain the same positional observations at measurement times. For higher fidelity modeling of applications ranging from single-cell RNA to vortex dynamics and even flocking of birds, scientific observables increasingly also include velocity estimates that serve to resolve this issue. However, incorporating this rich information in a scalable manner, free of simulation or unnecessary approximations, has remained a difficult challenge. In this paper, we introduce Kinematic Interpolants (KI), a novel framework that extends the seminal methodology of stochastic interpolants to infer nonlinear dynamics across multiple snapshots in phase space, i.e., observables with position and velocity measurements. Empirically, we demonstrate that across several scientific datasets, KI achieves superior reconstruction of the underlying dynamics compared to existing baselines while requiring only a fraction of the training time.
[LG-106] Predicting Privileged Information Effectiveness for On-Policy Self-Distillation
链接: https://arxiv.org/abs/2610.22512
作者: Kaijian Zou,Emiliano Penaloza,MohammadHossein Rezaei,Anisha Gunjal,Mohit Raghavendra,Lu Wang,Yunzhong He,Anas Mahmoud
类目: Machine Learning (cs.LG)
*备注:
Abstract:On-policy self-distillation (OPSD) enables large language models (LLMs) to learn from privileged information (PI) without a separate teacher model. However, prior work largely treats PI as a fixed design choice, leaving the effects of its content and construction poorly understood. This challenge is particularly important in multi-turn agentic settings, where trajectories contain actions, observations, and reasoning that can be transformed into PI in many ways. We systematically evaluate 16 PI configurations across three multi-turn agentic benchmarks and three student models that vary in scale and reasoning mode. We find that PI effectiveness depends on its content, generation source, and the student model. Information-dense trajectories and PI produced by more capable generators generally yield larger gains, whereas thinking models benefit less consistently. To assess PI before training, we compare PI utility with two KL-based metrics: sampled-token reverse KL ( D ) and error-step KL contrast ( \Delta d^-\mathrmerr ). The KL-based metrics track post-training performance more consistently than PI utility, which can be distorted by information leakage. Empirically, selecting each task’s PI by \Delta d^-\mathrmerr also improves post-training performance over random selection, most when the PI candidates differ widely. Finally, effective PI remains beneficial when combined with GRPO, allowing the combined objective to outperform a strong GRPO baseline. Overall, our results provide practical guidance for constructing and evaluating PI in OPSD training.
[LG-107] Generative Value Classifier
链接: https://arxiv.org/abs/2610.22497
作者: Alexander Swerdlow,Perry Dong,Chelsea Finn
类目: Machine Learning (cs.LG)
*备注:
Abstract:Q-learning is premised on learning a function to predict the value of a state-action pair, with the goal of using such a function to guide policy improvement. Most approaches learn this function from scratch using scalar or distributional return targets. This discriminative approach leaves another source of supervision for learning representations unused: the actions themselves. Rather than directly predict the return of a state-action pair, we ask the reverse question: how likely is the action under each possible return? This perspective provides a critic whose generative component is a return-conditioned policy, which can share the actor’s architecture. Training this policy supervises the critic with every action rather than with returns alone, and a squared-error loss anchors the critic’s Q-value to the bootstrapped target. We evaluate our approach on both offline and online RL tasks and show gains over leading discriminative methods, including distributional RL methods.
[LG-108] Lean4Agent icRL: Advancing Agent ic RL with Verifiable Multi-level Process Rewards
链接: https://arxiv.org/abs/2610.22448
作者: Ruida Wang,Gerald Friedland,Tong Zhang,Chris Kong
类目: Machine Learning (cs.LG)
*备注:
Abstract:Enhancing the agentic capabilities of Large Language Models (LLMs) to tackle complex software engineering (SWE) and general terminal-usage problems is a pressing challenge. Current state-of-the-art LLMs often go through large-scale agentic Reinforcement Learning (RL). However, they mostly rely on high-level outcome rewards, with minimal consideration of low-level toolcall and mid-level workflow-stage rewards, which makes the training less effective. To overcome this limitation, we propose Lean4AgenticRL, which, to the best of our knowledge, is the first framework that grounds multi-level process rewards in the unified Lean4 formalization schema for open-ended agentic RL. Lean4AgenticRL first builds a problem-specific workflow system that balances the agent’s own capability and controllability during execution, laying the foundation for further verification. Subsequently, our workflow formalization and multi-level reward annotation methods enable agent behavior verification within a general harness system. It provides multi-level rewards to support more effective agentic RL with the workflow-stage-normalized GRPO algorithm. Using Lean4AgenticRL, we train LAR-9B/27B, a family of models with enhanced agentic capability. Extensive experiments indicate that our model outperforms the leading model of similar size. LAR-27B achieves 77.3% on TerminalBench-2.1. On average, our models outperform the base model by 9.89% and surpass the second-best model of comparable size we know by 5.07%. These results indicate the potential of integrating formal methods in enhancing open-ended agentic RL.
[LG-109] Rank Measures of Fine-Tuning Updates Need a Noise Baseline
链接: https://arxiv.org/abs/2610.22438
作者: Muhammad Usman Altaf
类目: Machine Learning (cs.LG)
*备注: 24 pages, 6 figures, 15 tables. Under review at TMLR
Abstract:Studies of what fine-tuning does to a language model often summarise the weight update \Delta W by its rank. We compare the rank measures used for this against the same measures on a random matrix of the same shape, and against two harder references: the pretrained weights, and matched fine-tunes without alignment content. Measures that count directions put our fine-tuning updates only a little below noise; stable rank, computed from the same singular values, puts them far below. Both are right: an update has a few strong directions on top of a long tail. The pattern holds for thirteen released pairs up to 8B parameters, and it is not a mark of alignment: with training held fixed, instruction data does not stand out from Wikipedia or code, DPO gives the same update whether its preference labels are real or half swapped at random (one model, modest preference strength), and scrambled text gives the most concentrated update of all. Applied to a recent claim that post-training is superficial because the effective rank of the weights barely changes, the baselines show that a random update of the same size leaves that rank almost unchanged while wrecking the model, and that an edit built on the effective rank of the update has no measurable effect in our tests. Rank measures of weight updates need a noise baseline, and claims about alignment need a control without alignment content.
[LG-110] Generative adversarial networks with Monte Carlo dropout uncertainty quantification for structural health monitoring: Fibre Brag g grating data augmentation comparison in Hawk T1A aircraft
链接: https://arxiv.org/abs/2610.22436
作者: Alvaro Peiro,Marios Impraimakis
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注: 18 pages; 18 figures, published paper
Abstract:The signal generation capabilities of a novel data augmentation framework are examined in aerospace structures. First, the framework learns class-conditional temporal and spectral characteristics of the measured structural responses. It uses conditional generative adversarial network training and Monte Carlo dropout uncertainty quantification as a Bayesian approximation. This step enables stochastic signal synthesis and uncertainty quantification. Second, the generated responses are rigorously compared against experimental measurements across time, frequency, uncertainty, variability, and feature domains to assess physical realism, statistical consistency, and distributional fidelity. Finally, the synthetic data are evaluated to determine whether they should be incorporated into a supervised damage classification workflow. The framework is validated on the full-scale benchmark dataset from a BAE Systems Hawk T1A aircraft tested at the University of Sheffield’s Structural Dynamics Laboratory for Verification and Validation. The test involves acceleration, fibre Bragg grating strain, and nonlinear dynamic response measurements. The results demonstrate that the model accurately captures dominant structural dynamics, it guides on signal use for classification improvement based on uncertainty characteristics, and it provides meaningful uncertainty estimates linked to signal complexity and excitation characteristics. Importantly, the framework offers a realistic generative modelling plan as a robust digital twin component for data-driven structural health monitoring, supporting resilient decision-making for condition assessment in safety-critical aerospace systems.
[LG-111] BOLT: Multi-Bottleneck Bound-Guided Learning for Flexible Job-Shop Scheduling with Time Lags and Resource-Coupled Transport
链接: https://arxiv.org/abs/2610.22433
作者: Ziheng Zhang,Wei Zhang,Yunyi Zhao,Zuming Liu,Wei Zhou,Boon Ping Gan
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注: 12 pages main text plus 18 pages of supplementary material appended after the references (30 pages total). Code, data, and trained models: this https URL . Submitted to IEEE Transactions on Automation Science and Engineering
Abstract:Flexible job-shop scheduling becomes resource-coupled when a scarce vehicle fleet moves jobs between stations, because the schedule sets the transport waits. BOLT (bound-oriented learning for transport-coupled scheduling) builds a lower bound from chain, machine-capacity, and fleet-capacity relaxations. The bound equals the makespan at termination, so its per-step changes form a reward that sums to a constant minus the makespan. Because it never exceeds the attainable makespan, each candidate’s bound increase is a proven rise of the makespan floor, which the policy reads as a price. The root bound certifies a completed schedule’s suboptimality gap without a solver. On industry-motivated 10-module prefabricated-construction instances, explicit transport coupling outperforms a transport-penalty policy in every tested cell where loaded moves occupy at least 66% of fleet time and in none at or below 56%. Across five zero-shot 50- and 80-module production batches, BOLT outperforms the best dispatching-rule pair on four and the penalty policy on the three with long moves. In the three-vehicle 80-module batch, its mean makespan is 13.0% below the best pair and 4.2% below an hour-budget constraint-programming solver whose proven gap averages 34%. BOLT decides in 2.3-5.9 ms per event on one processor core and re-plans a disruption in seconds.
[LG-112] GPU-Accelerated Robust MA–RIS Pre-Optimization for Communication and Target Illumination
链接: https://arxiv.org/abs/2610.22431
作者: Tran Le Dung,Ho Viet Duc Luong
类目: Information Theory (cs.IT); Machine Learning (cs.LG)
*备注:
Abstract:Movable antennas (MAs) and reconfigurable intelligent surfaces (RISs) offer strong spatial control for integrated sensing and communication (ISAC), but repeatedly adapting MA positions and RIS phases to each channel realization can be impractical. This letter develops a GPU-accelerated robust pre-optimization framework that jointly selects slow-timescale MA geometry and RIS phases and returns a reusable Pareto set balancing expected worst-user rate and transmit-side worst-target illumination. The illumination objective measures directional transmitted power rather than complete echo-detection or parameter-estimation performance. The proposed Regime-Aware Multiobjective Pre-optimization (RAMP) preserves regime-specific Pareto populations, regulates cross-regime mating using Pareto-survival feedback, and constructs a worst-regime robust archive. Numerical results demonstrate competitive robust Pareto quality against the considered evolutionary baselines while avoiding online physical re-optimization.
[LG-113] CGDT-CESN: Correlation-Guided and Dynamics-Trained Clustered Echo State Network for Multivariate Time-Series Prediction
链接: https://arxiv.org/abs/2610.22430
作者: Hariharan Srinivasan,Masoom Bhargava,Sayan Gupta
类目: Machine Learning (cs.LG); Adaptation and Self-Organizing Systems (nlin.AO)
*备注:
Abstract:Accurate prediction of multivariate time series requires models that simultaneously capture variable-specific dynamics and cross-variable interactions. In this work, a Correlation-Guided and Dynamics-Trained Clustered Echo State Network (CGDT-CESN) is proposed, which comprises a modular reservoir architecture with each cluster distinctly specializing in one of the variables in the multivariate time series. An innovative reservoir design is proposed in which inter-cluster connections are enabled by empirically observed correlations among variables, providing a data-driven coupling mechanism. Apart from conventional readout training, inter-cluster connection weights informed by correlations are trained, which in turn allows the reservoir to capture time-varying cross-variable dependencies without sacrificing computational efficiency and further improves forecasting accuracy. The prediction performance of CGDT-CESN is found to be superior to that of the conventional ESN when evaluated on two fundamentally different datasets: Twitter information cascade evolution and solar wind forecasting. Also, the improvement in performance is preserved across multiple prediction horizons, even when presented with a low volume of data. By unifying cluster-specific specialization with learned inter-variable coupling, CGDT-CESN offers a principled, data-driven route to multivariate time series prediction.
[LG-114] rust but Verify: Designing Region-Inhomogeneous Forward Dynamics for Discrete Diffusion STOC NEURIPS2026
链接: https://arxiv.org/abs/2610.22380
作者: Gunmay Jhingran
类目: Machine Learning (cs.LG)
*备注: Accepted to 40th Conference on Neural Information Processing Systems (NeurIPS 2026). Workshop: AI for Stochastic Dynamics. 10 pages
Abstract:Continuous diffusion treats the forward process as a design space; discrete language diffusion often uses absorbing corruption, leaving context fully trusted throughout generation. This makes trust binary: corrupted context is copied as ground truth, propagating errors into generated spans. We reopen this design space by combining absorbing and uniform-mixing kernels with region-dependent rates. Missing positions follow absorbing dynamics, while context undergoes low-rate random substitution, exposing the model to observed but uncertain tokens during training. At inference, a heuristic sampler re-injects noise into suspicious context under an annealed intensity and freezes context updates late in the trajectory. An 8,000-step finetune of a pretrained 170M-parameter masked diffusion model takes under five hours on a 6 GB laptop GPU. The base model assigns roughly 50% probability to observed tokens even when they are uniform-random garbage. For uniform-random corruption, finetuning detects corrupted context with 0.96 recall and 0.91 precision, improves infill token-F1 by 3.8 points at 10% corruption and 6.4 points at 20%, and removes the accuracy dip next to corruptions. Clean conditioning shows no measurable regression, with a false-edit rate below 0.4%. Gains for plausible corruptions are weaker: about 0.5 F1 points at 10% corruption, without statistical significance. Our results suggest that detectability and damage depend jointly on departure from the data conditional: nearby noise is hard to detect and semantically mild, while distant noise is conspicuous and destructive. We interpret absorbing diffusion as the infinite-contrast limit of a trust axis, analogous to the signal-to-noise gap between known and unknown regions in diffusion inpainting. In our setting of a 170M model and a single text domain, intermediate trust is reached by changing forward dynamics rather than architecture.
[LG-115] Memory Bandwidth Not FLOPs: Profiling and Accelerating Query-Based Segmentation Decoders on Edge GPUs BMVC
链接: https://arxiv.org/abs/2610.22377
作者: Oliver Busemann(1 and 2),Torben Töniges(2),Martin Atzmueller(3 and 4) ((1) Osnabrück University, (2) CLAAS E-Systems GmbH, (3) Semantic Information Systems Group, Osnabrück University, (4) German Research Center for Artificial Intelligence (DFKI))
类目: Hardware Architecture (cs.AR); Machine Learning (cs.LG); Performance (cs.PF); Image and Video Processing (eess.IV)
*备注: Accepted at the British Machine Vision Conference (BMVC) 2026. Supplementary material included
Abstract:Query-based segmentation models achieve strong performance even under real-time inference on workstation GPUs. On edge devices, however, we show their transformer decoders are deeply memory-bandwidth-bound, a bottleneck invisible to standard FLOP analysis that worsens with input resolution and limited DRAM bandwidth. An FP16 roofline analysis with tensor-core GEMMs active and hardware DRAM counters shows that the non-tensor-core aggregates of FastInst and Mask2Former reach only 2.5% to 6% of the CUDA-core ridge yet account for 53% to 70% of decoder time and 61% to 71% of decoder bytes. The largest individual kernels sustain up to 90% of nominal peak DRAM bandwidth, effectively saturating the memory bus. The bottleneck is therefore not primarily low GEMM intensity, but cross-kernel activation traffic. CompactInst addresses this bottleneck using parameter-free spatial pooling at 4\times , which reduces the bytes moved within the transformer decoder, while a two-convolution refinement head recovers spatial detail. Under TensorRT FP16 on Jetson Orin Nano, CompactInst delivers decoder speedups from 2.35\times to 4.74\times and model speedups from 1.57\times to 2.62\times as input resolution grows, while closely matching accuracy ( -0.4 AP on COCO val2017, +0.2 AP on Cityscapes test). A heavier variant with 2.57\times more decoder FLOPs still runs its decoder 1.74\times faster on the most bandwidth-constrained device, confirming at the deployment level that decoder bytes, not FLOPs, govern latency.
[LG-116] ask-aligned spectral inductive biases in parameterized quantum circuits
链接: https://arxiv.org/abs/2610.23834
作者: Felix Frohnert,Adrián Pérez-Salinas,Sebastian Lange,Marcus Schwaebisch,Evert van Nieuwenburg,Vedran Dunjko
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG)
*备注:
Abstract:Parameterized quantum circuits are widely investigated in quantum machine learning, yet how their architecture can provide task-aligned inductive biases remains poorly understood. We study data reuploading quantum circuits with two distinct inputs: a signal that varies between samples and a fixed, task-defined spectral query that specifies how the signal should be probed. We show that encoding the query can change the circuit’s hypothesis class over the signal by modulating how its existing signal frequencies are combined, without introducing new sample-dependent information. This mechanism matches tasks built on nonuniform Fourier transforms, whose evaluation frequencies specify the query and whose sampling positions guide its encoding. We examine this connection in a classification task inspired by synthetic aperture radar, where acquisition geometry determines a query for classifying unfocused spectral patches before image formation. Across small-scale spectral and radar-like tasks, aligned queries improve learning from limited data over parameter-matched circuits with mismatched or absent queries. These results provide a route from known spectral task structure to the design and validation of inductive biases in parameterized quantum circuits.
[LG-117] Quality-of-Transmission Estimation in Quantum Networks: Compute Simulate or Learn?
链接: https://arxiv.org/abs/2610.23827
作者: Anis Bekri,Amar Abane,Van Sy Mai,Abdella Battou,Saddek Bensalem
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG); Networking and Internet Architecture (cs.NI)
*备注:
Abstract:Future quantum networks will rely on control-plane decisions such as routing, scheduling, entanglement provisioning, purification, and memory allocation. Each decision depends on a timely and reliable estimate of the quality of the entanglement that a candidate path can deliver, with end-to-end fidelity as a central metric. Inspired by the optical-networking experience with quality-of-transmission (QoT) estimation, this article asks when a quantum controller should compute fidelity analytically, simulate it, or learn a surrogate model. The optical analogy is useful but incomplete: optical QoT estimation often predicts a scalar transmission margin, whereas quantum QoT estimation must account for probabilistic entanglement generation, random memory waiting times, decoherence, state consumption, and measurement constraints that can make fidelity a stochastic target. We propose a compute/simulate/learn framework for quantum QoT estimation and illustrate it with a simulation-based proof of concept. Our study shows that analytical composition can be sufficient in simple regimes, that simulation becomes necessary when stochastic memory effects shape the delivered fidelity distribution, and that lightweight learned surrogates can approximate simulator labels at much lower runtime cost for online control. We close with open challenges for deployment, including stochastic targets, sim-to-real transfer, data-generation cost, physics-informed learning, and generalization beyond individual paths to full network topologies.
[LG-118] ENFR: Evidence Networks with Flow-Matched References for Amortized Simulation-Based Model Inference
链接: https://arxiv.org/abs/2610.23825
作者: Kai Lehman,Jameson Dong,Carolina Cuesta-Lazaro,Niall Jeffrey,Benjamin D. Wandelt,Francisco Villaescusa-Navarro,Sven Krippendorf,Jochen Weller
类目: Machine Learning (stat.ML); Cosmology and Nongalactic Astrophysics (astro-ph.CO); Machine Learning (cs.LG)
*备注: 28 pages, 6 figures
Abstract:The evidence, \mathcalZ = p(x | \mathcalM) , of a data-generating model \mathcalM is a powerful tool in Bayesian frameworks: it enables model inference, model misspecification tests, and goodness-of-fit assessment as an alternative to standard discrepancy statistics, e.g. \chi^2 . While versatile, the evidence is notoriously difficult to compute, even when the likelihood is available. In both likelihood-based and simulation-based inference, estimating \mathcalZ provides an immediate solution to high-dimensional goodness-of-fit or model misspecification tests: using \mathcalZ(x) as the test statistic induces a minimum-volume level- \alpha test. Such tests are essential to assess model fidelity, which is necessary to trust simulation-based parameter inference. Existing learned approaches largely target Bayes factors, which only measure evidence ratios. Instead, we provide accurate, fast, and amortized estimates of the absolute evidence. We present ENFR: Evidence Networks with Flow-matched References. Building on the Evidence Network framework of Jeffrey Wandelt (2024), the proposed method first learns Bayes factors with respect to a flow-matched reference, from which the absolute evidence is recovered. We show improvement in estimation of \mathcalZ over the flow-based baseline in all applications considered and the best performance of all density estimation methods considered in our benchmarks. We furthermore introduce and provide the proof for an architecture that jointly learns the individual evidences of a set of models \mathcalM_i\ . We demonstrate the performance and scaling behavior of these approaches on data from a variety of scientific applications.
[LG-119] Learning without relearning physical symmetry: an SO(3)-equivariant convolutional network (CNN) on the sphere
链接: https://arxiv.org/abs/2610.23821
作者: Joshua Williamson,Niall Jeffrey,Lorne Whiteway
类目: Machine Learning (stat.ML); Cosmology and Nongalactic Astrophysics (astro-ph.CO); Machine Learning (cs.LG)
*备注:
Abstract:Many of the most important datasets in the physical sciences live natively on the sphere, and these data often carry no preferred position or orientation. Using methods that do not reflect this symmetry leads to far higher simulation requirements, which can be catastrophic for scientific analyses (forcing reliance on cheap, imperfect models that bias results). We present a novel spherical implementation of a gauge equivariant mesh architecture that is specifically designed for the Hierarchical Equal Area isoLatitude Pixelization (\healpix) of the sphere. The \mathrmSO(3) -equivariant convolution is built from steerable filters, parallel transport and spin-weighted features. The architecture supplies the exact symmetry, so the network does not spend training data learning it. The convolution operations are constructed to give provably exact equivariance, given a band-limit. Our full CNNs can achieve SO(3)-invariance error at per-cent-level and lower, which is controlled through well-understood hyperparameters. On benchmarks, our technique gives the most accurate dense gradient regression and competitive cosmic microwave background foreground removal and denoising. The improved simulation efficiency matters most in simulation-based inference (SBI), where high-fidelity simulations are expensive and neural compressors require large training sets, so a fixed budget forces analyses with less accurate simulators. In cosmology lensing SBI, our technique yields the tightest posteriors at matched simulation count. This approach is a novel method that will enable new convolutional neural network analyses, including generative and inference tasks, for processes and data on the sphere.
[LG-120] ask-aware Kernel Stein certificates for Rényi differential privacy
链接: https://arxiv.org/abs/2610.23798
作者: Leoni Carla Wirth,Tara Trautwein,Gesine Reinert
类目: Machine Learning (stat.ML); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注:
Abstract:Rènyi differential privacy, based on Rènyi divergence, has become a notion of choice for privacy guarantees. However, outside the Gaussian setting, Rènyi divergence can be difficult to compute. In particular, Rènyi differential privacy guarantees that exploit task-specific information, as in Rènyi Pufferfish privacy, can be challenging. Here we use kernel Stein discrepancies based on escort distributions to provide certificates for Rènyi differential privacy. By designing the kernel to focus on the relevant privacy directions, we also derive task-aware privacy certificates. We provide exact identities for Gaussian and Laplace mechanisms and bounds for exponential mechanisms, with separate task-aware guarantees. A real data set example illustrates how the protected comparison set and the released subspace jointly determine the resulting certificates.
[LG-121] Parametric Resonance in Primal-Dual Learning for Slowly Non-Stationary Constrained MDPs
链接: https://arxiv.org/abs/2610.23788
作者: Maxime Baelde
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Systems and Control (eess.SY); Optimization and Control (math.OC)
*备注: 33 pages, 3 figures. Code: this https URL
Abstract:Primal-dual methods for constrained Markov decision processes are analyzed under a two-timescale assumption, with non-stationarity entering through two channels. In the additive channel, where the constraint level oscillates, the dual oscillation saturates at a floor set by the moving optimum and is amplified linearly by the timescale ratio. In the parametric channel, where the constraint sensitivity oscillates, a coherent modulation carries no resonant forcing at any depth: the dual update’s stiffness modulation cancels the primal update’s damping modulation. Instability instead comes from a mismatch between the constraint estimate used by the primal update and the measurement used by the dual update, producing a parametric resonance at twice the pair’s natural frequency. The tolerated mismatch grows with the square root of the forcing bandwidth and collapses with the square root of the dual-to-primal step-size ratio. These results hold on the tangent linear-quadratic reduction around an active saddle, for one constraint and for K constraints under a common modulation and an isotropy condition. Driving both updates from the same estimate removes the resonant forcing but introduces a stationary offset equal to the estimator’s bias. In a Lagrangian soft actor-critic agent, the multiplier locks onto the sub-harmonic at the timescale ratios its identified linear loop predicts and a coherent cost critic suppresses the lock-in, while a proportional (PID) term, predicted to leave the lock-in in place, suppresses most of it through the projection of the multiplier.
[LG-122] Sampling the XY model with an attention-based neural network
链接: https://arxiv.org/abs/2610.23783
作者: Piotr Białas,Christian D. Morales,Tomasz Stebel,Dawid Zapolski
类目: Disordered Systems and Neural Networks (cond-mat.dis-nn); Statistical Mechanics (cond-mat.stat-mech); Machine Learning (cs.LG); High Energy Physics - Lattice (hep-lat)
*备注: 19 pages, 6 figures
Abstract:We present a transformer-based variational autoregressive network for sampling lattice systems with continuous degrees of freedom and apply it to the two-dimensional XY model near the Berezinskii-Kosterlitz-Thouless transition. The architecture is combined with physics-informed ingredients: locality-informed attention and approximate probabilities. The resulting sampler significantly outperforms existing neural samplers. We train the model on lattices up to 96\times96 , verify with a mode-dropping estimator that no parts of the configuration space are missed, and reproduce the energy, magnetization and helicity modulus obtained with the traditional Markov Chain Monte Carlo algorithm.
[LG-123] Mesoscopic learning dynamics of extensive-width neural networks
链接: https://arxiv.org/abs/2610.23779
作者: Shu Zhou,Jean Barbier,Sebastian Goldt
类目: Machine Learning (stat.ML); Disordered Systems and Neural Networks (cond-mat.dis-nn); Statistical Mechanics (cond-mat.stat-mech); Machine Learning (cs.LG)
*备注:
Abstract:Understanding how stochastic gradient descent learns internal representations in wide neural networks trained on rich target functions is a central problem in machine learning. We study this question in two-layer neural networks learning target functions represented by networks of the same architecture, with arbitrary nonlinearity, and with network width and input dimension growing proportionally. In this regime, standard approaches for describing the learning dynamics require tracking a diverging number of overlaps, which measure correlations between learned and target features, leaving no obvious description in terms of a few order parameters. Here, we develop a mesoscopic description of learning by characterizing the distribution of overlaps. In the early stages of learning, we find that a Gaussian description of overlap statistics reduces the learning dynamics to a few collective variables that track learning of the target function’s linear and quadratic components. We then estimate the timescale of specialization, when Gaussian closure breaks down and individual learned features begin to strongly correlate with specific target features. We find that specialization in extensive-rank networks is accelerated via a quadratic shielding mechanism, whereby learning the linear and quadratic components of the target function filters gradient fluctuations. We finally describe how the activation function determines the pattern of specialization and characterize the subsequent refinement of learned features. Together, these results show how collective learning gives way to individual feature discovery and clarify when learning in large neural networks admits a simple statistical description.
[LG-124] From Flow-Box Invariants to Global Conservation Laws: Machine Learning-Assisted Detection of Singularities and Near-Integrability
链接: https://arxiv.org/abs/2610.23760
作者: Jimmie Adriazola,Ioannis G. Kevrekidis,Panayotis G. Kevrekidis,Wei Zhu
类目: Chaotic Dynamics (nlin.CD); Machine Learning (cs.LG)
*备注:
Abstract:Conservation laws reveal invariant structure in dynamical systems and constrain their evolution to lower-dimensional level sets. Locally, however, per the flow-box theorem, every d -dimensional system generically admits d-1 functionally independent conservation laws near a regular point even though, generically, only a few such laws persist globally. Data-driven methods typically learn conservation laws on a prescribed region of phase space from which training samples are drawn, leaving their behavior beyond that region unresolved. We formulate this local-to-global continuation problem and develop an expanding-domain neural deflation framework that tracks the number of independent conservation laws over nested domains. The method recovers the expected d-1 local invariants and detects their successive breakdown as the domain expands, relaxing to the expected global conservation law limit. It distinguishes the local invariants (associated with Frobenius integrability) from persistent global structure and shows that the domain scale over which additional laws survive reflects proximity to Liouville integrability. Pointwise losses can localize isolated obstructions, whereas in more complicated dynamics the breakdown of our loss functions is gradual and spatially diffuse. The framework thus provides a data-driven approach for studying the persistence and breakdown of (so-far-learned) conservation laws across phase space.
[LG-125] ECG-SDI: Single-lead Electrocardiogram Enables Continuous Sleep Depth Index Profiling and Scalable Digital Biomarkers for Sleep Health
链接: https://arxiv.org/abs/2610.23727
作者: Donglin Xie,Qingshuo Zhao,Songchi Zhou,Haoyu Wang,Jingyu Wang,Yuxi Zhou,Thomas Penzel,Shenda Hong
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注:
Abstract:Conventional sleep staging provides a coarse-grained representation that obscures within-stage depth variation. We developed ECG-SDI, a single-lead electrocardiography (ECG) framework for continuous sleep-depth profiling and digital biomarker extraction. Using 12,392 overnight polysomnographic recordings, we trained and validated a two-stage deep learning model supervised by conventional sleep-stage labels. Correlations with ordered stages excluding rapid eye movement (REM) sleep were 0.828 in the pooled internal test set and 0.848 in the external Cleveland Family Study; corresponding REM classification AUROCs were 0.954 and 0.977. Overnight trajectories captured within-stage variation. Biomarker-based phenotyping identified a disturbed subtype associated with poorer subjective sleep quality and higher all-cause mortality in the Sleep Heart Health Study (adjusted hazard ratio, 1.20; 95% CI, 1.05-1.38). Among 507 participants with paired recordings before and after continuous positive airway pressure (CPAP), normalized sleep-depth area increased by 6.1% and shallow-sleep burden below 0.3 decreased by 16.0%. Exploratory analyses linked reductions in approximate entropy and depth variability to reduced oxygen desaturation and increased sleep efficiency, respectively (Spearman rho = 0.24 and 0.12); both associations remained positive after covariate adjustment. Mean ECG-SDI within N2 increased by 0.017 (95% CI, 0.004-0.030). A wearable proof-of-concept illustrated the acquisition-to-report workflow. These findings support ECG-based sleep phenotyping and further evaluation of CPAP response monitoring at home.
[LG-126] Expressivity of IQP Feature-Map Quantum Kernels via Reproducing Kernel Hilbert Spaces
链接: https://arxiv.org/abs/2610.23692
作者: Valeria Repetto,Stefano Melchionna,Romina D’Aurizio
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG)
*备注: 19 pages, 11 Figures , Supplementary
Abstract:Quantum kernels are among the most promising methods in quantum machine learning, providing a natural way to embed data into high-dimensional Hilbert spaces via quantum circuits and capturing non-linear relationships in data. Their learning behaviour is governed by the function class they represent, which is formally characterized by the structure of the associated Reproducing Kernel Hilbert Space (RKHS). However, a theoretical characterization of this function class, and of how it depends on the choice of feature map and its hyperparameters, is still lacking. In this work, we focus on the Instantaneous Quantum Polynomial (IQP) feature map and derive a Fourier representation of the associated RKHS generators as finite sums of trigonometric functions evaluated on polynomially transformed input variables, with a finite set of accessible frequencies determined by the circuit construction and scaled by the bandwidth hyperparameter. This representation identifies three bandwidth regimes with qualitatively distinct expressive properties from localised, through multimodal, to periodic. Numerical experiments on synthetic datasets with controlled periodic and multimodal structure, across data dimensions and feature-map configurations, reproduce the three regimes in both supervised classification and unsupervised anomaly detection, benchmarked against a classical Gaussian kernel. These results make the relationship between expressivity and data structure explicit for IQP kernels: expressive power translates into a practical benefit only when the accessible frequencies match with the structure of the data, which provides a criterion for selecting the bandwidth when the periodic or multimodal structure of data is known.
[LG-127] Scalable long-range machine learning interatomic potentials with electrical response for atomistic material modeling
链接: https://arxiv.org/abs/2610.23643
作者: Yicheng Chen,Xiaoyun Yu,Hyejeong Song,Xintian Wang,Zihui Li,Zixi Xia,Junmin Chen,Peichen Zhong
类目: Materials Science (cond-mat.mtrl-sci); Machine Learning (cs.LG)
*备注:
Abstract:Accurate atomistic simulations of materials under electric fields require long-range electrostatics and electrical-response modeling at length scales beyond the reach of conventional first-principles methods. Yet these capabilities remain difficult to reconcile with the scalability of machine-learning interatomic potentials (MLIPs). Strictly local models enable efficient spatial parallelization but neglect long-range interactions, whereas existing electrostatic extensions introduce global dependencies that hinder distributed computation. Here, we overcome this limitation by analytically separating the long-range electrostatic forces and Born effective charges (BECs) into a global electrostatics calculation and strictly local derivatives of learned latent charges. This formulation preserves the spatial decomposability of equivariant MLIPs while enabling distributed evaluation of energies, forces, stresses, and BECs across different boundary conditions. Implemented as DPA4C-LR on the \textttnvalchemi platform, the framework scales to 5.2 million atoms on four GPUs at approximately twice the computational cost of its short-range backbone. Trained on energies and forces, DPA4C-LR accurately infers BECs and reproduces the field-dependent infrared spectra of water. We further demonstrate its capabilities by recovering the characteristic slim polarization hysteresis of relaxor \cePb(Mg_1/3Nb_2/3)O3, which is obscured by finite-size effects in smaller simulations; predicting a room-temperature field-driven ionic conductivity of 6.8 \pm 0.7 ~mS/cm in amorphous \ceLi_0.9NbO_0.9Cl_4.1; and modeling interfacial thermal transport across Si/diamond interfaces with slab-corrected electrostatics. By combining long-range physics, electrical-response inference, and distributed computation, this framework extends predictive atomistic modeling toward device-relevant length scales.
[LG-128] A general recipe for data-adaptive anytime-valid envelopes for iterative algorithms
链接: https://arxiv.org/abs/2610.23570
作者: Soham Bonnerjee,Aaditya Ramdas
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Optimization and Control (math.OC); Methodology (stat.ME)
*备注: 83 pages, 4 figures
Abstract:Most convergence guarantees for iterative stochastic algorithms are stated at a fixed iteration count and do not justify stopping at a data-dependent time. Recent work has constructed anytime-valid envelopes that do, but deterministic envelopes are typically too conservative for practical stopping, and existing data-adaptive ones are derived case-by-case, each coupling a specific algorithm to a specific noise model and boundary construction. This paper provides a general recipe for constructing such anytime-valid data-adaptive boundaries for a large class of iterative algorithms. Performance measures of many of these algorithms satisfy a one-step recursion that reduces the entire error analysis to a single general requirement: time-uniform concentration for predictably weighted sums of noise. We formalize this through a strongly adapted sub- \psi condition, and derive off-the-shelf boundary constructions via continuous and discrete mixtures, stitching, and product betting martingales. We show how these boundaries can be directly coupled with a separate algorithmic analysis that derives the one-step recursion to provide a class of readymade anytime-valid envelopes across diverse settings including stochastic gradient descent (SGD), temporal difference learning, and Bures-Wasserstein SGD. In particular, for SGD, our framework either recovers or significantly sharpens every existing envelope we are aware of. Our constructions come with explicit implementation details and are substantiated by numerical experiments.
[LG-129] Beyond Latent Personalization: Integrating HRV-Conditioned Personalization via LLM for Interpretable Stress Detection ICASSP2027
链接: https://arxiv.org/abs/2610.23555
作者: Wei-Hsiang Ma,Ya-Tse Wu,Hsing-Hang Chou,Chi-Chun Lee
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注: Submitted to ICASSP 2027
Abstract:Stress detection based on heart rate variability (HRV) has received increasing attention. However, existing personalized approaches integrate personal attributes with HRV through latent representations, limiting the interpretability of their interactions. In this work, we propose an LLM-enhanced personalized stress detection framework that integrates HRV features and personal attributes through LLM-based semantic interpretation. The proposed framework achieves a 29.9% relative MCC improvement over the strongest baseline on TILES-2018 while providing human-readable descriptions of individual physiological responses. The results highlight the potential of LLM-based semantic interpretation for achieving both effective and interpretable personalized stress detection.
[LG-130] Causal Discovery beyond Conditional Independence: Identifiability Algorithms and Statistical Inference for Directed Acyclic and Cyclic Graphs
链接: https://arxiv.org/abs/2610.23543
作者: Yang Ni
类目: Methodology (stat.ME); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Learning causal structure from data requires assumptions about mechanisms, noises, observation schemes, and interventions. Historically, causal discovery has relied primarily on conditional independence. This review focuses on modern causal discovery methods that exploit properties of distributions beyond conditional independence. Our organizing principle separates the source of identification from the statistical estimator and its computational implementation. This perspective connects conditional-independence methods with distributional asymmetries, conditional-moment restrictions, and variation across environments or individuals. We examine discovery under latent confounding and feedback, distinguishing causal information about observed variables from reconstruction of latent structure and emphasizing the additional semantics required by cyclic models. We also distinguish population identification, frequentist and Bayesian consistency, uncertainty quantification, and algorithmic guarantees. Throughout, examples show why the same observational distribution can support different conclusions under different model classes. We conclude with guidance for comparing methods at their identifiable targets and a research agenda for uncertainty, heterogeneous populations, non-scalar variables, and the intersection of discrete data, feedback, and hidden confounding.
[LG-131] Adaptive conformal prediction on the Grassmannian
链接: https://arxiv.org/abs/2610.23508
作者: Dimitrios Loukrezis,Zachary J. Grey,Dimitris G. Giovanis
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 21 pages, 6 figures, 3 tables
Abstract:We introduce a conformal prediction framework for regression with Grassmann-valued responses. The proposed method constructs prediction regions in the form of metric balls directly on the Grassmann manifold, using basis-invariant geometric distances as non-conformity measures and scaling the corresponding non-conformity scores through a learned prediction difficulty estimator. This yields locally adaptive, input-dependent prediction regions that respect the underlying manifold geometry, while retaining the finite-sample marginal coverage guarantee of split conformal prediction under exchangeability. We evaluate several Grassmannian distance metrics and compare their prediction-region efficiency on synthetic benchmarks and an airfoil inverse shape-design problem. The results highlight the advantages of geometry-aware non-conformity scores, while local adaptation improves empirical coverage uniformity across prediction difficulty levels and, in strongly heteroscedastic settings, can substantially reduce prediction-region size while maintaining the target coverage.
[LG-132] owards optimal selection schedules for online active learning
链接: https://arxiv.org/abs/2610.23477
作者: Hugo Cui,Francesca Mignacco,Francesco Mori
类目: Machine Learning (stat.ML); Disordered Systems and Neural Networks (cond-mat.dis-nn); Machine Learning (cs.LG)
*备注:
Abstract:We consider the problem of online linear classification, where a stream of unlabeled samples is available, and annotation is costly. We investigate the optimal query policy, that prescribes whether and when a sample should be labeled and used for a training step, or instead discarded, so as to minimize a given objective. We focus on the realizable case, under Gaussian design, in the high-dimensional regime where data dimension and number of unlabeled samples grow proportionally. We consider two objectives: 1) the generalization error under a labeling budget, and 2) a cumulative cost function where mistakes have unit cost and queries cost \Delta . We first show that the myopic policies that greedily minimize at all times the increment in the objective correspond to either small or intermediate-margin policies, that prescribe the selection of samples whose margins fall within certain intervals that shrink with time. Within this class, we numerically ascertain the optimal schedules for the selection thresholds that minimize the final values of the objectives. Taking inspiration from the myopic policy for the cumulative cost, we design a very simple query policy schedule that we compare to other proposed heuristics, on synthetic data and real medical image datasets. Finally, we discuss myopic policies for two-layer neural networks, and exemplify how naive margin-based policies can cease to be optimal.
[LG-133] he Loss or the Architecture? What Matters in Survival Analysis with Competing Risks
链接: https://arxiv.org/abs/2610.23322
作者: Julie Alberge,Gaël Varoquaux,Judith Abécassis
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Often, \empheg in healthcare and predictive maintenance, what matters is which of several competing causes will strike first, and with what probability: survival analysis with competing risks. Corresponding deep learning methods are developed as bundles: a new architecture together with a new training loss, tested on one to five datasets against a handful of baselines. Thus, a state-of-the-art result cannot be attributed to the architecture, the loss, or the evaluation protocol. The coupling is not accidental: with one incidence curve per cause, a network’s head is tied to the losses expressible on it. We vary the two axes independently, three losses, two of them strictly proper, on four backbones, across two modalities that tease the axes apart: 20 tabular and 3 pathology-image datasets. One loss wins in both: modelling cumulative incidence directly under censoring weighting. It brings twice as much as benefits a good choice of backbone on competing-risks data, and no more than it on single-event data: the loss is decisive for competing risks. The backbone matters too, and must match the modality: tabular networks lead on tables, pretrained vision encoders on images. The two choices compound rather than substitute. Proper scoring rules reveal these dynamics; the concordance index (C-index), which is not proper, does not separate models: neither axis moves it, not even tuning on the loss built to improve it. This is an avenue of progress: composed, the two give a model nobody has tried, RealMLP or TabM trained with the cumulative-incidence loss, which takes the best Brier score in our benchmark.
[LG-134] How the elastic axis folds the transonic flutter boundary: shock energy transfer and conformally calibrated active learning
链接: https://arxiv.org/abs/2610.23298
作者: Ashray Saxena,Sharad Shrivastava
类目: Fluid Dynamics (physics.flu-dyn); Machine Learning (cs.LG)
*备注: 18 pages, 14 figures, 2 tables. Code and data: this https URL
Abstract:Isogai’s case A, a pitch-plunge aerofoil whose elastic axis lies half a chord ahead of the leading edge, has a folded transonic flutter boundary in inviscid flow: near the bottom of the transonic dip several flutter speeds coexist at one Mach number. We map this boundary, and the family obtained by moving the elastic axis, with coupled unsteady Euler simulations, a Gaussian-process surrogate of the damping coefficient trained by batch active learning, and split conformal prediction on independent random simulations, which turns the surrogate into stability calls with a distribution-free error rate. Three failure modes of time-domain damping labels that leave no trace in the residuals are documented, and the half-degree mean incidence left by the common one-degree start is shown to raise the flutter speed at the dip by 2.8%, more than any numerical setting. With 80 simulations the learned case A boundary labels 64 of 65 independent runs correctly, and calibrated calls are wrong in at most 1.6% of cases. Over Mach number, elastic axis and flutter speed index, 193 simulations and 40 calibration runs show that the fold exists only for elastic axes more than about 0.4 chord ahead of the leading edge; it closes between 0.39 and 0.33 chord ahead, on two meshes. Chordwise energy budgets, closed against the solver loads and the growth rate within a few per cent, explain the boundary. The lower branch is crossed through the reduced frequency at unchanged shock phase. On the upper branch the shock is pinned at the trailing edge, and moving the elastic axis aft brings the pivot of the flutter mode towards the section, so that the damping of the supersonic surface falls faster than the residual shock input and the fold closes.
[LG-135] Learning angles for quantum approximate optimization
链接: https://arxiv.org/abs/2610.23296
作者: Johannes Jakubik,Isabelle Wittmann,Zaheed Gaffoor,Gciniwe Simphiwe Baloyi,Craig Mahlasi,Etienne Vos,Thomas Brunschwiler,Daniel J. Egger
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG)
*备注:
Abstract:The Quantum Approximate Optimization Algorithm (QAOA) and its variants are leading algorithms to help solve challenging combinatorial optimization problems. QAOA maps these problems to Ising Hamiltonians such that their ground state is the desired solution. Candidate solutions are then drawn from a variational quantum circuit. The performance of QAOA hinges on a set of variational angles. However, finding good angles is an instance-specific and computationally expensive non-convex classical optimization task. Here, we learn the mapping from the problem instance to high-quality QAOA angles. Our learned models replace the per-instance optimization, that we call the solver, with a single forward pass that is several orders of magnitude faster than the solver with a near identical quality. Our models further generalize to unseen topologies such as Barabási-Albert and 2D grid graphs. Experiments on quantum hardware confirm that QAOA samples with angles produced by our open-source models are indistinguishable from solver angles. Crucially, models sensitive to the topology of the problem instance increase the quality of the QAOA angles. Our work amortizes a major bottleneck in QAOA by an up-front data generation and training cost. Importantly, this cost can be amortized over time in settings where many similar problems are optimized on a regular basis, as is often the case in applications for industry.
[LG-136] Information-Geometric Blind Spots of Conformal Test Martingales
链接: https://arxiv.org/abs/2610.23131
作者: Johan Hallberg Szabadváry
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST)
*备注: 9 pages, 5 figures
Abstract:Conformal test martingales (CTMs) are e-processes for sequentially testing exchangeability via p-values produced by a conformity function A . However, any distribution shift that preserves the push-forward P\circ A^-1 leaves the p-value distribution invariant, rendering it invisible to every CTM based on A . We formalise this blind spot through the \emph A -equivalence class [P]_A and show that, for any target distribution T , there is a unique closest element Q^\ast\in[P]_A in KL divergence—the \emph A -cryptic I-projection—with explicit density q^\ast(z) = t(z),p_S(A(z))/t_S(A(z)) and a Pythagorean decomposition into detectable and cryptic components. When A is the Neyman-Pearson likelihood ratio, the cryptic component vanishes, connecting our result to optimal e-value design. We demonstrate the theory on high-dimensional image data by empirically constructing this I-projection. By conditionally resampling EMNIST letters to identically match the conformity scores of MNIST digits, we demonstrate that a distribution can be shifted to a disjoint support while leaving the test martingale mathematically blinded.
[LG-137] Distinguishability and Recovery for a Doubly Sparse Spiked Wigner Model
链接: https://arxiv.org/abs/2610.23033
作者: Ioana Dumitriu,JD Flynn,Zhichao Wang
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 50 pages, 9 figures, 8 tables
Abstract:We study distinguishability and recovery of a sparse rank-one signal in a spiked Wigner model with independently sparse noise through a variety of algorithms and parameter regimes. The signal and noise have separate sparsity parameters, while the signal-to-noise ratio (SNR) remains of constant order as the matrix size grows. For sub-Gaussian noise and spikes whose nonzero coordinates are bounded away from zero, a leave-one-out quadratic thresholding procedure (a generalization of Ding et al., 2024) identifies the support. Sign synchronization yields exact recovery for a Rademacher spike, and spectral refinement yields strong recovery. The quadratic thresholding search parameter gives an explicit tradeoff between signal sparsity, noise sparsity, and computational cost. We also study extensions to sub-Gaussian spike priors through diagonal and quadratic thresholding. For Rademacher spikes with continuous nonzero noise, we develop an identical search algorithm that exploits unblurred signal entries, the existence of which permits distinguishability; connectivity of the graph they reveal provides a sufficient condition for exact recovery. These guarantees do not require moment assumptions on the continuous noise distribution. Together with our previous PCA results (Dumitriu et al., 2026), our analysis covers almost the entire space of parameters and regimes for distinguishing and recovering in our doubly sparse spiked Wigner model.
[LG-138] Conformal Policy Selection: When to Deploy AI Decisions for Targeted Intervention
链接: https://arxiv.org/abs/2610.22930
作者: Junyu Cao,Ruijiang Gao,Maytal Saar-Tsechansky,Mingzhang Yin
类目: Methodology (stat.ME); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 28 pages, 6 figures, 4 tables
Abstract:Artificial-intelligence (AI) policies are increasingly used to make consequential decisions in medicine, finance, education, and public services, yet deploying them responsibly requires knowing not only whether a policy helps on average, but for whom. Existing tools for off-policy evaluation and safe policy learning provide population-level guarantees, but some individuals may experience adverse treatment effects. We recast targeted deployment as a multiple-testing problem: for each candidate unit, test whether the outcome under the AI policy would exceed a user-specified threshold, and control the false discovery rate (FDR) among the selected set. Our method, Conformal Policy Selection, combines weighted conformal p -values that account for the distributional shift from behavior to target policy with a two-step selection-and-pruning procedure. We prove finite-sample FDR control and establish a doubly robust asymptotic guarantee, where FDR remains controlled if either the outcome model or the propensity model is correctly specified. Synthetic experiments verify these guarantees across four misspecification regimes, and a semi-synthetic study built on real electronic-health-record covariates illustrates how the same FDR control can act as a deployment safety gate, bounding the harmful treatments that a globally trained policy would otherwise administer to a high-risk subpopulation where it misgeneralizes. The framework offers a principled method for safe AI deployment in high-stakes applications.
[LG-139] Unit-level aggregation breaks coverag e in causal foundation models
链接: https://arxiv.org/abs/2610.22770
作者: Gabe Schulman,Lucius E. J. Bynum,Yoav Wald
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 23 pages, 6 figures, 8 tables. Code: this https URL
Abstract:Causal foundation models (CFMs) show promising results for treatment-effect estimation, with unit-level estimates that are competitive on standard benchmarks. Many causal analyses, however, inform decisions through effects averaged over groups of units, such as the full sample, the treated units, or a subgroup, and for such decisions a CFM is useful only if its credible intervals are calibrated. We examine these intervals and find that their coverage degrades, both in simple synthetic settings and on standard benchmarks. On ACIC 2016, for example, temperature-calibrated 95% intervals from CausalPFN fall below nominal coverage beyond groups of 30 out of 4802 units, and intervals for the full-sample average contain the true value in only 20% of datasets. We trace this problem to how these models are trained and how their intervals are formed. They output a predictive distribution for one unit at a time and, at inference time, combine the unit-level variances as if the errors on different units were independent. We show formally that posterior uncertainty over the data-generating process makes these errors correlated. As a result, the combined variance is too small by a factor that grows with the size of the group. Empirically, this factor predicts where coverage fails and by how much. We then show that two neural process models, which make joint predictions for a group of units, account for this correlation and substantially improve coverage on synthetic tasks and, as heads on a frozen CausalPFN, on all six benchmarks. Comments: 23 pages, 6 figures, 8 tables. Code: this https URL Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG) Cite as: arXiv:2610.22770 [stat.ML] (or arXiv:2610.22770v1 [stat.ML] for this version) https://doi.org/10.48550/arXiv.2610.22770 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-140] Optimal approximation of smooth functions by ReLU neural networks
链接: https://arxiv.org/abs/2610.22693
作者: Thomas Nagler
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Numerical Analysis (math.NA)
*备注:
Abstract:We study uniform approximation of the Hölder ball \mathcal H_1^s([0,1]^D) by fully connected ReLU networks. Except in the one-dimensional Lipschitz case, existing constructions remain separated from the information-theoretic efficiency limit. This article closes the gap for every s0 and D\ge1 : networks with width O(N) , depth O(L) , and parameter magnitude O(1) attain uniform approximation error of order (N^2L^2\log N)^-s/D . No relation between width and depth is required. The construction uses new encoding and decoding mechanisms that exploit higher-order finite-difference bounds for local Taylor coefficients of the target function. As a statistical consequence, empirical risk minimizers over the corresponding clipped network classes attain the minimax regression rate n^-2s/(2s+D) without additional logarithmic factors. The paper also gives a detailed exposition of the key approximation techniques, covering both established and new mechanisms.
[LG-141] Quantum-Classical Neural Networks for Solving Lindblad Master Equations
链接: https://arxiv.org/abs/2610.22691
作者: Rafael Florencio Díaz,Julio Guerrero García,Elvira Romera Gutiérrez
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG)
*备注:
Abstract:Variational quantum algorithms have emerged as a promising approach for simulating open quantum systems, yet their expressive power and trainability remain poorly understood. We propose a variational framework for solving Lindblad master equations by minimizing a residual cost function using a compositional ansatz obtained by combining a classical neural network with a parameterized quantum circuit. The density matrix is represented through a purification on an enlarged Hilbert space, ensuring positivity, Hermiticity, and unit trace by construction, while all gradients are evaluated exactly via the parameter-shift rule. Beyond the algorithmic formulation, we derive an upper bound for the covering number of the proposed compositional model and compare it with that of purely quantum neural networks. Our analysis shows that the classical parametrization enlarges the effective hypothesis class while providing finer control over its complexity, thereby establishing a theoretical connection between expressivity, covering numbers, effective dimension, and the barren plateau phenomenon. Numerical experiments validate both the accuracy of the proposed method for representative Lindblad master equations and the theoretical predictions regarding gradient concentration. These results provide a unified variational and theoretical framework for the simulation of dissipative quantum dynamics using compositional quantum machine learning models.
[LG-142] Source-Aware Dynamic Pricing with Strategic Buyers
链接: https://arxiv.org/abs/2610.22516
作者: Jinjiang Zou,Yang Li,Zemin Zheng
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
*备注: 28 pages, 9 figures
Abstract:In modern commerce, firms and platforms typically use customer features from different sources to personalize prices: some features are verified using seller records or third-party data, while others are self-reported and may be strategically manipulated by buyers seeking lower prices. We therefore introduce a novel source-aware framework for contextual dynamic pricing with strategic buyers in a high-dimensional semiparametric setting, where buyer valuations follow a sparse linear model with an unknown noise distribution. To account for differences in manipulability across feature sources while enabling effective valuation learning, we develop the Source-Aware Localized exploration and calibrated Exploitation (SALE) pricing policy, which draws on verified features to localize price exploration and combines verified and reported features to correct for strategic manipulation during exploitation. Our localized exploration design allows converting binary purchase feedback into unbiased signals of conditional mean valuations, yielding a new population equation that supports consistent sparse estimation despite non-i.i.d. buyer arrivals. Building on this estimate, we develop a localized empirical-process analysis that exploits intrinsic sparse geometry to establish uniform guarantees for nonparametric estimation of the noise distribution in the high-dimensional setting. With these estimation guarantees, we establish SALE’s self-recovery property, which keeps localized exploration valid with high probability and leads to a sublinear expected regret bound. Both simulation studies and real-data experiments demonstrate the appealing performance of SALE. To the best of our knowledge, our study is the first to provide a theoretical and practical framework for incorporating information provenance into dynamic pricing with strategic buyers.
[LG-143] An Adaptive-Horizon Graph Neural Operator for coupled surrogate morphodynamic modelling
链接: https://arxiv.org/abs/2610.22432
作者: Ying Cao,Yifan Yang,Junqiang Xia,Julia C. Mullarney,Mário J. Franca
类目: Computational Physics (physics.comp-ph); Machine Learning (cs.LG)
*备注: 36 pages, 13 figures, 1 table
Abstract:Morphodynamic simulations require repeated coupling between hydrodynamic solvers, sediment transport modules, and bed updating. For long duration, multi-scenario simulations on large unstructured meshes, such coupling is computationally demanding. This study proposes an Adaptive-Horizon Graph Neural Operator (AH-GNO) that formulates bed evolution as a graph operator learning problem from recent hydrodynamic and topographic states to predict future bed elevation increments. The model operates directly on the native unstructured meshes. The approach integrates area-weighted graph learning, historical state selection, and forecast horizon estimation, allowing the effective history length (K) and update interval (B) to vary with the evolving flow and bed state. Using Telemac2D/GAIA solutions as supervised numerical references, the framework is evaluated on the official Bump and Yen bend benchmarks and on a field-scale application (Waiho River, New Zealand). Results show that the AH-GNO reproduces the main patterns of bedform migration, bend-scale erosion and deposition, and localized bed adjustment. Across the three cases, the tested CPU-GPU deployment reduced post-training bed-update workflow time from 12~1204 s with GAIA to 0.08~3.34 s with AH-GNO. This improvement corresponds to deployment-specific accelerations of approximately 150~360 times relative to 4-core GAIA CPU configurations, and with 105~186 times gains for 20-core CPU configurations. These results suggest that, while retaining hydrodynamic feedback from TELEMAC-2D, a learned operator for bed updating can provide an efficient acceleration component for morphodynamic scenario analysis and rapid coupled simulation.
[LG-144] Deep Learning for Forest Plantation Mapping in Godavari Districts of Andhra Pradesh India
链接: https://arxiv.org/abs/2610.22415
作者: Snehal More,Anuj Karpatne,Valerie A. Thomas,Randolph H. Wynne
类目: Image and Video Processing (eess.IV); Machine Learning (cs.LG)
*备注:
Abstract:Small-area forest plantations play a vital role in the socioeconomic well-being of farmers in Southeast Asia. Most plantations are established on former agricultural land, often on land less suitable for agriculture. Plantations that are converted from natural forest have adverse impacts on biodiversity. Mapping small-area plantations is thus important to understand the dynamics of forest cover in Southeast Asia and to study the social, economic, and ecological effects of this important land cover and land use change. While the small size of forest plantations makes it difficult to detect them using moderate resolution satellite sensors, the problem is exacerbated by the high degree of mixing between plantations, surrounding vegetation, and other land covers, which often show variegated responses in satellite signals across space and time. In this paper, we study the problem of mapping small-area forest plantations in East and West Godavari districts of Andhra Pradesh, India using deep learning methods. Remotely sensed cloud-free data from the Harmonized Landsat Sentinel-2 S10 product were classified using a pixel-level neural network and training data labeled using a field-based survey in concert with expert aerial photo interpretation. We compare the performance of deep learning methods with a baseline random forest classifier in our study region of 21543 sq. km over a period of 3 years and analyze the differences in the results across land cover classes and seasons.
[LG-145] Physics-Transfer Learning: A Physics-Constrained Theory for Zero-Shot Generalization Along the Ladder of Complexity
链接: https://arxiv.org/abs/2610.22376
作者: Yingjie Zhao,Zhiping Xu
类目: Computational Physics (physics.comp-ph); Machine Learning (cs.LG); Pattern Formation and Solitons (nlin.PS)
*备注:
Abstract:Reliable out-of-distribution generalization remains a central challenge in machine learning, particularly when models are deployed in data-scarce high-fidelity domains. Existing approaches often rely on statistical distribution alignment, or invariance assumptions that can fail when domain shifts reflect changes in underlying mechanisms. Inspired by a longstanding paradigm in physics that understands complex systems by starting from tractable idealizations and progressively incorporating additional mechanisms and levels of fidelity, we propose physics-transfer (PT) learning, a physics-constrained learning theory for zero-shot generalization along a ladder of complexity. PT constructs chains of anchor domains (CADs) that connect low-cost source domains to complex targets and introduces computable feasibility and reliability indicators for assessing transfer before target-domain labels are observed. Under physical-consistency and stable-representation assumptions, we derive an operational upper bound on target risk in terms of source empirical risk, the PT indicators, and finite-sample errors, and establish convergence as adjacent complexity gaps shrink and data coverage improves. We evaluate PT on excited-state energy prediction in quantum systems and dynamical-transition prediction in classical systems using neural networks trained without target-domain labels. Across substantial distribution shifts, PT yields accurate high-fidelity predictions, while the indicators identify incorrect physical mappings before target-domain prediction fails. These results provide a model-agnostic framework for designing, diagnosing, and controlling generalization in data-scarce learning problems.
[LG-146] Property Estimation in Geotechnical Databases Using Labeled Random Finite Sets
链接: https://arxiv.org/abs/2410.22659
作者: Changbeom Shim,Youngho Kim,Craig Butterworth
类目: ignal Processing (eess.SP); Computational Engineering, Finance, and Science (cs.CE); Machine Learning (cs.LG)
*备注:
Abstract:The sufficiency of accurate data is a core element in data-centric geotechnics. However, geotechnical datasets are essentially uncertain, whereupon engineers have difficulty with obtaining precise information for making decisions. This challenge is more apparent when the performance of data-driven technologies solely relies on imperfect databases or even when it is sometimes difficult to investigate sites physically. This paper introduces geotechnical property estimation from noisy and incomplete data within the labeled random finite set (LRFS) framework. We leverage the ability of the generalized labeled multi-Bernoulli (GLMB) filter, a fundamental solution for multi-object estimation, to deal with measurement uncertainties from a Bayesian perspective. In particular, this work focuses on the similarity between LRFSs and big indirect data (BID) in geotechnics as those characteristics resemble each other, which enables GLMB filtering to be exploited potentially for data-centric geotechnical engineering. Experiments for numerical study are conducted to evaluate the proposed method using a public clay database.
附件下载
